Day on the internet. We're excited to have you here today. We have a full agenda. I'll share the schedule for the day in just a minute. Before I do that, I wanna tick through a couple of logistical details. First, if I can ask everyone to check their phones to make sure that they're set to silent, just so that we have uninterrupted presentations over the course of the day. Also, throughout the day, we'll have beverages available for anyone who needs to rehydrate. You'll just go outside the conference room doors here, and you'll be able to grab something to drink. Also, we have charging stations here in the back of the auditorium if you need to recharge devices.
We've also got them in the back of the back of the building here, where you checked in at the registration desk this morning. Finally, we do have Wi-Fi. I understand that some folks are having trouble. We have one zone that's working, one zone that doesn't. The registration information for the Wi-Fi is in your white welcome packet. We'll get information and get that other Wi-Fi zone that's not working fixed, and we'll get that to you here shortly. All right. With that out of the way, I'll read the risk factors, and then we'll move on to the schedule for the day. Today's presentations contain forward-looking statements. All statements made that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially.
Please refer to our most recent SEC filings, including Form 10-Q, 10-K, and earnings release, for more information on the risk factors that could cause actual results to differ. If we use any non-GAAP financial measures during the presentations, you'll find on our website, intc.com, the required reconciliations to the appropriate and comparable GAAP financial measures. Okay, now moving on to the agenda. Andy Bryant, the Chairman of our Board of Directors, is going to kick off the day with a welcome to all of you and an introduction of our CEO. BK will then take the stage for the first keynote. He'll discuss the three strategic vectors that underpin our approach to driving growth and shareholder value.
He'll spend some time on each one of the business segments, with a particular focus on the role of IP reuse and efficiency and synergy across the breadth of the product portfolio. BK will hand off to Bill Holt, who's the General Manager of our Technology and Manufacturing Group. Bill's going to share an update on our 14 NM process ramp, followed by a discussion of some of the material science and transistor innovations that he and his team have been working on. After Bill wraps up, we will take a short break. Out in the lobby, we'll have refreshments, people can recaffeinate, and the executive team will be available at that point for informal networking. I will point out that lunch starts a little bit late this year, so you may wanna grab a snack during that mid-morning break.
When we return from the break, our CFO Stacy Smith will take the stage. Stacy's position immediately following Bill is no accident. You're going to see some explicit links between Stacy's material and Bill's material, specifically the discussion of our cost structure and our segmentation strategies. Stacy's going to begin with some observations about the evolution of our industry before he reviews the financials and our performance and outlook for next year. As Stacy finishes up, I'll invite BK and Bill to join me on the stage with Stacy. We'll have a Q&A session. I'll say a little bit more about the logistics of Q&A when that point in the morning comes. We'll transition to Diane Bryant. Diane is, of course, the General Manager of our data center business.
She's gonna spend some time describing the underlying trends driving data center demand with a specific tilt toward why Intel is positioned to win, across a whole range of workloads in the data center. Diane will hand things off to Kirk Skaugen, the general manager of the PC Client Group. He's gonna talk about PC innovation, the role of innovation, and the role of performance and new user experiences in catalyzing growth in that segment. We'll wrap up the morning with another Q&A session, this time with Diane and Kirk, and at that point we'll break for lunch. The lunch will be very informal. We're going to have at least one Intel executive, in some cases two, positioned at each table, which will be in the check-in area where you greeted or where we greeted you earlier this morning.
You're guaranteed to be seated with at least one Intel executive. Please feel free to grab any chair you like. Coming back from lunch, we'll take a new approach to the structure of our afternoon. We've not done it this way previously. Our President, Renee James, is going to lead us through a discussion of a series of strategic platforms, and then she's going to invite the general managers of several of those platforms up on stage to do deep dives on each of their businesses. We'll hear from Hermann Eul regarding Multi-Coms, Doug Davis on the Internet of Things, and then we'll wrap things up with Rob Crooke on nonvolatile memory. As Rob concludes his presentation, we'll do one final Q&A with the last few presenters, and that'll wrap up the day's formal presentations. We will invite you to join us for a reception.
That reception will be held in the same place as lunch, which was also the same place you checked in earlier in the day. As we do every year, we'll have the executive team, a really wide swath of the executive team, available for informal interaction and Q&A, and we'll also have a nice, rich set of demos for you to check out, hardware and software demos. That concludes the schedule for the day. With that, I'd like to invite the Chairman of our Board, Andy Bryant, up on stage to share some observations and to introduce BK. Andy?
Thanks, Mark. Good morning. I got a couple of good mornings back. I would like to also take the opportunity to welcome you to the Intel 2014 Analyst Meeting. I would like to begin by introducing a couple of our independent board members who will eventually be sitting over in this area. The first is Jim Plummer, recently retired, Dean of Engineering at Stanford University. The second is Frank Yeary, who is a Principal at Darwin Capital Advisors. Frank will be here sometime towards the middle of the morning, is what I've been told. They will be here. The foundation of Intel's business model is silicon technology, also known as Moore's Law. It's a lot more than, of course, Moore's Law, but that's how it's frequently referred to.
This has defined Intel's past and, quite frankly, will shape our future for years to come. Moore's Law teaches us that capability, silicon computing, however you wanna define it, will continually get smaller, lower cost, more capable. Could be better power usage, could be better speed, and by inference, will also become more mobile. Yet, many of you probably remember a year ago when I stood here and said I was embarrassed by the fact that we had not paid attention to the lessons of Moore's Law. We had as, by the way, not only had we not paid attention to our heritage, we'd not even looked around us. If we had looked around us, we would've seen that Moore's Law was driving new forms of compute capability, new forms of communication capability.
Quite frankly, we had missed it, and we had sat on the sideline for far too long. Today, I'm actually proud to say much has changed. In the last year, we've gotten ourselves on track to ship 40 million units into the tablet marketplace, which should make us, by most accounts, the second largest supplier of silicon into the tablet marketplace in terms of processing capability. We also recently announced a modem deal in LTE with Samsung, a flagship company, flagship product for us. Beyond that, you've seen the company announce some strategic relationships like with Rockchip, like with Spreadtrum. You know, we've been in business in China for 29 years. This continues to evolve our presence there and our ability to work with strategic partners like these two companies. You see a company that's changing.
You see a company that's listening to history again, watching the world around us and making significant progress. There's a new energy in the company, a renewed commitment to being successful in the marketplace, a new resolve to be willing to take informed business risks. This is much different than we saw, we'll say, a year ago. It's really heartening to see these changes. Now, by the way, it helps when your other businesses are performing well, right? We've seen the memory business, IoT, PCs, data center all have relatively good years. When things are going well, it gives you the time and the resolve to make the changes you need to make in other parts of your business. Now, of course, all is not well. We know that.
I would be remiss to tell you I'm not aware. Actually, it's more than I'm not aware. We, the company, we, the management team, we, the board, are aware that we're losing a lot of money trying to regain our presence in the mobility space. When I look at that, I'm not gonna tell you I'm proud of losing the kind of money we're losing, I'm also gonna tell you I'm not embarrassed by it like I was a year ago about where we were. This is the price you pay for sitting on the sidelines for a number of years and then fighting your way back into a market. We will improve this. We will not continue to accept a business with multi-billion dollar losses, but this is the price you pay to get back in, and we will get back in.
We are getting back in. The other thing that's interesting is, in general, it's been a good year for Intel. You could easily stand up here and say everything's wonderful. Many of you know me from the past. I'm kind of an Andy Grove disciple. In Andy Grove world, you don't get victory laps. You may get a moment to be happy about what you've accomplished, and then you have to move on. Every day, we have new challenges. It doesn't stop. We cannot become complacent. We have to continue to get a return on our investments. We have to continue to be the best in the world at silicon technology. We have to continue to earn high returns on these innovations. If we don't do these things, we're not doing our jobs.
I also want the investors to know the board listens to the investors. What we've heard from investors are a variety of things, what we are focusing on is making sure that we're comfortable with the strategic decisions and plans of the company. We review and monitor those. We are listening to the investors on board composition, we're listening on cash. As an example, we do monitor the mobile business unit. We do monitor this to approve the strategy. We do look at the losses. We do look at the plans to make those better, along with, quite frankly, the data center and how it's gonna continue to grow and be a profit driver for Intel. We've added a new board member who brings new experience and capabilities to the board this year.
As many of you have noticed, we've increased the buyback pretty dramatically, using our cash less on the balance sheet and more to return to shareholders. We think we're seeing real progress. We will continue to monitor, and we will continue to expect to see the good results we've seen be better in the future. The good news is, I don't have to do all that. What I'd like to do now is bring up the person who will be leading the company in delivering everything I just promised on. Our Chief Executive, Brian Krzanich.
Good morning. This is my second Investor Day as the CEO of the company, so I'm excited and I was sitting down in my chair, and I looked up, and I was kind of gazing at all the products, and I thought, "Boy, a year ago, I don't know if I'd have seen the diversity and some of the innovation that are on the stage here." I'm really proud of what we're gonna show you this morning. As we go through the day, as we, as each one of us comes up and does a presentation, we do the demos, you'll see the demos during the breakouts at lunch or this evening at the reception.
Really take a look and ask questions about what's driving the innovation, how are we listening to the marketplace, how are we thinking about where the market's gonna go with innovation. Definitely this is a different field than a year ago. I thought I'd start with a little bit At Intel, we're very good at grading ourselves, and I thought I'd start with what we said last year, and then talk a little bit about how I felt we did. We put up this slide last year because we thought with the new leadership, with Renee and I, we should really talk about what will stay the same and what will change.
What we said will not change are the fundamentals that Andy talked about, the relentless pursuit of Moore's Law, and I'll talk about that this morning, and Bill Holt will talk about that as well. That we would develop products that continue to make the best PCs in the world, and I think what you'll see up here today is PCs that are just dramatically transforming the way we think about PCs, the way we interact with PCs. Things that we would do differently is really that market-driven view of our industry. No longer would we say, "This is the way we'd like the industry to be." We've really been asking ourselves, "Okay, but where is it gonna go independent of what we do? Where do the customers want us to go?
Where does the end user want us to go? I thought another one that was really important here is that we create platforms, not just silicon. We're finding that as we drive this innovation, that's becoming much more important, that you can no longer hand somebody a piece of silicon when you talk about RealSense or when you talk about a product like Sprout. You really have to work on a platform and then work with the software and all of the integration that the customer requires in order to bring a product like that to market. There's innovation on both sides of that equation to bring a product like that. Then drive a focus on bringing that innovation to market much more quickly.
I think you've seen a lot of that, and one place I always like to kind of highlight that with, 'cause there's always the classic products that we've been able to do, PCs, data centers, was wearables. One year ago, we talked about that we saw that market, that we weren't gonna be absent, but now you see products coming to market. You saw our SMS headset. You see our MICA bracelet. All of those things are already in market and already setting trends and innovation. I think for me, I graded us pretty high. Andy said, in Grovian terms, you get your moment, so I thought I would take my one little moment and talk about grading of 2014. I promise it will just be a moment.
I t really was or is a record year of records so far. We're really setting up to be just really a year where the company's driving on all cylinders. We set a goal of hitting 40 million tablets. We said at the beginning of this quarter we were well on target to that, and that this quarter we felt like we would cross over that 40 million unit number. As I said, we're still on that target as we move through the quarter. Q3 was very special to us because we crossed the 100 million unit number for a 1/4 in CPU units, and that's a number that's kind of been out there for a long time for us, and for us to cross over that number, we were all really excited about that.
To me, that meant not only were we able to grow the top line through ASPs, which we've done in the past through mix and our great products, but there was true demand and increased demand coming for our products. Andy talked to you about shareholder return. It comes in both our buyback, our stock price, our dividends. 45% shareholder return for the year. The last is we forecasted a $55.9 billion revenue for the year, which will be a record year. I'll take my moment, and I'll move on to what we still have to do, but I thought it was important to share with the investors that we're very proud of this year and that we really feel like, as Andy said, we're on a momentum to carry change throughout the company.
Let's go now to the work that we have to deliver. As Mark said, we feel like the company is driven by three key vectors, the first being Moore's Law, and I'll talk to you a little bit about that this morning, but Bill's gonna give you a really fun, in-depth discussion of Moore's Law for those who really enjoy it. I do wanna make a point. Next year is the 50th anniversary of Moore's Law, and I don't think a lot of people think about this being a law that's been around for 50 years. Through that time, my 30-year engagement with that 50 years, myself personally as an engineer at Intel, I can tell you many times people have talked about the law ending.
Our job at Intel is to make sure it lives on for as long as possible, 50 years is a momentous milestone, and we will be doing some things next year to recognize that. The next one is integration, I think we've done a lot of work around integration. You see some really good products. We'll talk about them a little bit this morning. SoFIA is a great example where we've tried to really drive integration, you're seeing more and more integration with Broadwell. You'll see more next year with Skylake. We've got Cherry Trail. All of these products are driving more and more integration in. Then we'll spend a little bit of time showing you, giving you some examples of this shared IP.
There are examples here on the stage that as you go through the day, you're going to see, you know, when you have an IP like RealSense, you have this very unique capability in the computing and the imaging coming together that you can really drive that IP across many, many products. We'll talk to you a little bit about some of those, and I'll tease out some that you'll see coming as well next year. These are the three strategic vectors that we believe drive most of what we do. Our mission/vision strategy then is built on those vectors, and it's really to, you know, the mission to really utilize Moore's Law. We have it. We believe we lead in it, we drive it, we define Moore's Law as a company.
Our job now then is to really utilize the power that that could deliver in all different ways, and I'll show you later that there are three vectors that we think of Moore's Law that you can use, you drive with Moore's Law. Our vision is that it's no longer just that you want to provide a smart device, but almost every device that we talk about today will be both smart and connected. The criticality of IPs like modems and Wi-Fi and Bluetooth are obvious to us and important, and we'll drive those across our products. Our growth strategy is that we believe the highest shareholder return comes from a strategy that uses these core assets to move into profitable adjacent businesses.
As we talk about the Internet of Things, we look at that as an adjacent business. We look at it as, how can we move IP from those cores, from the Moore's Law work we do, from the innovation around connectivity that we have for the PC and for other mobile products? For all of the things that we do, how can we move that IP into the Internet of Things space in a profitable and efficient way? That gives us, we believe, real leverage to move into these markets. To kinda demonstrate that, I thought what we would do is spend a little bit of time, and first we'll talk about those core assets. I wanna talk a little bit about cores, a little bit about process technology, and a little bit about shared IP.
Then we'll go around and talk about some of those adjacent businesses this morning. That'll be how we kinda have a discussion of what's the work that we're gonna get done. As you look up here, let's talk about the cores first. It's really the core is that center of the PC. Kirk will talk to you about this morning, but we believe that we've really driven stabilization into the PC market. We'll talk about forecasts, but whether you consider it a flat forecast or slightly up forecast, I think most people have said it's not in that steep decline that we were seeing a couple years ago. You could argue about why and what is driving that, but we believe that what's really been driving that is choice.
We're across a lot more price points, and we'll talk about that. OS, can select almost any OS now. Real innovation in form factor, in usability, in user interface, all of those things have driven. You know, what we're gonna show you, and I think this next chart really kind of highlights it best, is that the lines of what makes up a PC, as you look across here, if you can tell me where the PC line ends and where the tablet line begins, it becomes hard. It becomes hard because we've driven innovation, right? As we move to Broadwell, we've brought out the Core M, devices are getting thinner. That's also driven a lot of what's driven the stabilization. It also is what we believe kind of drives why we see large distributions in forecasts for PCs.
A part of that is disagreement, debate, discussion about what is a PC. Is the Surface a PC or a tablet? Is this a PC or a tablet? We can each pick our devices out there and ask ourselves, "Are these truly PCs or tablets?" We're very comfortable with that. We believe it actually plays to our strength, that allows us to move that IP and that innovation across the spectrum. This also gives us a clear opportunity to be successful in forecasting because we can guarantee that somewhere in that range our forecasts will lie. That's a good thing, too. You know, the bottom line is that almost everybody is forecasting that this market has stabilized. What Kirk's gonna come up and talk to you about is he believes he can actually make this grow.
What Stacy will talk to you then about is, okay, even with a stable forecast, how does 2015 look? That's the stabilization of the PC. The next thing I've been working on is real innovation around the PC, and again, Kirk will talk to you about a lot of the devices here. I wanna talk about RealSense because to me, RealSense is a great example of IP that will go in many, many areas. A RealSense is a camera capability along with compute and software that allows the computer to really start to begin to see. If you step out of the, just basic PC and start talking about devices with compute capability to be able to see and interact, then you start thinking about all of the various applications.
It'll start in the PC, you'll see it first come out. You'll see it first be things like the Dell Venue 8, which is our first substantiation of this, where you can do different focusing, right? We announced this. You can do measurements. As we go through 2015, you'll see this one move into more and more capability as we drive this into our products. You'll see you facing and world facing. As you watch this come out, think about the future. Don't just think about the PC that you see, and am I comfortable with the fact that I'm using my hands? We're gonna talk about how people are stepping away from the keyboard and stepping away from the two-dimensional world into a three-D world. Think about the usage in tablets, in phones, in the Internet of Things, right?
Think about robotics that can see, truly see with depth and perception. You start to think about what you can do and what you can drive into the fields and segments of this business, that we can take this IP with very little additional R&D and carry it into new areas, right? That's one of the key strengths that we, as Intel, bring to these marketplaces. We also have wireless charging, security. We've got efforts in WiDi and WiGig for displays and connectivity. All of these things you'll see transform and move across the various segments of our business. Then on the right, we call it the options for everybody.
I believe this is one of the other things that's really driven stabilization in this segment, is that you've got price points from $199 to as high as you'd like. In fact, one of our fastest-growing segments is the real high-end gaming segment, right? That no longer does when somebody walks into the store and says, "I wanna buy a new PC," are they limited to some price points that are $600-$1,200. They can come in for a low price point if they need to, $199, $299, looking at different capabilities. Often they come in with one expectation, and they tend to buy up as they see the increments of performance that they can buy. Then we have options for the real high end, as I said, for the gaming devices.
OS of choice, that's how we call it, is that we want to provide platforms that really can go across every OS and are optimized and run best on our architecture for every OS. We're still the only supplier where an OEM can build a single platform and really choose amongst all these OSs at literally time of shipment or in the field. That's a unique capability we bring to this marketplace that we feel has added a lot of stability, a lot of capability, and a lot of choice for the customer. That's the PC. The next area I want to talk about is the data center. I noticed Diane was able to get a demo even with the data center, and it's-
It's very exciting.
Yeah, it's exciting. I seen it. You know, for us, the data center is really Intel's next big business. It's already a big business. We've said that we can continue to grow it at a 15% CAGR through at least 2018. It's got a broad customer base. We've got a good market position. We've got leadership products and leadership technologies. Diane's gonna talk to you a lot about this. This is an area that we've really focused on well as a company, and we have a plan to continue to focus on this and continue to move this at this fast growth rate. We're talking about a $14 billion business that we're saying we're gonna continue to grow at 15% rate through several more years at least.
We don't do that just by great products and such. We're gonna do that through real strategic planning, you've seen this come out and play out a little bit this year. One of the focus areas we call software to silicon. The idea here is that, again, our Moore's Law capability gives us transistors that are highly capable, we're able to add at a fixed cost, basically, with each generation, more and more of those transistors. We believe there's a unique capability that if you move certain segments of the software down into that silicon, you can provide a lot of either performance or feature. Now performance is obvious. It runs faster. It boots quicker. It goes at a, you know, the answer comes out sooner.
You can also provide features such as access into the encryption engine, and so security of moving data around the system is much higher because the software is in the silicon, and it is able to access the encryption much easier, much more directly. It is not passing through buses. A great example of that was our first Kind of effort into that, which was our partnership with Cloudera that we started this year. You saw us do an investment in Cloudera and start to build a real strong partnership. The idea there was that we wanted to build a very strong relationship between the two companies so that we would both open up the details of how our systems worked and allow the two to start building systems upon each other. You saw the first Intel-optimized Cloudera release come out a few weeks ago.
You saw Diane announce earlier this year a custom integrated Xeon with an FPGA product for our customers. That FPGA is designed to allow customers to bring that software down into the silicon. Then you'll see us continue to work on software-defined infrastructure using our OpenStack. Again, the same thing. The idea here is, in each one of these cases, bring the software down into the silicon, provide performance, provide features, provide unique capabilities that we're able to do through Moore's Law, and through IP that we've shared up into the data center that we believe nobody else can do. That is going to be a driver of a lot of the strategy in this area moving forward. You'll continue to see this effort of software to silicon and building out these capabilities across the data center.
Diane will talk to you a lot about that as she comes up this morning. I wanna move now to the next section, one of the core fundamentals is Moore's Law, and Bill's gonna talk to you about this in great detail, but I wanted to just talk about kind of my view of Moore's Law from this seat. I always remember that first you have to have leading-edge transistors. The real driver of Moore's Law is not only the scale function of the transistor but the performance function of the transistor.
You gotta really drive both variables, and that's really what's driven Moore's Law over the last few years and why you see it's not only scale, but it's architectural or the structure of the transistor changing as we went through, you know, strained silicon and High-K metal gate and FinFETs. All of those things are driving not only the scale but the performance of that transistor, the capability to drive, you know, the speed and the leakage and all in one, and Bill will talk to you about that. I always remind people that Moore's Law is not a single path. It's not a single road as a company thinks about what do you do with Moore's Law. There's really three vectors that you can then utilize Moore's Law for.
Those three vectors are price, cost, performance, and power. You get all of those as you drive the Moore's Law equation. As an architect then, you choose among those, and you balance those as you define your products. You can actually go in and move along those vectors, trading off one for the other often, but optimizing each of your products. I always remind people Moore's Law is not a single road, not a single path that you have to choose. You really have options to go down multiple paths here, and that's important. Okay, the last one of the key vectors I said was really pushing IP out. We talked to you a little bit. I used RealSense as a good example. I thought we, you know, we're talking to investors.
I thought, well, okay, they are always more interested in how we spend our money and how is it in the hardware. We put the graph up on the left, and the graph on the left is really our R&D spending. What you see is about 75% of our R&D spending is really shared R&D spending. It's either the fundamental spending for Moore's Law, the basic architectures, the validation and testing labs, all of those things that are simply required to build any product that you want to go do. The shared platform IP, as I call it, things like the connectivity, graphics, you know, modems, memory control, all of those kinds of things that, you know, they're used broadly across a wide swath of products.
75% of our spending is what we believe could be used in any one of our businesses, pretty much. The other 25% then is broken out into the various segments, and you can see the largest segment ends up being the PC. A fair amount of that, even though it's PC-specific, and these are things like RealSense or WiDi or WiGig or all of those things, they end up actually moving down into that orange segment down below eventually as they spread out across products. Any one of those things, you know, wireless WAN, you could have said was a PC R&D at one time, over time, they move down, that's really our process. Then on the right is actual diagrams. We used actual products, what we're showing you is Haswell.
I had to practice this a couple times last night because I am color blind, and I got the colors wrong a couple of times yesterday. I'll hopefully get this right this morning. The blue areas are the shared IPs. As you look at Haswell and you move over into the Xeon or the Bay Trail, the blue areas are things that whether it was in Haswell or whether it's in Xeon or Bay Trail, it's the same. We wanted to show you that it can work the other way as well. The green area. Yes, green area. They gave me my cheat sheet here, so I did the colors right. The green area is from the Bay Trail back into the Haswell, right?
Those are things like WAN and Bluetooth and some of the low-power connectivity capabilities that we developed for Bay Trail that we've pushed back into the PC for certain segments. These are really good examples. You can see how much of the die area is really shared amongst all of these products. To me, this was a great use or example of, this is real. This isn't just foils. The IP is simply moved from one side to the other. That's why you see us, you know, the work, the time is spent getting Haswell out, and then boom, boom, boom, every three to six months, you know, all of these other products kind of flow out as well, as long as the SKUs, as well as the SKUs of Haswell and things like that.
You're seeing the same thing on 14 NMs. Broadwell's come out. We said Cherry Trail will be out by the end of the year. It's the same process. The IP is reused and moved over. I wanna move on now to the first of the adjacent businesses, and I think we have to move to mobile first. The first thing I wanted to talk about is, as Andy said, I think, I remember a year ago and I said, "Yeah, I think we can, let's set an aggressive goal and 40 million tablets." Andy said, "No, no, no. You need to say you're gonna do north of 40 million tablets." I'm telling you still that we're on track to be north of 40 million tablets this year.
As Andy said, you can't stand up here and talk about 40 million tablets, and I am very proud of the innovation, the form factors, how we've moved into the China ecosystem, all of these areas, without talking about the loss that we have. As Andy said, it's not we're proud of it, but we're not ashamed of it either. When you come into a market that's that far along its maturity curve, and you're fairly absent from it, you've gotta go get your way in, and the products weren't really aligned to it. Our strategy wasn't to go there. We were shifting our strategy, and you have to make a choice. Do I wait until all my products are lined up?
You know yourself it takes a year at minimum, usually two or more, to move a product alignment to a market. You look at that and say, "Can I afford to be absent from a market for that period of time, or do I need to make my move, take my losses, but establish my relationships, establish my footprint, and start driving innovation and move my products?" I believe that it actually helps us move our products at a faster rate, too, right? We understand it. We were very transparent about this was how we were doing and why we were doing it. What Stacy will talk to you about is how we think we're gonna move through 2015, what kind of improvements we think we can make, and how we look at the future of this.
We're in this business to win, and we're in this business not to lose money, but to make money in the long run. You see a similar strategy being driven into this of cost optimization and differentiation, similar to what we do in the PC space. Actually, we think about this as an extension, as really a right arm of the PC space. The first thing we've worked really hard on this year is good, better, best. That's our classic segmentation of our products that allows us to segment the price points, the costs for us, and this is how you drive a lot of that cost reduction and improvement in the P&L of the business. SoFIA has been key to that. I remember roughly a little bit more than a year ago, SoFIA wasn't even on our roadmap.
As we saw our need to get in the mobile space, and we were looking for what the right pragmatic solution to get into this space was, the team came with the SoFIA solution, and we've acted very quickly. When Andy talks to you about the kind of different atmosphere around the company and the different behavior and the way people feel, I think SoFIA is a great example. The team came up with the option. We said, "That's great. We need it in market out by the end of next year." They gulped and said, "Okay." I can tell you that for the first substantiation of it, the 3G version, we're on schedule to have it qualified for production by the end of this year, and in the first half of next year, it'll be shipping.
We have the LTE version ready to be qualified in the first half of next year and shipping towards the middle to latter half of the year. That's a great improvement. Now we have a set of products that go from the low-cost good section with SoFIA all the way up through Cherry Trail, which we said was also on schedule. This, as these move through the segments next year, provide us right product, right cost for the right segment of the business. You see the Venue 8, as I said, was the first substantiation of RealSense moving into this space. As we move through next year, you'll see more and more of that.
We've demoed examples in a lot of places, whether it was CES or IDF or if you've come to Maker Faires, you've seen us demo it there. The ability to use RealSense in a tablet and scan people real-time, scan objects, all kinds of things. The Sprout is another good example of a RealSense application at the PC level. You talk about the mobile space though, you have to talk about phones also. It's not just a tablet discussion. A lot of people have said, "What's Intel's tablet strategy?" Our tablet strategy is to do it smartly. We feel like we can move into this space, and we can do it in a way that allows us to go a little bit more controlled, a little bit smarter along this path.
What we've done is we've taken the SoFIA product, remember about 1.5 year ago, a year ago, we said that we had developed a synthesizable core, an Atom core that was basically defined in software. That was a big deal for us. That was the first time Intel had done that, I don't think as we announced that, people grasped what that meant. That was actually the first step in setting up these relationships because you had to have a synthesizable core in order to go out to a partner and say, "I'd like you to build your products now with Intel architecture, and I can bring you a core that I can transfer to you in software and you can build with." That was critical for us to be able to do this.
You saw us set up new partnerships with Rockchip and Spreadtrum. Why them? Well, China is going to be the world's largest phone market. It's already the fastest-growing phone market in the world. Why not partner with two of the leading-edge silicon providers in those segments? Rockchip has more tablets, but they have a desire to move into phones. Spreadtrum is very much phones. They're in China, but also worldwide, with a desire to move up as well. We brought this synthesizable core along with other capabilities like our modems and things like that, they will build SoFIA-like products of their own roadmap that we both will be able to sell. They have that relationship in China. They have the markets already in place, and they can take this to market for us in phones and tablets and other products.
We believe this is a smart, efficient way to go into that market, to have an IA presence, and a large presence, but using partners that are already there, already established. That's what those mean. You've seen that we've done long-term agreements with some of our key OEM partners. ASUS and Lenovo are the two best examples. You see ASUS already, you see their PadFone that's already out. That was the first Intel modem phone in the U.S. For those who have been wondering when we would break into the U.S. market, you can go, and you can buy an Intel modem-driven phone today. Those modems aren't just critical for phones, and this is important. We believe the modems are critical for almost every product we'll build.
You can imagine in the Internet of Things, not having a modem or having access to a modem is going to be critical. You're talking about monitoring a pump out in the middle of nowhere on a, on a drilling platform, you're gonna wanna have some connectivity back, right? You can go all kinds of applications, you're gonna want a modem, you're gonna want Wi-Fi, you're gonna want various connectivities in Internet of Things. We chose two examples here, tablets and PCs, because we thought, okay, but these are areas that are critical to us today, and these are the forecast modem attach rates over time for each of these devices. We're already seeing modems attaching at different levels in the PC space with a lot of them as aftermarket add-ons.
A lot of the Chromebooks that are being sold today, because Chrome is really best when it's connected, are already being sold with modems built into the system as is. Modems will become more and more critical even in our core space. We felt investment in modems was not just about being in phones, but it's being in those adjacent products like Internet of Things and tablets, and it's critical for our core business like PCs. We believe an asset that over time will become more and more valuable, more and more critical, and one that we'll continue to invest in. I didn't think I could get through the day without talking about the organizational change we made.
We made it this week, and we did decide to do this, you know, prior to the investor meeting so that we could spend some time talking about it. It really was driven by two vectors. On the platform side or basically the product side, it's really driven by the fact that these are platform relationships. You're going into a customer. These customers are oftentimes the same customers that are buying PCs and servers and IoT or other products with us. You're working on a platform, a whole board that is going to build a product and a software stack and everything. One of the big feedbacks we've gotten this year is they think about these products as a continuum, just like we've told you as you look up on this stage, it's becoming a continuum.
They don't separate out in their own organizations, PCs and tablets and phablets and things like that. The feedback from the customers, and even then within our own organization was, these are really one continuum, one set of products. The first half of the reorganization was to move the platform organization in so that it was a client view of the world from a platform. It goes from PCs down to phablets. The other side was the modem itself, and we do As Andy mentioned, there was a Samsung announcement of modems. There have been some other modem sales we've done as well. When you do those sales, those are very engineering-focused sales.
When you go to integrate a modem into somebody else's platform, then you go to get that modem qualified around the world, it's very much an engineering-focused effort. What this reorganization does is allows us to provide one organization from top to bottom. It's within the same organization, no walls, no separation of leadership from the definition of the product through the engineering and architecture of the product, building the product, and taking it to the customer. It's one organization now. Again, our customers have given us feedback. Thank you. That's how we think about it. We wanna give you input on your roadmap, we wanna have insight into how the products are coming out, and we want help in integrating these into our products. I don't wanna have to go to three different people to do that.
Thank you for giving me one voice now that I can go to. That's what drove this reorganization. It's planning underway. It becomes effective in 2015, and we'll talk some more about that as the day goes on. I thought we'd talk about some of the other complementary markets that we're also moving into. I don't think we can go through the day without talking about Internet of Things and wearables. Internet of Things, and we broke out this year. We believe it's one of the next big businesses behind the data center. It's already a $2 billion business. We said that we could grow this at 20% per year.
You saw us break it out as a separate segment this year so that we could show you how big the business is. For us, this is a business we've been in for many, many years. This is the embedded business. What's changing? The cost, the amount of compute we can put out at the edge, the connectivity that's being added, and the usage models. It's a business that we're very familiar with and is already a $2 billion business and growing fast for us. The other one is wearables, we told you that we weren't gonna be absent from markets in the future. You've seen that we've already introduced products. We formed great partnerships. Our goal here is, like most of our products, not to have Intel-branded products, but to take key partnerships. You saw us announce partnership with Fossil.
As we go through next year, you'll see several other partnerships announced. We have the SMS Audio for ears. If you haven't used that product or gone and seen it, I personally use it almost every night on my runs. It's a great product. Great sound, gives you your heart rate. It's really fun. MICA, which if you've been reading the press, we really unwrapped all of the features that MICA has. That'll be shipping starting really in the December timeframe. I think great example of we're not gonna be absent, and we've already started to build these great partnerships in the segments that we have here. The next one is foundry. Foundry, you see us continue to work with our key foundry partners.
We've signed up several other partners this year. The one that we've been public on is Panasonic, but we continue to make good progress on our foundry business. We've tried to show you that we have a core set of businesses wrapped around Moore's Law, wrapped around the cores that we develop, and then the IP that we share, and how we take that into all of these businesses. I also started beginning with what was going to not change at the company and what would change. That was last year. That was last year when Renee and I were coming in and really starting to drive the change into the company.
To me, it's no longer a what it will stay the same and what will change, but really just how are we going to behave moving forward. If you take a look at moving forward, I think the things that are carried over from that comment last year is, you know, the ones you'd expect. We will continue to drive Moore's Law. It's the 50th anniversary next year. We'll do some fun things next year to celebrate the 50th anniversary of Moore's Law. We'll continue to develop products that enable the best computing experience. We're gonna talk to you about some of those today. We're gonna talk to you about the future of how we see those products today. A market-driven industry view. You saw even the reorg, I can tell you, I went out and spent time with the customers.
I couldn't tell them what I was going to do, but I'd ask them questions. How do you view us? What would you improve? How would you like us to interact with you differently if you could? How would you define Intel's organizational structure? Even that reorg was market driven. We're going to continue to open the foundry to all customers. Create platforms for the enterprise, not just silicon. We no longer think about a silicon sale. We think about what are we going to build with this. It's very different. Another key one is driving innovation to the market quickly. My personal opinion is we were too slow in the past. Things sat on the shelf. We didn't know what to do. I think this year you've seen us go much, much quicker. The two we're going to add is continued growth in the data center.
We believe that the data center is the next big business for Intel. It's already big, and it's going to be bigger. We really are focused on it's no longer just a compute world, it's really a connected computing world. You must provide connected capabilities. It's if it's smart and connected, it's gonna work best with Intel. That is our mantra. That is what we drive through all of our products. Now, I thought I would spend just my last couple of minutes talking about what has changed internally. I thought, as I was putting these foils together, what best way to do it than to show you actually the leadership team and how much that's changed, and yet how much that's also the same.
Many of these people are in this room this morning, but this is the leadership team that is driving, and many of them you'll hear from today, or many of them you'll have lunch. I thought one thing is, about 25% of you in the room, was my estimate that I got, are new to this meeting. I thought, okay, this'll be good. Let's show you the org structure. All of those that are in yellow are either new to Intel in the last year, since the last investor meeting, or new to their role. Like Kim Stevenson is not new to Intel, but she's new as our CIO, for example, and on MCM, our management team.
Steven Rodgers is not new to Intel, but he's our new General Counsel, as an example. Roughly 17% of this population is new to Intel this year. 25% are new to Intel in the past five years, and 33% are new to the role in 2014. We're changing at all levels. What was interesting is we just, about, I don't know, a month ago, we pulled the top 350 people, 350 leaders at Intel, together up in Oregon. Before the meeting, I was presenting a foil similar to this, to that team, and I thought, "Hmm, let's collect the data for the 350 that'll be in the room." It almost mirrored this data set exactly. It was off by 1% here or there.
What it told me is that these leaders are driving that same kind of change into their organizations. I think that's good. You don't want wholesale change. We have very successful organization, and we experience with Moore's Law is what counts. You have to know how to do these things and how to architect and build teams. There's an intimate relationship between the architect and the silicon designer in that early stage. Also, if you're trying to drive market-driven, if you're trying to move into new markets, you do need a fresh point of view. I really am very comfortable and very proud of the team and what they've done here, both bringing in new people, integrating them well. I think as you talk to some of the new people and new leaders here today, ask them how they've integrated it.
They'll tell you it's a very positive experience. They're helping drive change into the company and the views that we have. That's my last foil. I just wanted to kinda end with thank you for coming here today. I'm gonna bring up on stage now a great friend of mine and also somebody who I think can explain Moore's Law in much, more detail than I can, and that is Bill Holt, who will come up and talk about Moore's Law now.
Thank you, Brian. Thank you again for coming, and let me also add my thanks. I was here last year, and I made a few statements, and I showed you some data. In the same spirit that Brian said of looking back, I'm also going to start by recounting what did I say last year and what is that today. Last year when I was talking about technology, I was making lots of predictions, and that's because we were at a transition. This year, I'll also try to add a little bit of actual data to go along with those predictions and show you how things have really turned out. Make sure I'm synchronized here. Last year, I set this foil up. I said, what are the things that are key that I wanted you to take away?
One is that we will continue to deliver the benefits of Moore's Law. The second one was that true cost reduction remains possible in a capital-intensive environment. There was a lot of talk last year about the increasing cost of silicon wafers, about the amount of capital spending. Did that, in fact, eliminate the cost reduction? I said no. I'll show you that again. Finally, that the benefits that we can get from technology can be applied to all of our various product segments. Let me just go through and say that this year, none of those have changed. Those are all three the same key things that we're driving as we look at technology. Last year, I also showed you this as it related to 14 NMs. We said last year that we were having more difficulty than we would've liked.
It was harder than we had thought, and the fact is that we were somewhat behind. We said basically about 2.5 years, and we've been consistent in saying that. I think I should also update this for this year. If I take this foil and now update it, let me start in the lower right-hand corner. This is a reliability set status for the technology, and since we're in production, it had better be green, but it is all green. All of the basic fundamental capabilities are all taken care of. If you look above that, we're on track delivering the performance that we had projected. The other lower left-hand corner showed the matching of our factory parameters, and that was pretty good last year, and that remains on track. The one on top is yields.
I had been somewhat more optimistic about how soon we would close that. This is yield at that time on our test chips, and you can see that it has taken longer. It's continued to take longer, but you can see now it's converging, and that, in fact, the yield is now in a healthy range. You say, what about your product? Let's take a look at the yield that we're seeing on our lead product. This is the lead Broadwell, SoC product that we're now beginning to ship. This is aligned to when we began shipping those products. You can see it's safe to say that the yield is in a healthy range.
Again, we're not quite recovered all the way, but you can see we're very close now and on track to put this in the same yield category as our 22 NM technology. I'll just add this footnote that our 22 NM technology is the highest-yielding technology we've ever had, the bar that we're trying to catch up to there is a very high bar. That's essential because if you're gonna get cost reduction, you have to match yield, you have to match those other parameters to your previous generation. What you can see from this is that we are catching up. If you look carefully in the Q3 and Q4, that's when we began to ramp this technology. Our original projections had been that we would be aligned with the 22 NM yields.
Obviously, we weren't. Stacy will give you some details on how that affects some of our costs going into the beginning of next year. The parts that we're building on this in Q4 are the parts that we will ship and sell next year. There is an impact to that, there's no denying that. The good news is that as we progress through the year, you can see that the yields will match, and then we will begin to see the cost benefit in the latter half of next year. Stacy will give you a lot more detail on that. That's kind of a look back at what we had said. Now I wanna actually give you a little bit of an overview of our 14 NM technology. Last year, I just told you we were going to do it.
This year, earlier this year, we did a launch event and said, "What does it really consist of?" I'd like to review some of those foils and make a few key points. Let me start and say first it's a true 14 NM technology. I will give some details on the next page, but this is a very important point because there's lots of 14 NM technologies around, they're not all created equal. The second key aspect is this is our second generation of FinFET Tri-Gate devices, what you'll see is we've made substantial changes and improvements in this device. If you're going to drive performance, as Brian said, the transistor is the key to enabling performance improvement in our products, whether that's power reduction or frequency improvement.
Generally, that only comes if you have better transistors, and we've made substantial changes to continue that progression. I'll show you some data on the interconnect. This number is more esoteric to most of you, but as you'll see, we're delivering in this technology through 52 NM pitch on the interconnect. That's what's necessary to actually do scaling and make the die get smaller. Finally, kind of an interesting first is that this is the first time that we will have included air gaps in our technology in the interconnect system. I'll show you some pictures of that and explain to you why that's important. Let's go ahead and look at what a true 14 NM technology means. Historically, when you've changed the name, and the names, we can all admit, have lost their significance.
There is really nothing that's 14 about this or 20 or 22 about anything else. What we have preserved or had preserved was the ratio of those names was generally the linear scaling of the dimensions, okay? There was a reason that we went from 45 to 22 or 45 to 32 or 32 to 20 or, in our case, from 22 to 14. That's roughly the ratio that it takes in the linear dimension to get a 2x area scaling, okay? If you don't do that, you're really not going to get any substantial area scaling, and if you don't get area scaling, you will not get any cost reduction. For our 14 NM technology, we have, in fact, made those kind of scalings.
The pitch of the transistor Fin that I'll show you in a moment has gone down by exactly that amount. That's not just a coincidence. We know what it takes to get the scaling to happen, that's kind of a targeted number. In the case of the gate pitch itself, we were not able to get to that goal. It was a little harder, the number that we picked is slightly above, but it's still substantial in the amount of area reduction that it can drive. To make up for that, though, in the interconnect pitch, we went the other direction and actually scaled more aggressively.
That's where that 52 NM number is significant because that is a 0.65x scaling in the linear dimension for the wires at the bottom of the interconnect stack that will, in fact, enable the basic building block to scale by approximately this 0.5 area. When you compare, these are easy numbers to get. People publish them, as I'll show you in a little while. You can also easily find them in reverse engineering reports because people cross-section our parts and everybody else's. These numbers are not hard to find. They are the fundamental building block that enables the rest of the technology to actually provide real density scaling. Another way to look at that is to look at the memory cell. Memory cells also are common in almost all products these days, even in modems.
A large portion of the modems has a very large portion of SRAM. The scaling of the SRAM cell is another way to see if the technology is truly providing scaling or being a real true technology. Here, you see the outline of what the SRAM cell would be for our 22 NM technology, then again for 14. Again, you see the number is close. This also, we didn't quite get the desired 0.5, but we got a number that's very close to that range. Again, the interconnect pitch and what we can do in some of the graphics areas provides us even better scaling to make up for that. What you can do here is you also can find these numbers fairly easily and compare them and look and see what does it really mean when someone calls it 14 NMs.
I wanna move on to transistors, and I'm gonna do a little build here and show you step by step some of the changes that were made to provide increased performance for our transistors. On the left-hand side is the 22 NM. Right now it looks the same, the 14 NM, and we'll go through a few changes. The first thing we did was, in fact, reduce the pitch between the fins. That's that number I showed you on the previous page. If you're gonna make things smaller, they have to get closer together. The first step was, in fact, to reduce the pitch. That's not that easy. Those numbers are getting very difficult. If I was to relate where the challenges that we've had in our 14 NM ramp, they've all come from these areas where we've done aggressive pitch scaling.
That's the bad news. The good news is that we're actually making progress, and we're able to make those work, and those are the ones that deliver the actual cost reduction. If we wanna make the area smaller, eventually the transistor is going to have less room. If you wanna get more performance out of it, one way with this device now is to make it taller. We also made the fin taller, and that provides more current carrying capability within the same area. First you make them closer together. They're gonna actually get smaller in the direction into the board. To make up for that, you make them taller. That combination allows us to make a higher performance device, and while that may seem simple, making those kind of changes at these small dimensions is actually very difficult.
The other thing that we did is the transition from a standard planar device to the FinFET or Tri-Gate devices results in a lack of flexibility for the circuit designer. In the case of the planar device, they can largely control the size of that device. You know, there's X and Y dimensions that they can pick and choose from. With these devices, they can't change the height. That's set by our technology. They can't really change the size. All they can do is add or subtract fins. You've turned the design process into a quantization of individual transistors that you can basically gang together. If you wanna get down to the smallest area and you have a small device, you'd like to use as few fins as possible.
These fins are very small, and the variability of them is a key problem that the designer has to deal with. It is very common not to use the minimum number of fins or even to use 2. We put a substantial amount of effort in this technology to improve the process control, and now we've enabled them to very often use one or two fins. That is a substantial benefit in terms of, again, the area scaling that you can get, and that comes from driving process control. That also then allows you to use a device that is only as big as needed, and that reduces the overall loading on your circuits. That improves the power. Finally, you finish up and put on the High-K metal gate, and the transistor is now finished.
You can see kind of graphically that it's now taller, fewer, and closer together. This is what that looks like in real life. These are cross-sections. They're called baby pictures sometimes. You can see on the left, people are very proud of these pictures. On the left, you see what was a leadership device as little as two years ago, and on the right is now what we think it takes to produce a truly leadership device today. This has provided us with that fundamental improvement for the 14 NM technology that will lead to improved product performance. Let's just take a look back at the timeline. We have said that we have a lead, and we think that lead is still a defendable statement.
If you look at the first introduction of various technologies, I won't focus on the older ones, let's look at the Tri-Gate or FinFET device. We started shipping that device in the end of 2011 with our 22 NM technology. We are now shipping this second generation device in volume in the end of 2014. The earliest statements for anybody to claim that they're going to ship production quantities or quality FinFET devices will be sometime early next year, maybe a little later. We've put it early next year. That still is 3.5 years that we have built in experience and also shipping.
I'm obviously not gonna tell you what the next innovations are, but our roadmap is full because to continue to improve transistors, you have to make substantial improvements, and we plan to do that. While other people are working on perfecting their FinFET devices, we're gonna be moving on to looking at what comes next. Last year, I also showed you this graph, and I used it to indicate that different segments of our business want different kinds of performance or power. But in the end, it all comes down to what can you provide in a performance per watt improvement. The graph on the right-hand side here shows, and at the time, the 14 NM was starred because we were projecting, we can now take and add what did we actually get out of the Broadwell product line.
If we look at the performance improvement measured in performance per watt, the product almost delivered 2x. Now, the line and the slope that we have been projecting and have for a long time is about 1.6. That's the fundamental technology. The graphs that I showed were based on monitoring and analyzing transistors. When you get to the product level, clearly that's a factor. There's also all of the innovation that goes into the product, all of the improved power management and other things. I'm not implying that the 2x came from more technology improvement, but what I am saying is that we were able to deliver better than the curve between the combination of our product development and our technology. That line, which last year we were projecting would continue, truly has been delivered on.
As the year goes on, Brian indicated, you'll see products coming off of 14 NMs in different segments. For now, the statement that I made a year ago of enabling all different segments can only be validated with this first part, but I think it demonstrates clearly that the technology is able to provide the kinds of benefits we were looking for. Let's look then at interconnects. This is the other critical part of a technology. I showed you on that first page, the scaling that we were going to do for the lowest levels, the tiniest lines that were there. Here's a cross-section of the 22 NM and the 14 NM. This particular picture we typically call the Gordon Moore foil.
Years ago, he wanted to see cross-sections of the whole stack of the technology, and so there are specific little test structures put in the chip, in the scribe line area that you can cross-section very easily to get this particular kind of a picture. Obviously, Gordon doesn't care about it anymore, but the basic pretty picture still exists. What you can see, maybe you can see, but what you're gonna see is that things are smaller, tighter together, and that allows for the scaling that we're going to need to gain that cost reduction. What you can't see at this scale is that if you look very carefully, those little black areas, those are the air gaps. Now, the reason air gaps are important? Can you see them?
Yes.
Well, let's help. There you go. All right. Every generation, since probably the 130 NM, there's been substantial effort within the equipment and the material and the Intel world to change the dielectric constant of the material that goes between the metal lines. Now, why that's important is, as these lines get closer together, they interact with each other, and these lines are carrying signals that are going in multiple directions. What drives most of the power that, as the chip consumes, is moving these lines to transfer data across the chip. The closer the lines get together, the harder it is to make them move independently. That's caused by the capacitance between the two lines, that's a function of two things, how far apart they are. Obviously, we're making them closer together.
Then the dielectric constant or how the material between them causes them to couple to each other. We've done many things to try to reduce this. In general, this was glass as a particular dielectric constant. We've done many things, the equipment suppliers have helped dramatically to reduce that by probably 30% over the last decade. The ultimate that you can get, unless you make it a vacuum, is air. It's actually one on that scale from three to two something down to 1. By adding little holes in here, we're actually reducing that coupling effect. When you put holes in the middle of your technology, you also make it a lot weaker. The process of putting these holes in and making sure they're not filled and they're only the certain places is actually quite complicated.
It has enabled us to scale these lines together and have less of an effect on the performance. This been talked about for a long time, similar to Tri-Gate or FinFET devices. The industry has talked about air gaps for a very long time. This is the first time that we'll be putting air gaps into our technology. We also think it's probably the first time that anybody is going to ship product that actually has intentional air gaps. Sometimes you end up with air gaps when you don't want them, okay? That wouldn't be so bad, except usually the technology is not prepared to have holes in it, and it then falls apart. In this case, these are intentional, and they do provide a substantial benefit in the performance that we can extract.
If I take transistors and interconnect and take a look at a very simple metric, on the side you see a little picture, and it shows a little rectangle. In one direction, gate pitch decides the size of that box, and in the other direction, metal pitch. If you multiply those together, you will get an area, and that area factor for that simple little structure, which basically allows you to make a transistor, is the basic building block. I'm choosing to use this this year because recently, a number of people have published data on that. Samsung published data on their projected 10 NM technology at VLSI. TSMC published data last year in December at IEDM. For those parts that are already available on the market, you can easily cross-section them and go measure these numbers.
You can actually go extract this data for yourself. That's the vertical axis on this graph. On the horizontal axis, we've attempted to indicate where did a particular technology go into volume production. And this is obviously somewhat subjective, but I think it's a reasonable approximation of where most of these technologies went into production. In the white line are the Intel technologies, and in the green line are various best cases for competitors that we were able to extract data for. The 14 NM point here makes the point that not all 14 NM technologies are equal. There is publicly stated, no scaling going on. While people will be shipping 14 NM technologies less than two or three years behind us, what they're shipping is actually more in line with the 22 NM generation. In their own case, it's exactly the same.
In fact, it's a sales pitch that it's gonna be very easy to make that migration. That's a very reasonable thing to be doing because 20 NM technologies with planar devices have very serious transistor challenges. Generally speaking, higher leakage, higher power, very difficult to get those parts into the range that you would've liked to. We made that transition at 22. Our 14 NM technology clearly has that capability. We were able then to move on and work on scaling in addition to improving the transistor. To help your eyeball, if you kind of draw a dotted line between these two points, you see it's basically following the trend line pretty clearly. This trend line is roughly that kind of area scaling that you need to generate cost reduction in the product family lines.
We think that we have a substantial lead both in transistors and in our ability to deliver scaling into the marketplace. Why is that important? I also showed you this graph last year. In this graph, we were talking about the impact of capital cost and wafer cost increase. On the left-hand side is, in fact, that chart, and while it's hard to see on this scale, the last two data points have a substantially higher increase than the ones before them. Prior to that period, we're looking at roughly 10% kinds of increases in wafer cost. More recently, those numbers are closer to 30%, and that generates a lot of stress in the cost reduction. What we said was that we would continue to maintain the cost reduction by doing more scaling.
That's why I've emphasized you have to make the transistor smaller. You have to actually scale the Internet. You have to be able to make that little box get smaller so that everything that gets on top, built on top of it, in fact, can fit into a smaller die area because the little piece of area that you're going to build it on is actually getting more expensive. We're continuing on that trend. This year, we've taken the little star off of 14 because we are in production. This graph on the right is mature cost. We're not mature yet, clearly, the yields have to match for the cost to align.
We're well on track to that, and the fundamental built into the technology, when we get yields matched and we're in full production, will allow us to achieve this data point, which is slightly below this historical trend line. You say, "Well, okay, that's a lot of data again." Let's take a look at it in terms of what did it actually deliver with the Broadwell product. Here we're comparing the Haswell 2 + 2, which is the product that generally drives our Core i3, i5, i7 product line, and it shows the transition to Broadwell for the same 2 + 2 configuration. The Broadwell die is 37% smaller. That's pretty impressive. It's not 2x, but it's 30%. However, it also has 35% more transistors than the corresponding Haswell die, and that leads to this 2.2x increase in transistor density.
That is better than our 2x or half area. Okay? The combination of the technology actually does show itself in the actual product results. You can see that the Broadwell part provides clear area scaling. That area scaling enables the cost reduction that we were showing. It will also have higher performance and probably most significantly to many of us, it's going to have lower power and actually enable many of these form factors to see fanless 2-in-1 devices because the power reduction has been so substantial, driven and enabled by the product design and the new transistors that we put in the technology. One last little data point.
We're not ready to put a data point on for 7 NM, but we clearly are looking seriously at what that technology is going to look like, and we're very confident in saying it will be below the line. I've made the dotted line go out. I'm not ready to put a data point on it. I don't really wanna set that target. This long-term trend line, which is the fundamental economic driver of Moore's Law, is what we are so committed to continue to extend. While it will be a while before you see data on seven or even 10, we are quite confident that at least within this range, we can continue to deliver the promises of Moore's Law. If I summarize, we're still committed. Brian said it. It's one of our key corporate objectives.
That's really encouraging to the technology people that I represent, and it's our job to continue to deliver on Moore's Law. If we do that, we can continue to provide true cost reduction. That cost reduction enables more capable and therefore more valuable to the customer products. Without that cost reduction, you can build better products, but someone's gonna have to pay for them. Finally, because the technology is not just about cost, because we continue to innovate in the fundamental building block of the transistor, we believe that we can make real product improvements beyond cost and make that combination deliver truly compelling new devices. With that, I think I'm done.
Please welcome Stacy Smith.
Good morning, everybody. It's actually unusual to say good morning. I'm usually saying good afternoon. I also wanna just share my welcome. It's great to have y'all here. After the fact that we've already done the guidance, I was afraid many of you wouldn't come back. I was hoping that you'd at least come back for the haiku. I did do a haiku last year, I'm not gonna even take the 60 seconds of just reflecting on a great year. If you go back to that haiku from a year ago, I would like to say it was incredibly insightful. Came true almost exactly. It kept me out of trouble with the boss.
If you remember, I sometimes share jokes, and they sometimes get me in trouble with the boss, so I'm gonna stick with the haiku thing this year. I also, as I was reflecting on the 50th anniversary of Moore's Law, I realized that I have a little bit of law envy. You know, you have Moore's Law, you have Metcalfe's Law, we've shared with you in the past Simpson's Paradox. I don't wanna get ahead of anybody here, but you will hear a lot about Jevons Paradox today. You're gonna be very well educated in it. I couldn't come up with law that rhymed with Stacy, so I came up with Stacy Shroud, which is the art of using poetry to hide deep insight in plain sight.
It is a more creative form than the more convoluted Greenspan Garble. Without further ado, let me share with you the haiku, and after you see this, you can all leave 'cause now you know everything that I'm gonna say. This plus the guidance. The haiku, the guidance haiku for this year is that cloud forms, data shines, transistor density sprouts, client growth blossoms. Pretty majestic and insightful. If I do say so myself. Again, between that and the fact that we released guidance already, I won't be disappointed if you wanna take the hour off. I do wanna come back to the law thing.
'Cause if you look at those laws, you realize that they all have a sense of something doubling over time. There's a corollary to Stacy's Law, which is the number of haikus double every investor meeting. You're gonna get two haikus today. The corollary to the corollary is it's the second haiku that's likely to get me in trouble with BK. First, I wanna tell a little story about him. I don't know if you know what a true technology user he is. This man is deeply engaged in technology. As the people who run product divisions here know, he takes their pre-beta stuff, tears it apart, and then tells them everything that's wrong with the technology that's about to come to the market.
He has implemented a Intel CitySense sensor at his house, which is a citywide air quality monitoring system that's now running at his house. He builds stuff. He's big into the maker movement. He flies drones around and then sends us videos of the drones doing cool stuff like cleaning out his gutters, which is, this is true, a pretty cool thing. Without further ado, I'm gonna show you the second haiku. This is the one that might have me spending more time with my family. The second haiku is in honor of BK. It's big drones, cool robots, intelligence on the edge, BK spawns Skynet.
I'll also say that if you want to get the graphics department really excited and working on your stuff first, tell them you need graphics from Terminator. Between that and a five-hour ENERGY drink, they get very, very excited. All right. I have four relatively simple key messages today, all of which you should have gotten from the haiku, not the Skynet one, the other haiku. First, we're growing again. 2014 is turning out to be a better year than we thought at the beginning. We're seeing broad growth in the data center, in PCs, in tablets, in IoT, LTE, and NAS. Really across our business we're seeing growth.
Second, our core competitive advantages, the leadership IP portfolio that you've heard a lot about today, and leadership manufacturing that you just heard Bill talking about, are becoming increasingly valuable competitive advantages, and they're also becoming increasingly rare competitive advantages, and I'll build on that a little bit. Third, we're in a great position to benefit from this build-out of the cloud and data analytics. You're gonna hear a lot about that from Diane later today. Last, we continue to execute to our strategy of both investing in our business and in generating return for our shareholders via the dividend and the buyback. You saw at the break that we just announced the increase to the dividend, which I will talk about. Just a very quick retrospective back at 2014, or at least an in-process re-retrospective. We're seeing strong financial growth.
It's turning out to be stronger than we thought, we think we're on track for 6% revenue growth this year, and operating profit being up 25%. That's true both year to date and likely to end the year somewhere around there. We are seeing broad-based revenue growth. PCCG is up about 5%, DCG is up about 16%, IoTG is up about 18%. Again, you're seeing that strength across different Intel businesses. As you heard from Brian, and you'll hear more this afternoon, our investments in mobility and comms are paying off, with increases in share, increases in footprint, and some really big, important design wins that we're winning. I just wanna take a quick second to look at a longer-term revenue trend.
After two years of decline, we're growing again, and we're on track for this to be our best revenue year ever. Today I'm going to organize my talk by talking about our two key competitive advantages, that leadership IP portfolio and leadership in manufacturing. I will, as usual, talk about the various segments of our business, and then our strategy for returning cash to our shareholders. I'm going to start today by talking about our leadership in IP. You know, just to build on what you heard this morning. The highest performance, lowest power, least expensive way to bring a capability to the user is to integrate it into silicon. That has been true, and it will continue to be true.
When we do that, we can use Moore's Law to either double the performance, or halve the cost as we go from generation to generation. As you saw from Bill, we invest the Moore's Law benefit to do both. We bring die size down, we bring performance up. To integrate, you have to have those critical IP blocks. The thing is, to have a deep and broad portfolio of IP takes very significant investment. This chart shows five years of R&D investment for Intel versus other industry players, you can see that we're investing at a multiple to everyone else. We're doing that to build a very broad range of leadership IP. The reason we're doing that is simple. If you have this broad range of leadership IP, you can integrate.
That means that we can do products that are better than others, products that are more competitive in the market. Looking at this chart, you can get a sense of the scale that's required to build a portfolio of leadership IP. Remember, when we invest, we're one of the few IDMs left, integrated design and manufacturing, you know, we're investing in our process technology, and we're investing in IP. If you really wanted to get a fair comparison, you might take the investment, say, that a TSMC does and add it to somebody else, you can still see even with that, we're investing at a multiple to others in the industry. This chart shows the inverse. It shows what happens when companies no longer have that scale to invest.
What you're seeing here is you're seeing how we've seen a consolidation in critical IP blocks over the last eight or so years. If you start on the far left, this is looking across wireless WAN, graphics, and connectivity. We could do this across other IP blocks as well, but these are the ones that we chose because they're really critical to the platform. If you start on the far left of the chart, what you see is that 18 companies had complete modem stacks in 2006. Today, we're down to seven companies that compete broadly in the modem business, with only Intel, Qualcomm, and two others currently shipping commercially available LTE.
In the middle chart, you see that there were 16 companies that were selling graphics IP blocks in some form or fashion in 2006. Today, it's down to only seven, again, with only Intel and two other companies that have the range of IP that span from low power through high-performance PC-grade graphics. Similarly, on the right-hand side, you see the wireless connectivity. There's only a handful of companies that have that full suite of connectivity of Bluetooth, Wi-Fi, and GPS. 2 observations from this chart that I think are important. First, you need broad IP to be competitive in any one market. You can be locked out of markets or much less competitive in a market if you don't have the right IP blocks.
Second, the investment that's required is significant, so you have to service large end markets to make the economics of this work. I'm gonna build on something that Brian showed that is really showing the way that we get scale, which is by using the IP that we develop for the PC to go across other segments of our business. I'm gonna take a minute and just build this out. Again, what this shows is the IP blocks that we develop for our PC products. By and large, you should think of Intel as everything that we develop supports our PC business in some fashion. This image then shows how that IP becomes the foundational building block for our server products.
Much, maybe even most of the IP that we use, in servers is either developed or modified from our PC product line. Let me give you a sense, a couple of examples of how this works. The CPU core in server really is a direct pull from our PC products. I think many people don't really know that. Server would just have more cores to drive, the performance that's required in the high-performance compute segments of the server market. Audio, display, USB similarly are pretty much direct pulls of IP that get reused across the product family in servers from stuff we develop in PCs. In other cases, we modify the IP. The memory controller is a good example of that.
Some of the base technology is in the same. In the case of server, we have a product that has more memory channels and uses a higher performance DRAM. It's base IP that we develop for the PC and then modify for the server. We both pull and modify IP as we go from PCs to server, which lowers the cost to be in the data center business. And it's one of the reasons that our data center business is so profitable. And you see that across other verticals where we've been in the business for a long time. As we drive to low power across our product lines, you can now see that same dynamic as we're developing lower power IP for the PC. Think of this as like the Atom core.
You can see that same dynamic play out in our mobile products. Again, much of the IP for our mobile products is either pulled or modified directly from what we do for PCs. Now we're finding, as Brian said, that some of the low power and connectivity IP that we develop specifically for the mobile segment is finding its way back into PCs. It's now becoming this virtuous cycle where IP developed for one platform finds its way into a different platform and makes that product better. This ability to share IP is really important for us. It allows us to play in large markets. For example, building on what Brian said, we think that over the next few years, approximately 20% of the PC market and more than half of the tablet market is gonna require wireless WAN connectivity.
We'd be locked out or much less competitive in those segments if we didn't have the IP to integrate. Now looking at how this plays out, we invest about $11 billion in R&D. As you can see from this chart, about 60% of that is in foundational capabilities. More than half of that is spent on things like process technology and developing the base cores, like Xeon, Atom, Core cores. The next block on the chart shows the investments we make in these shared IP blocks, so Brian took you through that. Examples are wireless connectivity, wireless WAN, graphics. With those, you see that about 70% of our investment is common across our businesses.
The last 30% is investment that we make to develop specific products for our business, so PCCG and DCG being the largest part, then MCG, and then relatively limited investment specific to the products that we're developing for things like IoT and some of the other businesses. An observation on this chart, the overall investment that's required to have a broad range of IP is enormous, but the specific investment to participate in a specific market segment is relatively modest. Our strategy is to take advantage of the IP that we develop for the PC to give us a lower cost of entry as we go after other market segments. A great example of this that you can see in our financials is I talked about DCG before, but if you look at the IoT group's profitability, you can really see this playing out.
We reuse a lot of IP in that business. You'll hear some of this from Doug. In fact, most of the IP that we use in IoT is reused from things that we're developing for the PC business, and that has a direct impact on our profitability in that segment. Okay, I'm now gonna talk through some of the key focus areas in our business. I really think of our business as two ends of the internet. We have the clients, and IoT kind of at the edge, and we see more and more intelligence moving out to the edge, and they're all connecting back to the data center. In addition, we have large businesses in software and in memory.
I'm gonna go pretty fast through this section, and in some places actually faster than others because we have focused topics on these businesses later in the day. I may give you just a quick run-through of some of the financials, and then you'll hear a lot more from people like Kirk and Diane as they talk in great detail about their business. I'm gonna start with the Data Center. Just a quick retrospective on this year's performance. We think this business is gonna be over $14 billion in revenue this year, we've seen really nice growth over the last several years. That's up 16% year-on-year. We're on track for this business to be about $7 billion of operating profit, which is up 26% year-on-year. Great growth going on here.
The combination of the build-out of the cloud and data analytics is driving significant growth for us in the data center. The mega trend here is that there's this explosion of devices, a bunch of new applications, an enormous amount of data that's being collected and analyzed, and all of this is driving this secular trend as computing capability and the associated storage and networking is being deployed to deal with those underlying growth drivers. I think we are uniquely well-situated to benefit from those trends, and you can see that playing out in our results. The combination of cloud, telco, high-performance computing, so kinda everything that sits above enterprise, is growing well over 20% year-over-year. By the way, do you like that tagline of cloud plus big data equals big deal? No, Diane's shaking her head.
When she saw that, she said, "I'm really glad you're CFO and not in marketing," I did like it. It may not be a great slogan, I'll acknowledge that, but it actually is a really strong business driver for us. We expect the DCG growth rate as we go forward to 2016 to be slightly above the 15% CAGR that Brian talked about, a little bit of a tick up from what we're seeing this year. I'm sorry, consistent with what we're seeing this year. We expect that operating profits to grow a bit faster than revenue next year. I'm now gonna talk about our PC Client Group and our Mobile and Communications Group.
As you've heard over the course of the morning, we think about this as a continuum of clients that spans from, you know, high-end PCs down to tablets with a lot of fuzziness in between now of the different form factors. In order to stay very transparent and specifically to be able to show you how we plan to reduce the loss in MCG, I'm going to talk about the financials today in the current reporting segment so that you have a clean baseline to work from. Here you see on the left-hand side, the PC Client Group, on the right-hand side, MCG. The PC Client Group has exceeded the expectations that we had at the beginning of the year. They're on track to grow about 5% in revenue and operating profit growing 25%.
As we've talked about over the course of the year, we've seen both the enterprise and mature markets growing, relatively strongly, and we're achieving very good costs. MCG is down in revenue, and with a larger loss than we saw last year. This is driven by the factors that we talked about a year ago. We're seeing a reduction in 2G and 3G volume, and we're seeing that the bill of material offset dollars that we've been providing to our customers as contra revenue is driving a bigger loss. I want to say upfront, this is a very large loss. As Andy said, it's not something that makes us proud. I will talk about what we're doing to improve the profitability of this segment.
That said, we're making very good progress in the strategy that we had of building our footprint in tablets and growing share and catching up to the market leader in LTE. I'm gonna talk more about all these business drivers over the next few slides. I just want to give you a sense of some of the high-level trends in our PCCG business. After an unprecedented two years of volume decline, we're growing again, and we're actually growing at a pretty robust rate. We think that the growth rate this year is on the order of 9%. Switching to ASP, I love showing these kind of longer term trends. That's one of the things I really like about the investor meeting. You can see that our 2014 ASP trends are down slightly versus 2013.
If you look at this on a longer term trend, what you see is that it's been pretty flat. If you looked at these curves, going back in time, so up until about four or five years ago, what you'd see is that we would expect a normal ASP decline of about 5% down year-on-year. You can see that that really has changed. The thing that has driven that to change is this combination of technology leadership and us segmenting our product line, and I'll talk more about that. That really has allowed us to bend the curve here. Now switching to cost per unit. Here is a long-term cost per unit trend. You can see that Moore's Law results in our ability to lower costs over time.
You can see that this year we're on track for a nice reduction in costs. It's due to really a couple of things. 1 is, as Bill showed you, the 22 NM yields are as good as we've achieved or better than we've achieved on any process. We're seeing a higher mix of some of the lower cost products like Bay Trail this year, and I'll talk about that coming up as well. This chart now puts both the ASP and the cost on the same axis. I have to say, this is one of my favorite charts because it really shows on 1 chart the margin expansion that we're seeing as a company by achieving our segmentation and via our manufacturing leadership. Let's dive in a bit.
Some of these are old favorites of yours, I know, and I'm gonna update them in terms of how we're segmenting our costs over time across different segments of the PC market. What this chart shows is our average cost for performance, mainstream and value. It shows actuals through 2013, and it shows what we expected for 2014 a year ago in the investor meeting. Just a quick refresh. You can see here a couple of trends. Our costs come down over time, and that we're achieving a nicely segmented cost structure.
Specifically for 2014, you can see that in the performance and mainstream segments, we expected that costs would be pretty flat, and the phenomenon that we told you about last year that was driving that was as 14 NM-based Broadwell products came into our product line, it would cause kind of a flattening out, maybe even a little bit of an uptick in that in that cost trend. Here's what we actually achieved in 2014. You can see that across the board costs were better than we expected, really as a result of the two factors that I've been talking about: great cost on 22 NM, and our 14 NM products were a smaller mix. We had more of a mix to 22 NM in 2014. A smaller mix of 14 NM versus what we expected.
Here is the forecast for 2015. You can see that we expect an increase in costs as we ramp 14 NM process technology and as those products become a bigger percentage of the mix. That's a pretty normal phenomenon for us as we ramp a new process technology. Specific to 2015, the costs here are a bit higher than what we normally expect as a result of lower yields in the early part of that 14 NM ramp. Bill showed you the yield curves and how we're catching up as we get into the back part of this year, first half of next year. It's very consistent with what he showed you. I do wanna dive in a bit on the 14 NM cost curve.
What this shows is it shows Broadwell, and it shows the cost trend by quarter or so. It kinda starts it in the quarter after we achieve 1 million units, which happens to be Q1, and it shows how that cost trend comes down over time. It doesn't just come down, it comes down actually at a pretty steep rate. Again, that phenomenon is pretty common for the first product on a new process technology. Now I've added to the chart the comparison to Ivy Bridge and to Haswell at the same point of their respective ramps. As a reminder, Ivy Bridge was the first 22 NM product, and so that's the one that's synonymous with Broadwell.
Haswell was the new core that came later on the 22 NM process technology, that's the product that's in high volume manufacturing today. Bill showed you that the yield curve on 14 NM as we're catching up. This cost trend reflects that. The Broadwell cost in the early part of the ramp is higher than we saw with Ivy Bridge because the yields have been a little bit lower. You can also see that by the time we get into the second half of the year, we expect that Broadwell is gonna have a lower unit cost than we saw on both Ivy Bridge and on Haswell at the same point of their respective ramps. That's Moore's Law at work, I'm gonna talk a bit more about that on this slide.
I'll talk about more of that in a second. I want to switch gears here for a second and talk about segmentation, because that's really important. What this chart shows is how we're segmenting our PC product line. You saw that the ASP curve has been relatively flat for the last several years. A big part of that is because we're seeing growth in the high end of the market. You can see in this graph that we're seeing unit growth in our core product line. The top part of this graph is core. The bottom part of this graph is Pentium and Celeron. In 2014, the core segment of our PC product line grew by about 5%, so nice growth at the high end of the market.
You can also see when you look at the right side of the top part of the chart, that the mix in Core has moved more towards i5 and i7. This is happening because of the performance and the features that we're bringing into the high end of the market. Kirk's gonna talk a lot more about that later today. At the lower end of the PC market, we're seeing even more robust growth. What you see there is that the low end of the market, Celeron and Pentium in PCs, grew by about 14%. What's notable here is that we're bringing that lower cost Bay Trail product, and that's allowing us to lower system price points, which is enabling us to gain share.
It's not just gain share from the traditional IA competitors, but we're gaining share from ARM-based tablets, and Kirk's gonna talk some about that as well. The value proposition is clear. For a few hundred dollars now, you can go buy a touch-enabled sleek new device, and when that's compared against somebody wanting to replace maybe a newer tablet, the value proposition's clear, and they're buying more and more the PC. The next page is gonna show how cost becomes that really important tool for us to gain share. We've showed you this page before. This is an update to it. What this shows is our lowest PC client Celeron costs. You can see here how in 2014, Bay Trail came in, replaced Sandy Bridge, and lowered the cost by about 40%.
That we expect another 30% reduction in costs as we go from 2014 to 2015. The cost improvement here enables us to grow the low end of the market and gain share. Having a segmented product line generally, when we think about the high end and the low end, allows us to bring features and performance at the high end of the market, grow that segment, lower costs to the price-sensitive segment of the market. All of that is enabled by our manufacturing leadership.
The ability to segment our product lines and utilize Moore's Law is enabling us to grow share, grow the business, and still maintain an overall gross margin for this year that's 63%, and a gross margin next year that's forecasted in the high end of the range. I'm now gonna shift to talk about several trends in the MCG segment, and I wanna start with multi-coms. As you heard earlier, we're in a very small group of companies that are selling commercially available LTE data and voice modems. LTE is ramping now, and more significantly, we're winning designs across some of the biggest players in the industry. I won't say more on this other than we're ramping today. We expect this business to grow next year.
The vast majority of the business for us next year will be on LTE versus the older platforms. Hermann's going to talk a lot more about this in his section. I'm now going to talk about tablets for a second or for a few slides. This chart adds volume to the PC trend that I showed you earlier. It's just reinforcing we're on track to the 40 million tablets that was our goal. This will make us the largest merchant supplier of silicon, the second-largest manufacturer of silicon that's going into tablets. The unit growth inclusive of tablets is almost 20%. When you include the tablet unit growth, it's one of our best growth years that we've seen in a long, long time.
You can see from this chart how we're building this footprint of Intel architecture-based devices. You can see that we're outgrowing the market. We've bent that curve in terms of share, and in 2014, we think we're gonna gain back some share, so we're actually growing faster than the market. I now wanna go back and talk about the strategy that we outlined last year to build our footprints in tablets and then carry that forward into 2015. As a reminder, Bay Trail was a product that was designed for the high end of the tablet market and also for the PC market. To increase our footprint in tablets, we decided to take it broadly across the various segments of the tablet market in 2014, and we've been successful in doing that.
We're shipping to the 40 million units, we're across operating systems, and we're across price points, which is all very important to us. Bay Trail, as a reminder though, carried a pretty high bill of material delta at the system level because of things like the memory subsystem, the power management subsystem that it required. The Bay Trail bill of material delta was as high as $20 in some configurations, and in almost all configurations was more than $15. We worked that down over the course of the year, so Bay Trail CR came in and reduced that by about half. Even with that, we were taking products designed for the high end of the market.
We were taking them more broadly through all of the different segments of the market, and that was driving a significant cost delta, and it was driving some contra revenue dollars. Early in 2015, we will bring out the 3G version of the SoFIA product family. This product is designed for the value and entry segment of the tablet market, and it has an SoC cost that's about one quarter that of Bay Trail, and it carries a competitive bill of material cost at the system level. That's really important. To be clear, we will still be paying some contra revenue dollars in 2015 because Bay Trail and Clover Trail will be significant volume. They're designs we've won, and they'll be shipping over the course of the year.
For the new SoFIA platform, our expectation is that we will not have a bill of material delta which requires contra revenue support. Bottom line, the SoFIA product line will improve our margin structure on tablets pretty dramatically. Zooming out for a second to talk about just kind of general profitability trends in MCG, and I'll get to a forecast in a little bit. This combination of ramping LTE, so starting to get volume that's paying off some of the investment we're makingImproving our SoC cost, which you can see as we bring out products into the value segment of the market, and reducing the bill of material delta at the system level, is gonna improve our gross margin pretty significantly for MCG over the course of 2015. We're not yet where we need to be, but we're making progress.
And as I think out and look at the product roadmap and where we're heading, I expect that we can get to the point that we're gross margin positive for MCG sometime in the first half of 2016. Not where we need to be, but making progress. I just want to come back for a second to the shared investment model that Brian showed you and that I showed you. When we look at this for MCG, you can see that only a relatively small portion of MCG's spending is unique. The rest is allocated spending for things like IP blocks and the capability that we use across the company, like our process technology, process technology development.
As we improve our product portfolio and ramp the products, we have a relatively small amount of unique spending to cover before we start generating an economic return. Again, this is our shared investment model at work. You can see it today in the data center, you can see it today in the IoTG. I believe this will play out as we gain share and improve our product portfolio in MCG. For relatively low incremental investment, we can go after actually quite large end markets. Getting into some specific expectations for PCCG and MCG for 2015. For PCCG, we expect approximately flat revenue units, or flat units and a slight reduction in revenue relative to 2014.
If I was putting a plus and minus around that, I'd say +1 Gartner, - 1 IDC to give a range. You'll see from Kirk later today that he's not satisfied with this growth rate, so he has ambitions to grow it more than that. That being said, our baseline forecast for next year is relatively flat units, a slight reduction in revenue. Diving under that, we think we'll get slightly less of a tailwind from the enterprise market, a little bit more participation as consumer comes back. PCCG, as I showed you, will have higher costs in 2015 associated with really two things.
One is it's the year we start getting hit with 10 NM startup costs and high Broadwell costs that I showed you. Both of those will come down nicely over the course of the year. For MCG, we're ramping LTE, we're introducing SoFIA, we're reducing the BOM contra revenue support on the new platforms that we're shipping, things like SoFIA. These actions we think will result in an operating margin improvement for MCG that's on the order of $800 million next year. Again, not where we wanna be, good progress year on year. Now I'm gonna very quickly talk about IoT software and services and NAND. One interesting observation as I was preparing for this presentation, as I was looking at them, you know, all three of these businesses have now crossed over $2 billion in revenue.
Against the backdrop of an Intel that's, you know, in the high 50s, they're not that big. Against the backdrop of the universe of companies in which they compete, they're actually quite large. They're all growing and they're all profitable, and I'll just kind of flash through a few of them, and you'll hear more about all of this this afternoon. The revenue for the IoT business is expected to be north of $2 billion at $2.1 billion. That's up about 18% from last year. In this business, we're seeing strength in manufacturing, strength in retail, and strength in the automotive segments of the business. Operating margin for this business is 26% of revenue at approximately $600 million. You get a sense of that IP reuse that's really positively impacting the profitability of the business.
We're investing in this business. You may have noticed the operating margin flattened out. We are very consciously making some investments in this business because we see a huge opportunity as we go forward. You're gonna hear from Doug how he's taking cores that we're developing elsewhere in the company, things like Atom and Quark, and using them to go after segments of the market that require lower power and lower cost. Again, that reuse at work, and he's opening up significant portions of the TAM as a result. You'll also hear how he's taking Wind River and now approaching the market from the standpoint of being able to provide more vertical solutions, which we think is a core competitive advantage. It's those things that's leading to a very robust design win pipeline.
If you look at the yellow line on that's the design win pipeline. In prior investor meetings, we've talked about the fact that this is a business that design win pipeline really is a great leading indicator of future revenue because the design win pipeline here is fairly long. And once something goes into manufacturing, it tends to have a pretty long tail. It's that design win pipeline that gives me confidence that we're gonna see even an uptick in the growth rate on top of the 18% that we expect this year. A business that's been growing nicely and we expect to continue to grow nicely. Moving to software and services. Our software and services business is on track to grow to $2.3 billion this year. It's moving from break even to profitable.
In addition, I think it's notable that we had all-time record in terms of bookings for Intel Security, including the McAfee product line. Pretty strong bookings pipeline for this business. More broadly, I'd say that our software capabilities are becoming an even more potent competitive differentiator. A great example of this is the work that we've done with Google to create the Intel reference design program for tablets. This program allows our partners to bring tablets to the market more quickly with Intel Inside, so Android-based tablets to market very quickly. We think that Intel-based tablets will be first to market on Lollipop behind only the Nexus systems. That's a huge shift. If you have been listening to us over the last few years, you know, you'd see that we were six to 12 months behind on Android-based systems.
We now think that we're first to market. We don't get a lot of direct revenue for this, by the way, but it's a very important competitive advantage for us. Renee will talk more about this in the afternoon. Last for this section, I want to touch briefly on our NSG business. You can see the business is growing nicely, and that it's nicely profitable. Revenue's going to cross over $2 billion this year, and it's generating $hundreds of millions of operating profit. Rob will talk more about that later. I'm now going to shift gears and talk about another competitive advantage, which is our leadership manufacturing. You're going to see me tagging on to a lot of the stuff that Bill showed you earlier and taking it into some of the financial implications.
This is the chart that Bill showed you. It has updated a couple of dots and lines from what you saw a year ago. I do wanna say that when Bill puts a dot on the chart, it means something. It's not just PowerPoint. He agonizes over the dot, and then he agonizes over the dotted line. These aren't representative, they're actually what he thinks he's going to achieve. Just a refresh from what we talked about last year. Advancing process technology is getting harder and more capital intensive. Fewer and fewer companies are able to do it. We're able to offset those increases in capital intensity by improving our density/our scaling, and this is keeping us on that historical curve of improving the cost per transistor.
It sounds simple. It's actually huge for us. This is the heart of Moore's Law. It's what allows us to keep bringing more features, improving performance, and reducing costs. It's hugely important. What Bill showed you in his presentation is that we have line of sight of in plans for 10 NM. In fact, that 10 NM dot is slightly better than the historical trend. Based on everything we know about 7 NM, we believe that we can stay on that historical cost curve. Now I just wanna take a second and show you some of the details behind those cost curves. We've shown this to you in the past. Again, it's updated. The first block on that prior slide is that capital intensity is increasing.
It was a little hard to get the sense of it from the scale. What this is showing is our actual expectations of capital intensity. It's actual through 22 NM, and then what we expect for 14. You can see here that 14 NM capital intensity is up about 30% from 22 NM. Starting at 22, there's been an uptick in the curve in terms of how fast capital intensity is increasing. That's happened because at 22 NM, we really started to see multi-patterning and lithography, which is driving this. One other observation on this chart, which I'll come back to in a little bit, our unit growth CAGR has slowed.
As a result, you can see that our peak wafer starts per node are actually down a little bit from 32 NM. When I put that on a scale of how much capacity we're putting in place because of the increasing capacity we get, you'll see it's a pretty noticeable amount. The second block on that earlier chart showed that we're able to improve our density and offset that increase to capital intensity. The practical impact of that is that our die sizes come down. This chart shows the actual weighted average die size for the company through 2015, and you can see that it comes down nicely over time.
In fact, you can see that starting in 2012, which matches when Bill showed you we started to increase our scaling and our density, that we've actually seen this curve start to come down at an even faster pace. You should note, however, that this curve can be impacted by mix. For example, if we do more server volume, our, as a percentage of the total, the curve flattens out some. If we're doing more Atom, the curve can accelerate down. Bill and I are both showing you this slide. We'll make slightly different points, but this chart, this comparison of Haswell to Broadwell really isolates that scaling benefit because it's looking at our Core that we're developing at the higher end of our market.
What it compares is the 2nd Generation Haswell Core versus the 1st Generation 14 NM Core, which is Broadwell. You can see, as Bill showed you, Broadwell is 37% smaller than Haswell, even though it has 1.3 billion transistors versus less than 1 billion transistors for Haswell. In addition to that, you get these nice performance and power benefits. It's sub-5 W, so it fits in a fanless system. It has great graphics performance, so we can continue to bring new features to the end market. To tie this together, and to tie together two things I've just showed you, our 14 NM capital intensity is about 30% more capital intensive than the prior generation. This product has a 37% smaller die size. We're getting the higher performance, more energy efficiency.
This is the advantage that we get from our manufacturing leadership, and this is why it's important for us to continue to advance as fast as possible as we go from node to node, because we get this kind of advantage in our products. The other way to look at that advantage that we get by advancing with Moore's Law is that we can increase unit capacity without having to add more overall loading capacity. Again, as you shrink the die, you get more die per wafer. You can see that we've slowed down that addition of capacity now consistent with the peak wafer starts per week. This chart also shows the relationship between the capacity that we put in place and loadings, and the difference between those two lines is the utilization of our factory network. Two observations on this chart.
First, the obvious one, which is Moore's Law is letting us grow our business at a lower cost. The second one is you can see in the chart, in terms of how we've tried to match capacity demand, that we're trying very hard to be tactically responsive to changes in demand. You can see that, you know, as we hit the financial crisis and loadings dropped, with a relatively short period of time, we were able to get those lines back into alignment by rolling capacity forward. We can now do that in a couple of quarters, and before it used to take us a much longer period of time. I have to bring back tech for a minute here.
Maybe this will be the last year I have to use them, I know that we still get tied up here in, "Okay, I get that, but you're putting in less unit capacity, or, you know, minimal increases to unit capacity. Why is it $11 billion of CapEx as you go from, why are you spending $11 billion of CapEx?" The simple answer is, that's what it takes to advance to the next manufacturing node. The slightly more complicated answer starts with a breakout of CapEx, and this chart is updating information that I showed you last year, and it shows that about $7 billion of the CapEx that we spend is for factory capacity.
I showed you this last year for 2013, it's not terribly different. The other two big chunks of the pie are the spending that we do for R&D. This is the spending that Bill does for kind of advanced research on next-generation processes, and for things like office building, the engineering computing environment that for a company like Intel is actually a fairly big outlay of CapEx, for which we think we get a very high return. I'm now showing you a four-year trend. This is the first time I've showed you that. In fact, you can see that the trend has been pretty consistent back over the last several years. We've been spending at or slightly less than $11 billion per year.
The average for capacity-related spending is, you know, somewhere between $7.5 billion-$8 billion of that total spending. The way to think about this is it takes $7 billion-$8 billion of CapEx to move from node to node with minimal increases to overall volume. Over this time period, our unit CAGR is kind of in the low single digits up. You can tell from the earlier slide that we're not ahead of our skis in terms of getting unit capacity in place. In fact, the utilization of our factory network actually is quite high. On top of that spending then, we spend another $3 billion or so of CapEx on things like R&D and non-manufacturing. I'll come back to those numbers in a second.
Just quickly, then reinforcing the guidance for 2015, we're expecting CapEx to be about $10.5 billion, down a little bit from where we are this year. And in general, the same general distribution across manufacturing, R&D and some of the non-manufacturing buckets. Not a huge change or not much of a change really in terms of the distribution of CapEx. One more text. The other conversation I have with some of you is that, okay, even at that $7 billion-$8 billion of manufacturing-related CapEx, you're gonna see an increase in depreciation, and won't that depreciation impact your gross margins? First, a clarification, only about three-quarters of our depreciation flows into cost of goods. The rest is going into below-the-line spending.
For example, the depreciation for the development factories, by and large, goes into R&D. The depreciation for office buildings, as you expect, goes into G&A. The second thing to note is that depreciation is going up as a % of cost of sales. If you step back, this makes sense based on everything you've heard this morning. Capital intensity is going up, overall costs are not. If you say it another way, this increase in capital intensity is being offset by efficiencies in smaller die sizes, which is translating into us running fewer wafers, and that means that we have less labor expense per wafer. We have less chemicals and gases and direct expenses per wafer. This chart shows that we're approaching a point where depreciation that goes into cost of sales is nearing 30%.
Depreciation as a percent of cost of sales is nearing 30%. I expect that as we start seeing the depreciation for 14 NM and 10 NM moving in, that percentage is gonna go up, and in fact, it'll go up into the low 30s. We tried to do this in a way that's gonna look familiar to some of you, which is we're sitting in a bar doing manager math, napkin math here. This is not meant to be a long-term forecast of our gross margins. This is just meant to isolate the impact of depreciation as it goes up on gross margins at kinda current business size. Yes, depreciation will go up over time to match CapEx. That's just how it works, right?
If you're spending between 7.5 and $8 billion of CapEx over time, the depreciation that's flowing into cost of sales will rise to that number. The rest of the costs will come down as a percentage of the total. When you do this back-of-the-envelope math, it says that even with no revenue increase over the next several years, we just stay at the size that we're at in 2014, as depreciation increases to match CapEx, we're still in our long-term gross margin range of the company.
We're kind of in the mid to high 50s, with zero increase to business size. With a 5% or 10% increase in revenue, and in fact, we're forecasting five just for next year, what you see is that we're in the middle or even the high end of our gross margin range. The thing to take away from this, there's many things that impact gross margin, but this increase in depreciation is not gonna be the thing that takes us out of the gross margin range because, again, depreciation as a percent is gonna go up. We're gonna see savings in other places. That page, again, was a static what-if look, not a long-term forecast. Please, I don't wanna see Intel forecast long-term gross margin in the 50s. That looks at the impact of depreciation on gross margins.
We've demonstrated that we have lots of levers to manage gross margin, I love this chart because it shows kind of a long-term trend of gross margin. You know, the gross margin is impacted by the competitiveness of our product portfolio, how much capacity we're putting in place, how efficiently we're running the factory network, our mix, our product cost. In the real world, we've been above 60% gross margin for four of the last five years, in fact, we're forecasting a gross margin midpoint of 62% for next year. It is down a tad from where we are in 2014, I wanna give you a little bit of the puts and takes there.
We expect that we're going to see elevated startup costs this year. This is the year where we start seeing startup spending on 10 NM. We also expect to see our cost per unit being up a little bit on the year. As I showed you, Broadwell starts out pretty expensive, then comes down nicely by the time we get to the back half of the year. Partially offsetting those two negative drivers, we expect that we're going to see ASPs up a bit in 2015. It's not that we're raising prices. We think that server will be a larger percentage of the mix, that helps. Also, the reduction in the contra revenue dollars will show up as an increase in ASP.
Those two things give us a little bit of an offset to the 10 NM spending and the Broadwell costs. Lastly, to just give you a quick shape for the year, based on what I said, it shouldn't be a surprise, we expect that the first half of the year will be kind of below the 62%. The back half of the year will be above the 62%. Again, costs come down, startup or Broadwell costs come down, and startup costs also come down as we get into the back half of the year. One last observation in this section, similar to what we saw in IP, as advancing Moore's Law becomes more difficult and more capital intensive, fewer and fewer companies have the scale to continue to build their own factories.
I thought Bill did a great job of talking about how it's that partnership between the product side and the technology development side that's allowing us to stay on that traditional cost per transistor decline. We think that that's one of the reasons that others aren't able to do it, because they no longer have the manufacturing side of the house. You can really see it play out in this chart. If you go back 10 years ago, there were dozens of companies that built their own factory. Today, it's just 2. It's us and Samsung. This is the second key competitive advantage for us.
We have this broad portfolio of leadership IP, but we also have leadership in manufacturing, which allows us to do the things we've been talking about, integrate, improve performance, and reduce costs, and do that at a rate that's ahead of the rest of the industry. All right, I'm gonna go quickly through my last and favorite section, which is cash and shareholder return. It's always fast, but that doesn't mean it's not important. What you see here is we continue to generate strong cash flows. This chart shows our cash from operations and our free cash flow. We're on track to generate $21 billion of cash from operations, $10 billion of free cash flow this year. You got to love that.
The priorities of what we do with that cash, I think, have been very well understood by the investment community. They're unchanged. It's first invest in our business. Next, increase shareholder return via the dividend. We target 40% of our free cash flow to return to shareholders via the dividend. Third, we modulate our cash balances and return cash to shareholders also via the buyback. I love this chart. It shows a cumulative 10-year look at kind of how we've executed to those priorities. I think this is the first time we've put this all on one chart like this. You can see over the last 10 years, we've invested $153 billion in CapEx and R&D.
That's the investment we make, and we think that if we do that well, that's how we generate long-term return for you all by making those smart investments and staying ahead of the rest of the industry. From that same date, we've returned, I was really hoping this was gonna be over $100 billion, but we're just to the edge there, $98 billion to our owners via the dividend and share repurchase. I mentioned that 40% of our free cash flow to the dividend, we were a bit above that 40% in 2013.
As we moved into the back half of 2014 and as I look at the cash flow I expect to generate in 2015, we're now, you know, I think we're moving into the 30s, which is what then predicated us coming out with an increase to the dividend that you saw on the break. I'm really pleased to be able to stand in front of you and announce the increase to the dividend. We're increasing it by $0.06 per share, which brings our annual dividend for 2015 up to $0.96. To close this section, I just want to focus on the journey we've been on in bringing down our cash balances and using the buyback to also return cash to our shareholders. We've been repurchasing shares since the early 90s.
Our philosophy, I think as many of you know, is not to time the market, but to just regularly distribute the cash that we're generating to shareholders and manage to a prudent cash balance. We've actually been bringing down the net cash balance that we think we need to maintain based on our responsiveness to the market. Since the start of the buyback, we've repurchased about 4.6 billion shares. Since re-entering the market in 2010, we've brought down our diluted share count by over 0.5 billion shares. Pretty significant activity in this space, as Andy talked about. Just putting it all together, you saw some of this at the break, I'll put it in context now. We're forecasting revenue to be up in the mid-single digits.
PCCG is expected to be down slightly in revenue. MCG is expected to grow pretty fast, off of a depressed base. You have to be a little careful with that math. DCG and IoT are expected to grow at a very fast pace for the size of those businesses. You'll hear more about that. We expect gross margin in the high end of our long-term range. As we discussed, we expect it to be down a little bit from 2014 due to that increase in startup costs on 10 NM due to the Broadwell cost at the beginning part of the year, offset some by ASP as we mix higher to servers and as we see some of the abatement in contra revenue dollars. Moving to spending, we expect spending to be approximately $20 billion.
In terms of that as a percent of revenue, we're down some in 2014. We expect that we'll be down some more in 2015. Under the hood there, we're seeing increases in investment in our Data Center Group and in the IoTG. Those are two places where we see big opportunity. We're investing a bit more in sales and marketing, and you'll see some of that coming out into the market in not too long. Then decreases in investment in MCG and in internally developed SoCs targeted at phones. Last, we're forecasting CapEx at about $10.5 billion, which is down some from 2014. I'm gonna end right here, just reiterating my key messages. We're growing again, and we expect to grow again next year.
Our competitive advantages are delivering real value, and I think very importantly, they're becoming increasingly rare. There's fewer and fewer companies that have the scale to do the things that we do, and I think that'll bring value in the future. We're in an amazing position, I think, to benefit from these trends of the move to the cloud and data analytics. Last, we believe that we continue to be prudent stewards of our owners' capital. With that, thank you very much, and I'll bring Mark back up to introduce the Q&A.
Great. Thanks, Stacy. All right. I think we are about ready to go here.
Yes.
With that, why don't we start here with Gary?
Vivek, please go ahead.
Thank you. Vivek Arya from Bank of America Merrill Lynch. Thanks for all the presentations today. I think, Stacy, you mentioned that. Sorry, I'm here.
Where is he? Oh, there. Thank you.
You expect to be gross margin positive in MCG by early 2016. Would you go as far out as saying that you could even be break even on an operating basis by 2016? The larger question there is, given how the market is maturing in mobile, why hasn't Intel managed to say that, okay, these are the areas where we can be reasonably profitable, these are the areas where we are focusing in? Because I look at the investments you're making today, I look at all the balance sheet investments you're making in China, why can't you be operating even break even at MCG and in some reasonable timeframe? I'll follow up.
Yeah. I guess I'd say that the progression that we're making, you know, we expect an $800 million improvement in operating profit this year. I'd say in 2016 we expect another big improvement in operating profit, probably not break even for the year. I think that happens further out. What you really see the company doing is adjusting our product portfolio and our strategy in terms of how we're going after different segments of the market via partners. It takes some time for that to play out. Good improvement, not yet profitable, and I wouldn't forecast profitability for 2016, although I'd say it's probably a big improvement.
Yeah, I want to add. You think about it, SoFIA will, the 3G in the first half, LTE really mid to second half. Rockchip will have their first product probably sometime towards the latter half of next year at Spreadtrum. We just signed that deal a few weeks ago. The engineers are working. You gotta think their first product probably won't come out till late 2015. Really Rockchip and Spreadtrum kind of start to play a role next year. SoFIA starts to kick in. I think we've made a lot of progress. We've got good momentum, these are gonna take a little bit of time to play out until they're a major portion of our volume there.
I just come back on one thing, 'cause I don't want to lose the point that I was making, which is, you know, what you see in the MCG P&L includes a big slug of allocated costs for things like process technology development, IP blocks. You know, it's just, it's how you do segment reporting as they bear their share of those company-wide costs. When you look at it on an economic basis, so, you know, let's assume we're successful in getting to gross margin breakeven sometime in the first half of 2016, the actual incremental investment that we're making to participate in those segments is relatively modest. Think of it on the order of a $200 million a 1/4 maybe.
You're getting awfully close to the point that we can look you in the eye and say, "Hey, on a economic basis, these things are carrying their own weight." It'll take longer for it to get to the point that it's then covering their portion of, you know, Bill's development costs and the core development and things like that.
Got it. Just as a follow-up, I understand, you know, all the points that, you know, Bill, you made, and Brian, you emphasized on transistor, you know, manufacturing and leadership. There's a big player in Cupertino who has done extremely well with 20 NM and is sticking with that same foundry structure even for 14 or 16, regardless of the disadvantages that, you know, you alluded to for the foundry structure. My question is there something inherent with ARM architecture or design that lets them, you know, continue to do so well in that low power part of the market? Because if PCs and mobile are blurring, then you might be very good in transistors, but is there something in ARM that continues to, you know, make it do extremely well in that part of the market? Thank you.
I'll start, and then Bill can jump in. The simple answer is no. The reason I give you that simple answer is you look at the products that we have, whether it's a phone or a tablet, I've been using the Dell Venue tablet. I charge the thing like every three or four days. I mean, we've proven that we can go into low power. The fact that we have fanless designs here on PCs is another example. The reason for Apple choosing whatever they choose is you really need to talk to them about their strategy and their direction. Our products, as far as being able to do low power, being able to have the performance we want, absolutely not.
Has that been the focus of the company? Have we driven our product roadmaps there? No, other than in the last year and a half to two years, we've really beared down, focused. You see the products coming out. I think you'll see a lot more in the future. We have some great plans ahead for products that operate at very low voltages and powers with great performance.
Bill, care to add?
No. Great answer.
Great. All right. Let's come back over here. Ross?
Thanks. Over here. Ross Seymore from Deutsche Bank. Question on the OpEx side of things. Stacy, you talked about the $20 billion you're gonna spend, and it's great to see that coming down as a percentage of sales a little bit. If I remember right, I thought 30% of sales was pretty much your target for what OpEx was, and by some of the rough manager math, it sounds like it's still down a percent or so, but maybe 34%. I guess the two-part question is 30% still the target? Are you only gonna get there through the revenue growth side of the equation, or are there currently some investments that you're making where investors should think that there’ll be a step down, a market step down to get you to that target margin number a couple years out?
I'd say Brian and I are still committed at 30%. We've been very clear in saying it's a long-term goal and that there's a period of time where we're gonna be making investments above that. If you look at the investment portfolio of, you know, go back a couple of years, this is where we were investing in building our communications capabilities, so we weren't as far behind the market leader, bringing in low power cores and SoC capability, those kinds of things. Today, the investments, as you looked at the shift that I articulated that we're making, is more towards us putting bigger bets on some of the areas of the company that are growing really fast and where we think we have an opportunity to actually get out ahead of the rest of the world.
We're making bets in the data center business. Diane will talk about that. We're making some bets in the IoT business, where we're seeing, you know, growth rates that are approaching 20%. We'll still come down towards that, but it'll take some time, and I do think it's more likely to be seen as revenue growth versus big reductions in spending from here.
I would agree. I mean, I think that's the way you want us to achieve that goal. It would be fairly quick and simple to simply cut spending and get to 30%, but you then would look around and say, "Where's the growth?" Right? As a shareholder, we're trying to balance between those two, and we think we have a roadmap to get there that combines both. You know, we talked about we'll spend less in some areas that may not be growing as fast, like PCs and some of the other areas, but we are gonna invest in those areas we think we can grow at or faster than the market.
Great. I guess as my follow-up, you mentioned the growth side. Transitioning over to that 5% growth target, a lot of it seems to be coming from the DCG side of things, growing again 15%, 16%. That was a great achievement of your target this year, but the last couple years prior to that, you were markedly below it for macro reasons, whatever the case may be. Without front-running Diane too much, because I know she's next. What gives you confidence that you can stay at a rate that is relatively elevated versus any sort of 3-year trend? Thanks.
Yeah. Yeah, if you go back to the years that we're below the trend, it's because we had a very depressed enterprise market as a result of macro trends. If you look at everything I put in my little cloud above the enterprise, which is cloud computing, telco, all of the networking and high-performance computing, that segment of the business has been very consistently growing above 20%. It really hinges on we're banking next year that we continue to see those secular drivers, the build-out of the cloud and data analytics driving that segment of the business, and that we'll see enterprise being kind of a normal year in enterprise. Based on everything we see, that's a rational forecast.
Great. Let's come back over here. I think it's Doug, if I can see him in the shadow.
Great. Thank you.
Great.
Thanks for hosting the day.
Oh, it is.
If you could, I find it interesting that we just heard your presentations, there was no mention of tick-tock. When we look at where you're at, and we look at the year that just occurred, your startup charges clearly came in a little less than expected due to the 14 NM slower than expected ramp. How do we think about what you're planning on 10 NM for next year, and what, if any, would be the gives and takes to your startup charges that you're accounting on for next year?
I can start with just the bigger picture, then Stacy can tell you about charges. You know, we felt like we went out a little early with 14 NMs as far as timing and performance and features, and we saw actually competitors adjust to that. We're gonna be a little bit more prudent, a little smarter about signaling to the industry exactly when, what, and where. And you'll just have to trust a little bit the 50-year history we have with Moore's Law, and that we should be able to keep it going for 51 or 52 years.
53, 54.
Yeah. We're going to be a little bit careful there about signaling exactly when, what, and where. Stacy can talk to you about the timing.
Yeah. I mean.
Spending, sorry.
I would just say on the spending side, it's a fairly normal cadence of spending, and I'm not gonna be any more discreet than Brian was. We do think we've been giving too much insight too far in advance. We'll talk about 10 NM some time in the next 12 or 18 months, and when it's appropriate.
Great. Thank you.
All right. Let's come back over here. Trey.
Hi, JoAnne Feeney with ABR. I wanted to go back to your discussion about the mobile side of the business, Stacy. In particular, the $800 million in operating margin improvements you expect. I was hoping you could clarify what you mean. If this year we're on track to see about $4 billion in losses in that segment, do you actually expect the absolute loss number to drop to something like $3.2 billion? Or are you saying as a portion of the revenues for next year, the percentage of operating losses will drop?
No. Take operating loss minus $800 million. That's what we expect for next year.
Real dollars.
Real dollars. Yeah. I'm not creative enough to even know how to do it the way you said. I wish I'd thought of it. I kind of did it.
It's a bigger number.
I kind of do it the old-fashioned way, which is we're gonna lose X, and we're gonna lose X minus $800 it next year, if we execute to the plan I put in front of you.
As a follow-up, you did slip a comment in at the end about how you're reallocating your research dollars. One of the things you said was that you're going to be reducing allocation of R&D to phone SoC activities. Could you elaborate on what that means for the importance of the phone SoC to Intel's future and how that might be made up through your partnerships?
Yeah. It wasn't meant to be a slip in. It was meant to be pretty specific. What I was trying to signal there is that as Brian talked about, we have a strategy now for the low end of the phone market that revolves a lot more around enabling third parties to go do designs on our architecture. What you're gonna see at that low end of the phone market, you're gonna see more energy happening from third parties and less direct investment. What I meant to signal is we're gonna do more on enabling others, so there's a increase in spending on that side, and less on internally developed SoCs for the low end of the market.
No, here's my simple model, 'cause I always have to break things down into fairly simple terms. We will generate the IP that allows others to go and design. You'll see continued investment. You know, the modem we talked about being highly important, the other connectivities.
Yep.
You'll continue to see same or more investments in those key IPs. The synthesizable core. What we won't have to do is spend as much money on the iterative products that you typically do once you've created kind of the master product, an SoC that has all of these. The partners will be able to do that now, and they'll be able to then take those to market in their channels.
All right. Why don't we come back to the other side of the auditorium? Looks like it might be Jim. I can't quite see you. Go ahead, Jim.
Great. Yeah, thanks. Stacy, a cost accounting question for you. When you show us the mobile, the MCG group losses, there's the dedicated piece out of R&D that goes to MCG. Did those losses include any allocated R&D from the broader chunk of shared R&D?
Sure. In the Mobile and Communications Group segment is what you're asking?
Correct.
Yeah, absolutely.
That $4 billion, there's an allocated piece from the broader group R&D. It's not just the R&D from that one slug that you're showing us in yellow.
Correct.
Yeah. Correct. If you let me say it differently, make sure we're saying the same thing. Pretty much everything that's in that yellow and orange is allocated spending. I'll use my friend.
We'll just spend it whether they're there or not.
Use my friend Bill here as a good example. He spends a lot on process technology development. He's got activity going on in 10 NM, 7 NM, and beyond, right? We have a methodology to allocate that spending amongst the different business units. MCG would pick up a chunk of that. That's actually a large part of that orange bar in MCG. It's in their segment. As Bill Holt was just saying, even if MCG didn't exist, we would still be spending that money to, by and large, or maybe a slight incremental amount, but by and large, we'd be spending that money to develop 10, seven, 5 NM process technology.
Awesome. Thank you.
Welcome.
All right. Let's come over to this side of the room. Looks like Mark Lipacis over in this corner.
Thanks. Thanks for the presentations. I think it's a question for Bill. Bill, I walked in a little bit late, so I'm strategically placed behind the speaker, and I didn't see your chart fully. I think in the one corner you said that 7 NMs you expected the transistor to be on the same or at the historical projection or lower. Is I guess I'm hoping that you can correct me on that if I'm wrong. The question is, do you assume that EUV comes into the market and helps you achieve that?
I think you're talking about the cost chart where at the very end I put a few little dotted points out at the end. That was intentionally ambiguous. Well, what we want to make clear to you is that we believe that the cost reduction historical trend can be continued. We're extending the dotted line on the historical trend. What we didn't say is exactly how far below that line we think we will be and how that trend will be with the previous generation exactly, because that's stuff we don't know. Part of it depends on how mature EUV is and whether we get to use it or whether we have to do that technology without. The positioning of the dot, which we will eventually put on that graph.
Is still in discussion. It's a pretty wide bar, but what we can say is it will be below the historical trend. That much we know.
Excuse me, hence Stacy's comment that when Bill puts a dot on a graph, it really means something.
Well.
When he puts a dotted line on a graph, it means something too.
The follow-up is, can you get below that line without EUV?
Yeah.
Thank you.
All right. Let's come back over to this side of the room. John.
I don't want to. I want to be further below the line. Okay.
Thanks, Gary. It's John Pitzer with Credit Suisse. Thanks for the presentation this morning. Brian, a question for you. To date, marching down Moore's Law has helped you tremendously in your core businesses, and you can just look at the financial model to see proof of that. It's unclear whether or not it's helped you get into these new markets, there's a lot of skeptics in the audience that the hard-fought tablet share this year was just bought tablet share. Is next year the year where your Moore's Law advantage really starts to play out in these new markets? You gave us a target for tablets in 2014, for 2014. Do you have one for 2015, definitional issues aside?
Okay. I see you asked a couple questions. You, you can't argue that Moore's Law necessarily is what got us into the tablet market. Certainly, we had to have products that could perform. I actually think with Bay Trail especially, as it came in at 22 NMs, it was quite competitive relative to the others. Same with Moorefield products. You look at the battery life and the performance, it's very, very strong. I don't want to say that Moore's Law didn't play a role, but clearly, our leadership in Moore's Law hasn't what's led us to our position. Absolutely.
2015 is the first year where you'll start to see 14 NM or leading-edge products in that space, and the first ones will be Cherry Trail. 2016, you'll see probably SoFIA products down there. I think that's 2015 and 2016 is when Moore's Law will start to kick in into this mobile space, and you'll see it increase more and more year after year after that. The second part of your question is, you know, we set a target of $40 million, and what are we thinking about for next year? You know, we wanted to get to a number that was sizable, that was, you know, got us into a market leading position that we could have a footprint, so developers would come and develop on IA and all.
We think we've achieved that. We think now just within Intel, just not including Rockchip and Spreadtrum and/or what's a 2-in-1 and all that, we think we need to grow at about market from this point, right? There's not a need to go gain share or gain footprint. Think of us as a we'll grow it relative the market. There'll be some pluses and minuses in there.
We're toggling the priority from significantly more share to improving the profitability. That's, we've had in-depth conversations about the levers we want or to improve the profitability while making sure we're not growing below the market rate.
Right.
Stacy, as my follow on, a lot of the presentations this morning talked about your ability to leverage IP across multiple businesses. That, to me, sounds like a really op margin leveraged strategy. I'll go back to Ross's earlier question. You know, if you exclude the mobile wireless losses today, your op margin today is already kind of, you know, mid-30-ish type levels. Why isn't 35%-40% op margin longer term the right way to think about this business given the IP leverage everyone talked about this morning?
You know, I don't know that I wanna talk about long-term op margin. I'll say our goal is to bring down spending as a percent of revenue over time to 30%. If we're, you know, at the midpoint of the long-term gross margin range we've talked about, that kind of puts us at a 30% op margin. We, based on the technology lead that we've had on others, I think that's been the kind of core driver of why we've been at the higher end of that long-term gross margin range. A lot of it depends on can we maintain that lead, and can we execute on the product side. I guess I'd say we certainly have ambitions to be higher. The long-term model that we've articulated is 60% long-term gross margin and 30% operating profit.
30% spending as a % of revenue.
All right. With that, I believe we are out of time for this Q&A session. Thank you all. I'd like to invite the General Manager of our Data Center Business, Diane Bryant, to take the stage.
All right. You gonna help me up? Oh, look at you. What a gentleman. Oh my goodness, chivalry isn't dead. Okay, it's certainly my pleasure to talk to you about the data center business. There's three things that I hope to leave you with. The first is that there are a couple of big industry trends that are fueling this increased deployment of data center infrastructure. The second is that we are investing to win in this rapidly evolving and transforming market. We are investing to ensure that Intel architecture delivers the best results for all workloads across the data center infrastructure, across servers, storage, and network. As has been said a couple times already, we are continuing to forecast a 15% revenue CAGR out through our planning horizon, which is 2018. Okay?
First I wanna start with the big industry trends. The move to the digital service economy is one that we've talked about a bunch. This move to where IT is fulfilling the business, public sector, and consumer needs. That trend certainly continues, and it's this virtuous cycle of you have more and more connected devices, whether they're IoT or personal devices. Those devices drive demand for new services, and those new services drive demand for the build-out of the data center. At the pace of this cycle, the rate at which new services are being deployed, the rate at which new devices are being connected is tremendous and it is accelerating. This drives a pretty significant transformation in the information and communications technology sector as a whole. The second, as my friend Stacy commented on, is Jevons paradox. That's him there.
We were thinking it may be an early relative of Andy maybe. He kind of has some family similarities there. That could be. That could've been your eight-year distant uncle. If you don't know Jevons paradox, this is 1865, William Jevons declared that as technology progresses, as the investment in technology continues and technology becomes more and more efficient, that there will be new usages, and the consumption of those resources will actually expand instead of contract. We see that, right? We see that technology continues to become more and more efficient.
The result of these two big industry trends are four big drivers of our data center business, and that is cloud computing, network function virtualization and software-defined networking, which you can think of as just virtualization and cloud computing for the telco infrastructure, high-performance computing and big data. A quick example for you of the pace of the digital service economy because we do continue to talk about this. Last year, we showed you this slide, and this is November 11, 1111, China's Singles' Day. The digital version of a hallmark holiday where everyone's encouraged to go out and buy something for themselves online, invented by Taobao back in 2008. We showed you last year that in that 24-hour period, there was $5.7 billion of online spend.
Fast-forward 1 year, last week was Singles' Day, the results were even more staggering. $9.3 billion in spend, an increase of 63% year-over-year, 278 million transactions handled through online ordering and payment systems. You step back and just consider the scalability of the data center solution that can deliver that kind of growth in demand in a 24-hour period, it's a perfect proof point for the value proposition behind cloud computing, the ability to deliver on-demand rapid services in support of the, of the changing in the demand environment. Jevons paradox has played out in the data center many times, I have a couple of them here for you. First, you go back to 1996 when Intel entered into the server market.
We entered in and started the movement from proprietary, vertically integrated, risk-based and proprietary Unix-based servers. This big movement then onto horizontal, server solutions running on Intel architecture with Windows and Linux, standards-based environment. That shift from proprietary to open has had a significant decline, an impact on the efficiency of servers. You can see here that over that time period, the cost of servers has declined by 80%, and over that same time, the demand for servers has grown by 625%. A more recent example of Jevons Paradox is with the virtualization of servers. In 2009 is when we introduced Nehalem, the Nehalem product, Xeon product, and it included in it the second generation of virtualization technology, which really drove the adoption of virtualization.
As virtualization occurred, there was a server refresh, so that with, you know, CIOs and enterprise IT taking advantage of both the CapEx and the OpEx benefit of virtualization. That virtualization process drove utilization of the server infrastructure up from, you know, sub 10% in a standalone dedicated server environment to about 50%. The result of that efficiency gain was growth in server demand of a 17% CAGR over that time period. New usages emerge as the technology becomes more and more efficient. Cloud architecture, the move to cloud architecture is this next big transformation in efficiency, and we believe it will continue to unleash new demand. The resulting four data center growth drivers that I talked about, it's important to note that they permeate all market segments.
This is a view we've showed at a bunch of times of how we manage the data center business. We manage it according to the end users that are actually purchasing in data center infrastructure, deploying that into a data center, and managing it. It's enterprise IT, the public cloud service providers, telco service providers, and then technical computing. If you look at the call-outs here, you can see how these four growth drivers are spanning all of those segments. You see enterprise IT is now deploying cloud, private cloud solutions. They're also picking up from the telco market, and they're deploying Network Function Virtualization. They're virtualizing the network in order to take the advantage of those efficiencies.
If you look at the public cloud service providers, they're now deploying technical computing solutions, high-performance computing solutions. There was a great example at Amazon's re:Invent event in Vegas last week, where one of their customers had deployed a high-performance computing workload into Amazon Web Services across 70,000 Xeon cores. They were able to go from a, you know, 0 FLOP to 729 teraFLOP HPC compute environment in just 60 minutes. That's equivalent to the 63rd, number 63 on the TOP500 list. You see public cloud service providers now deploying super computers. The telco side of the house, they're turning into public cloud service providers. You can see China Telecom partnering with an enterprise IT supplier, VMware, to deploy infrastructure as a service.
In HPC, in the high-performance computing space, you see PayPal, a cloud service provider, using high-performance computing clusters to detect anomalies in transactions and doing so real time. There's this blurring of the workloads between the various market segments as these new architectures emerge and these new usages of the infrastructure emerge. When we talk about the growth of the data center business, you folks often ask about the mix between ASP growth and volume growth. I wanna show this here because we have continued to benefit from ASP increases over time. Often we get asked is why is that? Why are ASPs continually going up?
As you can see here, while CPU unit growth has been at 5%, the compute capacity that we have been delivering generation after generation after generation is 39%. The result of that is that we see our customers choosing to buy up the stack, to buy our products at higher and higher price points in order to take advantage of that performance that we're delivering. We obviously continue to provide the benefits of Moore's Law that those organization gives to us, that we deliver a minimum of 20% performance generation over generation at the constant price point, so SKU to SKU, 20% performance improvement at a minimum. Over the past five years, what you see here is 70% of our volume has moved up the stack.
70% of our volume has shifted up to higher price points, beyond that 20% gain that we give gen to gen. Significant shift up, and obviously, that results in ever-increasing ASPs. You could say, "Well, why would a customer see such value in continuing to buy up your stack?" The answer is very easy. It's total cost of operation, total cost of ownership. There's two examples here to demonstrate that. The one on the left is an enterprise IT example. This is an example of enterprise IT making a server purchase decision for a virtualized environment, virtualizing their applications. They could choose to buy 19 servers with a 6-core Haswell CPU, or to get the same compute output, they could buy 10 servers with a 10-core Haswell CPU.
The server price of the high-end 10-core Haswell product is obviously greater, but the total cost of operations, the total cost of ownership is a 44% savings over four years. It's pretty obvious because you have fewer servers. Fewer servers means you've got lower network costs, you have lower power and cooling costs, lower operating system licenses and maintenance costs, applications, virtualization, licenses and maintenance. You just go right down the row, and it simply makes sense to buy higher capacity, a smaller number of higher capacity servers than to buy many low-end servers. The data on the right is from Amazon. James Hamilton is the chief architect at Amazon, for their data centers, and he published their total cost of operations pie chart, as you can see.
In it, he made the case for why it made sense for him to buy a higher-end custom Xeon processor from Intel. Servers make up 57% of their total cost of operation. As he says, if he can get a 14% boost in processor performance, so 14% performance increase out of his server capacity, he can do that with simply a 2%-4% TCO increase. It makes very logical sense for him to buy up, to pay a premium, to have that custom processor solution. With those underlying growth factors, we continue, as I said, to forecast a 15% revenue CAGR over the horizon.
We don't have any real expectation change in the growth by segment, with the caveat that it will increasingly get harder. It is getting harder today, but it will increasingly get harder for us to tie our CPU sales directly into one of those segments, enterprise, telco, HPC, or cloud. It's because of that melding of the workloads that I talked about, the fact that telcos are becoming cloud public cloud service providers, public cloud service providers are deploying HPC clusters, enterprise IT is deploying network function virtualization. These lines, because the workloads are melding, our ability to associate a given CPU sale to one of those segments is going to become increasingly more difficult. The other point that I wanted to make regarding our growth is the diversification of our product line.
Increasingly, as we look forward, our revenue is coming from products other than our CPUs. This includes things like silicon photonics. Our high-performance computing fabric that we've been talking about is coming into production late next year. That's both the fabric controller as well as the switch silicon. We have a new distributed Ethernet switch product that'll be coming out, and then we have next-generation memory solutions as well. All of those then start to drive our revenue growth beyond our standard CPU business, and you can see there getting to 22% of our revenue out in 2018.
We are investing to grow our portion of the silicon components within the storage box, the server box, and the networking boxes. I want to click through and give you a little bit more insight into those four big growth areas and why we're confident about them. The first two areas are driving our business because of the new architectures that they're bringing, architectures that are being deployed, and that's both cloud computing architecture, and then for the telco side, as I noted, NFV and SDN. The other two growth drivers are growing our business because of new usages for the data center technology, and that's high-performance computing and big data. To start with the cloud, as I mentioned, this is the next great example of Jevons paradox.
Cloud computing provides a significant improvement in efficiency, the next big step after virtualization. The value proposition is compelling. If you're an application developer, you're gonna get faster time to market of your new application or your new service through this on-demand, rapid delivery, self-serve capability that cloud computing delivers. If you're the person that's actually running the data center and managing the data center, cloud architecture is gonna give you operational expense reductions through automation, so you have fewer IT folks running around the data center manually configuring your infrastructure. It's also gonna lower capital expenses, so you get greater utilization of the infrastructure, as you can see in the picture, so lower CapEx cost.
From the move from virtualization to cloud computing drives the server utilization from about 50%, the general industry benchmark, to 80% or even higher based on the sophistication of your cloud orchestration solution. That cloud orchestration solution can make very intelligent decisions about where to land the application to get the greatest use out of that hardware infrastructure. And that efficiency, which is a very compelling value proposition, is what's motivating the significant growth in cloud computing. Since 2009, when we actually started to track this segment, very explicitly and develop products targeted at that segment, cloud computing has grown at a 52% CAGR, so significant CAGR. And the cloud deployments are now 35% of our total data center CPU revenue. It's a significant portion.
Whether that's the tier 1 cloud, public cloud service providers, the tier 2s, or now enterprise IT solutions deploying private clouds. If you look at the number of players, the top seven have remained the top seven, have very consistently over the march of time, benefiting tremendously from the economies of scale. Some of these seven, you know, folks have gone public and said that they have, you know, over 1 million servers running in production. Tremendous scale, and that brings them a lot of benefits in that efficiency, the efficient use of their infrastructure. If you look at the tier 2s, the number of tier 2s has grown dramatically over this period. Back in the early days, I mean, we could count them, you know, there's 10s, maybe a dozen or so of them.
There's hundreds of tier 2 cloud service providers. Folks like, you know, Workday and Salesforce.com and Dropbox and Rackspace and on and on and on. There's just many, many, many. We are very pleased to see the growth in enterprise IT. Enterprise IT now moving from the virtualization process that they went through in 2009 and 2010, now moving into a cloud architecture with the deployment of private clouds. We regularly survey enterprise IT CIOs, and in 2013, 7% of them said they were actively building out a private cloud environment. This year, that's up to 12%, we expect that obviously to continue to grow. As this market has grown, we have grown tremendously with it.
It's been a great industry for us to contribute to and participate in. We have included capabilities into our standard Xeon products. We have performance features that make it easier to deploy cloud computing. Beyond performance, there's security is an obvious one, so security features. We've also included in our latest Xeon Haswell platform, features in the processor that allow the cloud automation solution to operate more efficiently, to look down inside of the processor and understand what's the utilization level, what's the power level, what's the security level, and make more intelligent decisions about where to place applications. If you see, there's now 45 cloud, public cloud service providers that are signed on.
It was led actually by Amazon. Now we have 45 that are actually branding their cloud services with the Intel Inside logo. Acknowledging that that service, that infrastructure as a service capability is running on Intel. SoftLayer is a recent example. IBM SoftLayer, which is a cloud hoster for enterprise, they're promoting their cloud instances that include Intel's trusted execution technology. Noting if you land your applications on this instance without TXT, you have a certain level of security. If you land it on the instance that has Intel's trusted execution technology, you have a higher level of security and protection for your enterprise. Of course, they charge a premium for that capability. Amazon is doing a fabulous job for us in exposing the capabilities of our processors.
You can see, when you log on to Amazon Web Services, and ask for an instance, if you look across the 10 different attributes that you choose from, five of the 10 are directly attributes of Intel processors. You'll see, would you like, you know, Intel's encryption instructions, the AES-NI? Or would you like Intel's floating point acceleration instructions with AVX or Intel's turbo mode or what clock frequency? You can see here, the public cloud service providers see direct benefit in exposing the attributes of our technology up to their end users and out to the application providers. They see value in it, so then obviously they're willing to pay for that value, and we see value in it as well.
Over the course of time, we have grown our market segment share with the public cloud service providers from 84% to now 94%, so that is the share consistent with rest of market. Through that same process, we have increased our revenue with the public cloud service provider market by 69%. Significant value that they see in the technology that we deliver. Next is the equivalent of server virtualization and cloud computing, but for the telco market, for the communications infrastructure. It has been coined by the industry NFV/SDN. They could have called it virtualization and cloud and made it a lot easier for all of us, but they didn't. Let me talk about NFV and SDN.
Our overarching strategy in winning the network is in this move from proprietary fixed function appliances, moving off of that proprietary architecture onto Intel. This is the objective for the network, whether it's the network inside the data center, the switching and routing and load balancers and firewall boxes, or the network, the bigger network, the telecommunications infrastructure. The WAN, the wireless and wireline infrastructure, all the way from the backbone out to the access layer. We want that entire network end-to-end to run on Intel. We're very happy to say that we are now able to apply Intel architecture to all of those workloads end-to-end. We've gotten to this point through some significant investments over time, as Stacy and BK were talking about investments that we've explicitly made into the data center business.
First is just the standard Xeon product line that we've been developing. As that product line has continued to grow in performance, as the core count has continued to increase and as the performance per core has increased, we can now reach more and more of those networking loads. We've also made significant investments in accelerators. We have our data packet processing accelerator line, deep packet inspection, encryption accelerators, compression accelerators. We've invested in accelerators targeted at the network. We've also made some very targeted acquisitions, particularly this year, including last week. Earlier in the year, we acquired Mindspeed, and they gave us the ability to play in the small cell section. They gave us signal processing IP that we didn't have.
Just two days ago, we solidified the acquisition of the Axxia product line from Avago. That product line then allows us to have a footprint in the macro base station space. That acquisition actually allowed us to accelerate our roadmap in macro base stations by four to five years, both the revenue and the share. Significant gains there. With those actions, we now can cover the entire network end-to-end. The second step then in the strategy of moving first from proprietary onto Intel architecture is to seize this industry architectural transformation to network function virtualization and software-defined networking. A proof point of how even the largest and most legacy-ridden industries, like telco can move rapidly when the value proposition is compelling is this move to NFV and SDN.
There is tremendous momentum in the industry. You can see the timeline there. The value proposition for the telco infrastructures, for the carriers to move off of proprietary fixed function hardware onto open standards Intel-based infrastructure running in a virtual environment is huge. It gives them both the CapEx and the OpEx savings that enterprise IT found in moving to virtualization, as well as it gives them the ability to rapidly deploy new services real-time, self-serve. The pace of adoption is really quite impressive, especially if you know the telco industry. It started back in October of 2011 in Paris. All good things happen in Paris, I guess. There was a carrier community event, and the British Telecom presented their lab findings that they could run network functions on Intel architecture in a virtualized environment.
One year later, that industry had come together through the ETSI forum and have established nine use cases and 12 POCs that they would collectively work on to build out the capability. 2013, they were deploying proofs of concepts and moving into pilots. Now this year, there will be first commercial deployments as well as many more pilots, but commercial deployments have been announced by China Mobile, AT&T, Telefónica, Verizon, and SK Telecom. Tremendous rapid movement of the industry onto network function virtualization. I'd just like to play a very short video to have the leaders of AT&T and Telefónica network tell you why this move, this transformation of their architecture is so important to them.
Evolution is not enough to stay in business today, so we have to face a lot of challenges. We need to change the way we deploy and operate the networks. Network transition is one of the main ingredients to transform our networks today. Here is where Intel is playing a key role in this new transformation. By using DPDK, we can get the real performance and the enough performance to execute these network functionalities in the network. DPDK was key for that.
AT&T's strategy is really simple. We're moving everything that we can possibly move into software-defined networking. How do customers perceive it? What are the differences? The differences are really around the, what we define the user-defined network cloud is what we call it. When you talk to the customers that have seen and touched this, it's the user-defined portion. Not only are they elated with how simple it is to interface, but first customer actually said, "Wow, I actually feel like I'm programming your network." We have a long history of working with Intel to move through some of those changes that have hit our industry. We have some joint R&D investments. We have people working with one another. When you have two organizations like AT&T and Intel philosophically aligned and tactically deployed, special things happen.
As this movement has been happening, we've been investing to increase the capability of our product line for the network workloads, as I noted. We have in this period, we've from 2008 to today, we've increased our addressable market in that end-to-end network by 11x. In that same period, we have tripled our revenue into the communications infrastructure market. This year, we're really happy we crossed the $1 billion point for our Intel silicon into the telecommunications industry. That now gets us to a 7.5% share of the network market. That's up two points from last year. About one point of that is due to the Avago acquisition that just closed.
We're making gains both in growing the market that we can address, as well as growing our share within the market. Tremendous opportunity remains, clearly. We've also been investing then to build out the ecosystem around network function virtualization. It takes a lot of partners to do this. If you think about the ecosystem that Intel built out many years ago around standard high-volume servers, all of the software and solutions and systems partners. We started the network builder program, and just one year after we launched it, we have over 100 participants, 100 members, and these span, you know, folks from the traditional telco market, well into the enterprise IT space. You have Alcatel-Lucent from the telco industry world, as well as enterprise players like VMware and HP.
The carriers are showing clear results. Telefonica now has 23 virtual network functions across 17 different vendors running in their environment. Dell has a network function virtualization starter kit that is in deployment. Red Hat is investing in a carrier-grade Linux solution. The industry and the ecosystem is building out around the network market running on standard high-volume servers running on Intel architecture. Next is HPC and the growing usages of high-performance computing. HPC growth is across three vectors. The first vector is in the government sector, and that is an area that has the investments have existed for a long time, but we see worldwide those investments growing. One example is Trinity. It's a very large $174 million system that's procured by the National Nuclear Security Administration.
It's a Xeon Phi design win for late 2015. If you happen to know, this week was Supercomputing 2014, the new results of the TOP500 list came out, and we're very happy to say that 97% of the new systems on the TOP500 list are Intel architecture. There were only two that were not Intel. We now have 86% share of the TOP500 systems. Something else we're very proud of is if you look at those systems that have a coprocessor or an accelerator in them, 15% of the TOP500 list have a coprocessor. We now have 33% share of those systems, and this is just two years after we entered into this market with Xeon Phi.
That's, you know, five years after our competitor had already entered and made gains. The second growth area for high-performance computing is in the area of commercial deployments. There is this continued move from physical to digital, from, you know, the physical prototyping and testing to computer-modeled and simulated solutions. The third area is, as I spoke of before, cloud. Is big data and the use of high-performance computing in the cloud. We see new usages. Big data analytics is a big driver for HPC when you talk about the complex analytics required for things like real-time predictive modeling of the PayPal solution or complex analytics in support of genome based personal medicine.
In high-performance computing deployed through a cloud, that model is also occurring, so it is a new usage for HPC, and the % of the total HPC spend that will be in the public cloud is predicted to double over four years. New usages for high-performance computing. We're obviously investing to win in high-performance computing to continue that beat rate. We're meeting or beating our targets for Xeon Phi for this year. We see strong demand for our coprocessor environment, and we're actually making very strong gains on the second generation of Xeon Phi, which is code-named Knights Landing, that arrives late next year. The wins in Xeon Phi are significant for us.
It's a very strong ASP, so the ASP of the Xeon Phi product line is over $1,000, so it's a wonderful contributor to our business. With the second generation, Knights Landing, we're on track for production. The A 0 silicon is out, and I have to say it has as many transistors on Knights Landing as there are people on the planet. 7.2 billion transistors, thanks to Moore's Law, so one transistor for everyone. We're very proud that we already have 50 systems developers committed to that product line. The middle there is beyond the CPU, so we have not made any R&D investments in the True Scale InfiniBand product line that we acquired from QLogic a couple years back.
Yet our design wins of that product have doubled year or 50% growth rather year-over-year of the TruScale fabric. It's a clear indicator of the desire for our integrated Omni-Path fabric solution that comes out late next year with Knights Landing. Customers are recognizing that a good on-ramp onto the integrated Omni-Path solution is adoption of our TruScale InfiniBand product line, so we're seeing nice growth in an existing product line. On the last column there, it's key that the growth in Xeon Phi is fundamentally driven by the applications being able to extract value, performance value out of it above and beyond what you could get with a standard Xeon system. We're investing to continue to build out their worldwide.
We now have 41 Intel parallel computing centers. These are centers where we work with the local universities and research institutes to parallelize and optimize common applications for Xeon Phi. The last is big data, the last of the four, and this is currently the biggest of the big industry buzzwords. It's very big, a big deal, as Stacy says. We will take some credit for the big buzz around big data because the cost of storage and compute have continued to decline over time thanks to Moore's Law and the growth of standard high-volume servers and storage systems. Over the past 10 years, the cost of service has declined 40% and the cost of storage has declined 90%. It's now cost feasible to actually store all this massive amount of data and compute on it.
We believe this is another great example of Jevons Paradox. As the resources have become more and more efficient, new usages like big data analytics have emerged. It's clear that big data analytics is the next big technology disruptor, with an opportunity to transform all industries. The IDC predicts that the market will grow to $41 billion by 2018, which is a 25% CAGR, and it's interesting to note there how much of that growth is in hardware. With big data analytics solutions, there's a large range of open source software solutions. You know, like Hadoop and Hadoop analytics platforms, the investment in building out these big data analytics solutions can go then to the hardware infrastructure build. It's obviously good for us.
There is pent-up demand for big data analytics solutions, we believe, and so we're working hard to unleash that demand. We have the partnership with Cloudera that BK spoke of earlier, where we've worked with them to harden their Hadoop distribution for enterprise. Integrating, for instance, our security instructions into their distribution, so now it's possible for enterprise IT to encrypt data real-time, both in transit as well as at rest. The performance is great enough using our instructions and the power of Xeon that they can do that. You can provide a enterprise trusted secure data warehouse solution. We've seen very strong growth year-over-year in servers that are deploying Hadoop.
We've also done things such as launched a developer platform, Aware, we've launched that out on Amazon Web Services, inviting people to develop analytics solutions on our platform. We've with healthcare industry being the biggest of the big big data opportunities, we've partnered with a couple folks to just demonstrate to the industry the things that can be done through analytics, whether it's research and into new treatments for Parkinson's disease with Michael J. Fox Foundation or partnering with Knight Cancer Institute on integration of the many different types of data with the goal of through cancer research and genome research. It's very clear that big data analytics solutions value the full range of our data center portfolio.
To demonstrate the growth of IoT devices and how, with the build-out of all these IoT-connected devices, we're driving the growth in new usages and new applications for Big Data Analytics solutions, I'd like to invite to the stage Ron Kasabian. Ron is the General Manager and Vice President of our Big Data Analytics business. Ron, I have been wearing this Basis watch for a very long time, and at last you're gonna tell me why.
Yes. Well.
Come on over here.
While you have been wearing your beautiful Basis Peak watch, we've been streaming data from your watch, your activity data, into the cloud and loading it into a distributed database like Hadoop.
Okay.
Running on Amazon Web Services. We've been collecting that data, and that data can be used for a whole bunch of different things. Today we're going to look at how we can use it to increase your quality of sleep.
I see. Okay.
So, um-
Are we gonna project this up here?
To do that, yeah.
There we go. Oh, I'm sorry.
Yeah. To do that, what we've done is we've actually taken a cloud-based machine learning tool called Neuron.
Okay.
Developed by some partners of ours called ColdLight, and we've loaded, five years' worth of data from 10,000 individuals on activity and sleep data into Neuron.
Okay.
We've got that huge base of data there. Now we're gonna compare your data against that, and we're gonna see what kind of patterns and predictions Neuron can develop for you to determine whether you can sleep better and maybe how we can help you sleep better.
Well, I'm very-
Right?
I'm glad you're so concerned about my sleep.
This is gonna happen. I do care about you.
Thank you.
This is gonna happen in real time.
Okay, real time.
Less than a second, so don't blink.
Okay, don't blink.
Okay. What we're gonna do is we're gonna take data from the last couple of days. We haven't refreshed it lately.
Okay.
We're going to take data from the last couple of days from you and your watch.
Okay.
We're gonna compare it against this huge database, and we're gonna develop some insights, which we'll look at real quick.
Okay.
I'm gonna click Refresh, and now we can see up in the top right that 24%. That's pretty good. People are recommended to at least get 20% REM sleep per night.
Oh.
24% based on your activity over the last couple of days, so that's good.
I'll sleep better after I ingest your data.
You after this demo and based on your activity. The other thing is down on the, in the right, bottom right half, there are a couple of predictions. There are some short-term predictions, again, generated on the fly.
that basically say if you go to bed an hour earlier, you're gonna increase your quality of sleep by another 1%. Long term, there are some predictions like if you increase your vigorous activity by two to three days a week, then, you could potentially increase your quality of sleep by 5%.
Okay.
Okay?
So-
That is lovely. It's probably not the most helpful piece of data I will get today, how to get a better night's sleep, so I'm sure that this is all about one example of a broader range of data analytics solutions that are of great value to the world.
It is. In fact, you know, lots and lots of data analytics and machine learning are being used today. It's being used in healthcare. It's being used in manufacturing and retail, lots of different industries, as you pointed out earlier. You know, the ability to analyze massive amounts of data is really bounded by the amount of compute, and network, and storage we have available to us. For this demo here, we loaded the data into Neuron, and we're retraining the machine learning algorithms on a regular basis. To do that and to generate your predictions, it took 94 Xeon cores.
Wow, right.
That's based off a base of 10,000 users. If we were ideally to increase that to 1 million or 10 million, the number of Xeon cores required would increase by a factor of over 100.
Right, right.
So great-
Great use of-
Yeah.
Our technology across server, storage, and network.
Yeah. These applications require not just, you know, high-end Xeon processors. There's opportunities potentially for Xeon Phi.
Mm-hmm, fabric.
A high-speed fabric and more memory and more storage. You know, the future of analytics at the end of the day is really based off the ability to apply increasingly complicated algorithms to increasingly large amounts of data and find that insights that no one person could find by themselves, using compute to find a needle in a haystack.
Unleashing new demand.
Yep.
Thank you, Ron. I appreciate it. Yep.
You're welcome.
Mike, Val, I apologize, I like the MICA a lot better than the Basis, I'm gonna switch over. Sorry. No offense to the Basis watch. No offense. Okay. Stacy's giving me the hush. Okay. I do wanna hit on three other areas of investment. These are areas that we've been making significant investments in and will have an increasing impact on our business going forward. Let me click through those here. First of all, as the world's dependency upon IT grows, the number and the types of workloads continues to grow. As I said, we are committed to making sure that Intel architecture, that single architecture, covers the entire workload space.
One architecture across server, storage, and network provides tremendous efficiencies for the developers and for those folks that are actually running and managing a data center. We have over 100 standard Xeon and Atom SoC SKUs at any one point in time in production across compute, storage, and network. But we are equally adept at providing customized solutions with the Intel architecture, and we have a broad range of implementation options that you see across the way here, all the way out to full custom CPUs. Last year at this event, I noted that we had delivered 15 customized Xeon processors. Today, that number is 35, and just in the last four months, we've had some real fun announcements come out. One of them, Oracle, an elastic CPU was announced in support of their Elastic Cloud.
At the Amazon event last week, they announced that their new C4 compute optimized instance was running on a custom Xeon E5 CPU. Microsoft just launched a couple weeks ago Azure G-series, which is their an instance of Azure Platform as a Service that's targeted at big data analytics, and that is running on a custom Intel processor, Xeon processors, that we did for them to meet those workload needs. HP just launched their next generation of Moonshot running on the Haswell processors, and one of the cartridges that we did was a custom cartridge with the integrated graphics, Iris Graphics solution, targeted at media and remote graphics workloads.
Overall, if you just looked at the cloud service provider market, 23% of our CPU volume is now custom processors, and we project that that will grow to almost half of our processors for the public cloud service market next year. It's a tremendous capability that we have. Another new and exciting product line for us is silicon photonics. I have to get my props here. We started shipping our first silicon photonics products just last month. There are three targeted markets for silicon photonics, I wanna hit on them briefly. First is in the data center, this is from the top of the rack switch on up to the core network, replacing existing traditional fiber optics with silicon photonics.
The ability to do that, we have a very disruptive cost structure with silicon photonics. We're the only solution that has a fully integrated onto a single piece of silicon on this module here. That allows us to deliver a solution at a cost structure that is half that of other silicon photonic solutions and a third of traditional fiber optics, and that's with today's 100 Gb per second bandwidth. As that continues to grow, the distance and the ability we have to distance the traditional and other silicon photonic solutions continues to grow. It becomes a very strong competitive advantage. The second market is high performance computing, and the high performance computing market has a, is I/O. This is awfully big.
The high performance computing market cabling and the node-to-node interconnect with copper is a constraint to the performance and it drives up power, which then again constrains performance. If you look at the I/O spends in high performance computing, 30% of the total HPC spend is going into I/O. The promise of silicon photonics with high integration, low power, small form factor, lightweight cables is a very compelling value proposition to high performance computing. The last is Rack Scale Architecture, which this is an example of the Rack Scale Architecture here. Rack Scale Architecture is all about driving compute density into the rack. As you drive greater compute density, you need to drive greater bandwidth down into that into those compute nodes.
Copper, the copper cables that I showed you there, is limited at 3 m at 100 Gb. If you're gonna go beyond 25 Gb by 4 Gb, you need silicon photonics to do that. Very compelling value proposition across the three different markets. As part of our investment, we have created an industry standard around our silicon photonic solutions, around the cable and the connector. Prior to our work here, each one of the silicon photonics or traditional fiber optic solutions had a proprietary cable and a proprietary connector. That's an investment, that's a cost, that is kind of no value add. By standardizing the industry, we're driving down the cost of the cabling and connector solution, and we have multiple players there.
Then you can see at the bottom, we have multiple, many, many, many design wins. Some of them we were able to put on the slide and show you today. We are shipping first production now, but we see tremendous ramp and opportunity. The market is about a, we project in a couple of years, it'll be about a $1 billion-dollar market, and we think we have a great opportunity to win there. Lots of work, and I should also add, just like we've have the processor, we've integrated the fabric into the processor. Over time, we would also have the ability then to integrate the silicon photonics with the fabric into the processor, so it's a great opportunity. Another use of Moore's Law. Last, as I pointed to, is the Rack Scale Architecture.
This will be available, this architecture will be available from multiple vendors in 2015. It's an architecture that's purpose-built for cloud computing, so it's kind of the next generation of the rack moving from, you know, 2U rack mount servers, moving to a fully integrated cloud-based solution. What this architecture is all about really, it's about the ability to deliver on-demand applications down to the resources that get the best utilization of those resources through the pooling of compute, storage, and network. For instance, here you have a tray of Xeons, here you have a tray of Atoms, you have the pooled storage. The opportunity for us is that it can accelerate the CPU refresh, 'cause you can refresh those processors without having to refresh the entire server.
It also is CapEx savings 'cause you have fewer, as Dell calls it, UCPR, unnecessary crap per rack. You have fewer fans, fewer power supplies, fewer of all that stuff that's really no value add. If you're spending less on that stuff, you can spend more on Intel silicon and valued products. And then it also gives us the opportunity to grow our share of the footprint, as is noted there with the variety of products. I do wanna spend just a minute to talk about the competition. With the first shipment of ARM 64-bit processors, there's a new round of discussions that are occurring. Obviously we take competition very, very seriously, and there will always be competition, especially when you have a business opportunity that's as strong as ours.
There is now 12 ARM vendors that have stated that they are entering into the data center business. That's down from 16 last year. IBM is also, over the year, they've opened up their architectures. They have openpower.org. Obviously they're trying to increase their footprint of Power as that market continues to shrink so that they can sustain the support and services revenue. AMD obviously continues on. Our strategy is pretty clear. We are committed to delivering leadership solutions across the data center workloads. As I noted, we have a full range of implementation options at our disposal to meet those needs. And we have an annual product R&D investment, just for the Data Center Group at, of about $2 billion. It's a significant investment that we make in order to sustain that leadership.
We have, obviously, a very expansive ecosystem running on Intel architecture. In doing so, we can continue to deliver the absolute best performance per TCO at whatever the workload is. As you can see there, James Hamilton, again from Amazon, when he was at re:Invent last week, he was interviewed and asked for his thoughts on ARM. As he says there, Intel continues to be the right option. When you do the numbers and you're trying to maximize the performance you can get at the lowest cost of operation, Intel comes out ahead. To conclude, our industry is in a period of tremendous change, which creates tremendous opportunities for Intel.
We are absolutely investing to win and to expand our position as the industry moves to these new data center architectures, to net-new network architectures, and in the adoption of new usages. As we stated, we will continue to grow the business at a 15% CAGR out through the horizon. I encourage you all to come visit our demos. We have many of our General Managers here, and hopefully you can drop by. With that, I wanna say thank you and introduce to you Kirk Skaugen, Senior Vice President and GM of the PC Client Group.
Okay. Well, I realize it's been a long morning, but the way I look at it's about $1 billion of revenue a minute I'll be up here to explain, so hopefully we'll keep you interested for the next 35 minutes or so. Although I'm excited about the Client Computing Group and the org announcement, today I wanna talk about why I'm still invigorated about PC clients in 2014 and 2015. There's really three things I want you to walk away with today. One is, at the end of the day, PC meant personal computer, and maybe for a few years, we lost that personality.
I think there's innovation that's coming back and the strategies we've laid out for the last three or four years, and that innovation is driving some of the growth that I'm confident we'll have in the future. The second question I get from a lot of you, and I was just in New York a few days ago, is, "Hey, does performance really matter? You're going across new operating systems like Chrome. You know, does high-end gaming really matter?" I'd like to just give you a perspective. I think performance is becoming more important than ever, especially as we go forward, and what that means to our average selling prices.
Third is we are investing, as Brian Krzanich said, in a lot of new user experiences that we think can help significantly refresh our existing installed base and get people excited to want a new PC, not just need a new PC when something breaks. Let me start off and talk about the innovation strategies. As was said by Stacy Smith and BK, last year we were here talking about revenue down mid-single digits and operating profit flat. As you heard, we've been delighted with the year, with operating margin now up, you know, over 20%, revenue up single digits. It's been fantastic, not just driven by XP refresh, and hopefully you'll see some of those as I get through today, but really by reinvigorated companies in the PC market that are innovating, I think, faster than ever.
I'm also excited that if you look at the other businesses around PCCG, hopefully you saw that with Diane Bryant on silicon photonics and some of the other things that were talked about today, our non-processor chipset businesses, so things like our gateways, home gateways. We're now number one in the world in cable gateways, with our Puma silicon, more than 100 million units shipped into customers like Comcast and Charter Communications. Thunderbolt silicon with more ports on the next generation Macs. Things like our Wi-Fi silicon, the leadership we've taken on 802.11ac. This is the first year we'll cross $1 billion in that business as well. Not just the processors, but also increasing our share of wallet in some of the peripheral businesses as well.
Okay, Stacy showed you this chart. Yes, I can confidently say we'll be in this range of the TAM. This is really part of the reason we created this Client Computing Group next year, is that, you know, as I went to New York last week, I spent a third of my time just explaining what chip was in which business unit, and if I buy a vHRL-T or a vHRL-M, our customers were saying the same thing. You know, IDC doesn't include 2-in-1 detachables in their PC number when you see this forecast. You gotta dive under the covers and say, what is included? Do they include a Surface Pro 3? Is a tablet or a PC?
Although Stacy talked about flat units, you know, I'm here as the business unit GM saying I think there's innovation and excited customers that we can do better than that, not just in 2015, but beyond. That's probably a good bet that Stacy and I will take this year, even though I think it's the right thing to be financially conservative. This is four now consecutive quarters of year-on-year PC growth, we're excited about that. In the third quarter, our notebook business was up 21%. Desktop, which a lot of people had said was dead, was up 6%, and our top-line revenue was up 9% year-on-year. If you look at that growth curve, why did it occur?
I mean, certainly XP refresh helped, we still think there's about 170 million PCs out there that are still running XP, even after the installed base this year moved over. We're also seeing tablet saturation in the mature markets, as tablets move down to phablets and 7 in and 8 in, the productivity that you can do on a 7-inch tablet is very different than you can get on a PC. A lot of people now have got a tablet in their house, one or two, and now they're back looking at the innovation you'll see today and saying it's time to buy a notebook. If you look at the installed base, I think 1 thing that's relatively shocking to people is the installed base of PCs now is 1.8 billion units.
When the iPad came out, people were elongating their purchase. You look at now the number of PCs that are four years and older, and it's more than 600 million units out there. What we're finding, and we interview about 2,500 people a month, we do about 250,000 interviews a year, believe it or not, is they're now waterfalling their PCs down. For example, putting them into the kids' bedroom to do their homework or to have games and things like that. While the refresh rates are getting longer, the total available installed base is getting bigger.
When that breaks, the kids are coming back to their parents and saying, "Hey, I've had a PC in my room the last 1.5 years, I still want a PC." It's actually causing a refresh, so the overall TAM actually and refresh rates get bigger from that perspective. Even a small change in this refresh is a big deal. With Microsoft now coming out with a new operating system next year, operating system refreshes are always good for hardware refreshes, and we're excited about that. I contend, as I've talked to you, that this year there's been more innovation in the PC market than in the last several combined.
If you just look at the traditional desktop space, we still have an all-in-one where you put the processor behind the screen of about a 17% compound annual growth rate going forward. We're seeing the increase in density, where you're taking desktops out from under the office and putting them into the kitchen as a TV, with gesture or with touch, just to check the stock price or do things you would do, and that's increasing the density of desktops per home. In the hotels, for example, if you go over here to the Marriott next door, you'll see that they replaced all their desktop towers with all-in-ones behind the counter. All the way places you get your ticketing for airlines or check-in or go to the concierge are now all-in-one.
We're seeing business increases in the number of PCs purchased per business as well. We also have a new category called mini desktop, which is one of the fastest growing categories at 20% annual growth rate. This happens to be a Lenovo Tiny, HP has a mini, Dell's launched one, this is growing at over 20% compound annual growth rate. This is a full Core i7 vPro, another benefit of the Moore's Law that Bill talked about. As this technology continues to evolve, what we'll be announcing this year are basically compute sticks like this that can plug into the back of anything from a TV or a smart monitor and bring intelligence into that.
That market is tens and tens of millions of units that traditionally had run on ARM and Android that now we can put into a fan-less Intel envelope. Last thing I wanted to show you is if you can just hand me over the portable all-in-one. This is probably one of the fastest growing categories now. Best Buy is putting a store-in-store in about 400 of their stores. This is not your father's Oldsmobile desktop, right? This is a Lenovo Horizon 2s. It's 5 lbs , four hours of battery life, 1080p display. You can lay it flat. It has touch screen in it. Basically now is movable around the house. This is the kind of thing you're seeing now go into places like your kitchen and this kind of area. Lots of innovation on the desktop.
Likewise, on the notebook side, just colors. For five, six years, you had a choice of black or gray in notebooks. Now you can get red and blue and yellow and green and just about any color you want. I'll show you more of the innovation mechanically in a second. The other thing that's happened, as BK said, is we're now embracing all operating systems, right? For decades, it was Wintel, then it was Microsoft and Mac, as we continue to see Mac growth. In the third quarter, you saw Mac sales pass iPad for the first time in a long time, just based on the outstanding battery life you can get on a Mac Air. Look at the other logos on here. I'll talk about Chrome at length. You're now gonna see Android portable all-in-one.
If you like just a big tablet. Is this a tablet or is this a desktop or is this a portable all-in-one? Again, another reason why all the areas are blurring. One that's might be interesting to you is how many people know of SteamOS or have heard of SteamOS? Yeah. That's the logo there, the third one in. SteamOS now has over 100 million users. They have an entire ecosystem where people go and get games and bring it into the console, and we're very excited about what's happening there as well. Port of choice, as BK said.
What I'm happy about is, you know, sometimes it takes a while to get these strategies in place, but we've had a very consistent strategy on how to reinvent the notebook and get growth back into this business. We brought core microarchitecture in 2010. We drove basically the entire world to thin and, you know, everything is thin now. We've basically cut the weight and the thickness of notebooks in half since 2011. Touch has been the fastest ramping technology that's gone into a notebook, faster than 2D cameras, faster than even Wi-Fi came into a notebook, and that's despite the, you know, shaky start that Windows 8 had.
In 2013 and 2014, we've added Chrome support, and now we've moved to this 2-in-1 best of a tablet and best of a PC in a single device. How is that 2-in-1 going? Again, 2-in-1 was really our concept of why do you have to carry around both a tablet and a notebook? If you can get all-day battery life, and we can make these things incredibly thin, why can't you get the best of a tablet and a notebook in a single device? There's really two focus areas for 2-in-1. The first is, let's just get people excited to go back into the store, not because something broke, but because they want a new PC, not just because they need a new PC. What we're finding here is with year-on-year retail sales now of 2-in-1s up 150%.
We do exit surveys now over 1,500 people in U.S., China, India, Brazil, Japan, and England. What it's saying is that people that come in to buy a 2-in-1 are refreshing their notebooks one whole year earlier than people that buy a clamshell. 1 whole year earlier. When you ask them very specific questions, "Did you want to buy this, or did you need to buy this?" Increasingly, people are saying, "I wanted to buy this." That's kind of for the 13, the 14, and the 15-inch screen sizes. The second thing we wanted to do is basically drive share back from premium tablets. There were, at one point, 100 million tablets, you know, selling even before the keyboard and the cover at pretty significant price points.
In this 10-in, 11.6 in, or 12.5-in category, it was really about giving people a reason to buy a PC, a tablet first model where you detach the screen and it has every bit as good of a tablet experience. In that space now, in all 9 in to 12 in, including all tablets, all iPads, et cetera, two-in-ones now represent more than 10% of the market. In the case, again, of people walking out the door with two-in-ones, 56% of them said they probably would've bought a tablet had they not walked out of the store with a two-in-one, which is a big number.
Which means we do believe we've arrested some of that erosion that was going to 10-inch tablets, and that's happening in the 10 to 12.5-inch screen size. More than 70 designs last year, basically nothing under $699. Now more than 50 designs at system price points below $699, or about 40% of the designs below $699. The other thing we did this in the third quarter is we announced our first new brand for PCs in five years, that's Core M. We did that because we felt we had a very unique value proposition to go after premium tablets and value 2-in-1s. We literally have a product now that doubles the performance of any ARM or Atom product in the marketplace.
You say, why do we have confidence that we can grow and people are excited to come in versus a 4-year-old laptop, again, 600 million in the market, twice the performance, 2-in-1 versus a laptop-only, twice the battery life and half the weight. This is literally what You know, you kinda forget what a 4-year-old notebook looks like, and I'm gonna do my best to keep it holding out here. You look at what Lenovo is now putting into Best Buy as of yesterday, which is the new Yoga 3 Pro, and you just get a perspective of how different it is. I can't hold that much longer. Got it. If you just look at it, how amazing this is. This is a full Core performance, and will sell up to full Core i7.
Obviously, even though this is a PC first, tablet second, it's still incredibly thin, please go to Best Buy and get one after today. That's also with 10 hours of battery life. As BK said, we don't have an issue now on battery life anymore. We're still getting that same kind of performance and battery life. Okay, the next area. Drilling down into the Bay Trail space or Bay Trail-M and Celeron and Pentium. This is an area where we hadn't invested significantly in the past. As a result, our share in the sub $399 PC space versus AMD had fallen below 50%, and our Chrome share had fallen all the way to 24%. With Bay Trail-M and Celeron and Pentium, we've made a significant increase now.
We're now the largest provider of Chrome products in the market, going from a low of 24% to over 65%. I don't think we're done here. The second reason I have confidence is that Chrome units are expected to triple by 2017 from Gartner. Our brand at Intel, because of our teach and our teacher initiatives, is very strong. As a result, if you look at someone like Samsung, last year they had their own Exynos ARM processor in their Chromebooks. This is now the Samsung Chromebook 2. 9 hours of battery life, under an inch thick, less than 2.5 lbs at $249. Even someone who had their own internal Exynos processor is moving over to the Intel solutions.
As Stacy said, Stacy showed you a slightly different graph on Bay Trail. If you remember, we had Sundar from Google up on stage launching Haswell-based Chromebooks. Those were on our big core. Now we're taking Bay Trail and we're able to get to much lower system price points, which expands the TAM. For notebooks, now what you can see is Q3 going into Q4, 83% of our Celeron and Pentium mix is now in small core. We're getting product margins at or better than our big core, but we're able to get down to price points as low as $199, and Black Friday you'll see things even lower than that. What Stacy showed you, just so you can reconcile, is all both notebook and desktop percent Bay Trail.
This just gives you more data into what we're seeing on the notebook side. On business. It seems like every week you read about a major retailer in the world getting hacked or a government being hacked by cyber terrorism. vPro today is viewed, I think, undisputedly as the world's most secure and managed desktop. You can manage it out of band. We've been working on simplifying the deployment for several years, we've now seen a 13% increase in vPro, which is only Core i5 and i7, because security is becoming the top concern in the IT space. We're also seeing vPro get into devices in Doug's IoT world because they're putting them into point-of-sale terminals, where point-of-sale terminals are getting hacked, or vending machines or kiosks as well.
The big transformation, I think, in the last year has been you used to get a super thick notebook from your IT department. You could go down to Best Buy and get something incredibly thin. If you look at what's now available from someone like Lenovo here, this is 9.7 millimeters, 820 grams, 12 hours of battery life and full LTE. A fantastic notebook, and Lenovo's obviously known for their great business keyboards. With just like that, you've got yourself a tablet with, you know, outstanding battery life and Core level performance with all of vPro security and manageability features. What we're gonna do this year is we're expanding that to benefit the line of business managers. We're gonna put vPro wireless display in.
This is the basically, every meeting that you go to, you try to find the dongle to connect to the dongle of the projector. Basically, by the first half of next year, when we launch the next generation Core, you'll just be able to flick your screen onto your projector, and it'll be secure and managed just like an endpoint so that people can't hack into your network, but you don't have to worry about dongles ever again. We're committed to go into another new business called WiGig, which basically, I'll show you a demo in a second, enables you to connect to your projector or to your screen and all your IO wirelessly, anywhere between 3x and 10x faster than Wi-Fi.
This will grow to a multi-hundred-million-dollar business for us because we're providing the silicon not just on the notebook side but also on the display side because it's an end-to-end connection and experience. Again, Intel's uniquely served because people trust the security that we're putting into these kinds of things. Okay. I just wanted to quickly summarize then sort of the multi-segments of notebook and desktop, and then we'll move on to performance. The way to think about this, and you'll see demos out there, is in Entry, we're fully supporting Windows and Chrome. You'll see that down to $199 price points, maybe even below. The 14 NM versions are on track, called Braswell. I can confirm everything's on track to get 14 NM into the Celeron and Pentium space.
In Mainstream, we're gonna continue the move to thin and light. We plan to ramp the fastest Core generation we ever have. The benefit of having 14 NM a little later than we had desired is that all the motherboards are there. We plan to take the next generation Core family all the way from Core i7 all the way down to Celeron when we launch it. It's gonna be a very fast transition in the mainstream. In Premium, we're creating new enthusiast SKUs. I'll talk about that. We're ramping our Iris and Iris Pro brands. I'll talk about that as well in terms of raising the average system price with our graphics. In 2-in-1, we're right in the middle of the Core ramp, Core M ramp.
You'll see about seven to 10 products hit the shelf here over the next few weeks as we ramp Core M detachables. In the business space, you're gonna see products like Helix, where some of the thinnest systems in the world are now business PCs, not just consumer PCs. Increase our vPro security and management, create a whole set of small business SKUs that have kind of lightweight security and manageability that is more affordable for small business. In the desktop space, in Entry, we're seeing a huge engagement now from the China tech ecosystem. As far as I know, basically 100% of the desktops and notebooks out of China tech ecosystem are Intel-based. We learned from the aggressive growth of tablets there.
We embraced it, and we have a very strong relationship with the CTE ecosystem there. We're gonna create a new set of enthusiast SKUs for the high-end in towers. I'll show you a few of those in a second. In the mini space, this is again, one of the fastest-growing segments. You're gonna see a new Steam Box and the SteamOS launch on Intel architecture. In all-in-ones, we're driving price points down to $449 and even below. Things that used to be $999 are now $449. New relationships with the display vendors to cut 4K displays in half in the all-in-one space. In business, really a ton of form factor innovation, again, vPro for IT. Again, the purpose here is this isn't really just a one-stop, one form factor world.
When I managed data center, a lot of our time was explaining why even though enterprise towers were going down, HPC and cloud were still going up. You've got all-in-ones and minis here growing at 17%, 20% a year, even though the traditional tower business obviously is going down. Let me transition to performance and why I think the performance requirements are gonna be more important than ever as we go into 2015. There's four areas I wanna talk about. One is does performance matter in Chrome? Is Chrome just a value operating system? The second is, what's the value proposition of this new brand with Core M? The third is, what happens when we put 3D cameras in everything from these very thin tablets all the way up to notebooks and 2-in-1.
Last is, what's going on with enthusiast gamers? What we have here is just a general performance statement of our two processors, kind of entry processors. On the left is a Core i3 and on the right is a Core M versus ARM. You can see on WebXPRT, which I think is a pretty understood third-party benchmark, we're anywhere between 2.4x and 3x the performance versus ARM. If you really look at an Acer product, this is an Acer, You know, one of these is the Acer Core i3 product, and the other is the ARM product.
At less than $300 on the Core i3 product, you can basically open a Hangout session, pull up about nine websites, and you can still smoothly go browse through Chrome. On the competitive systems, once you load Hangouts, and you put a couple websites on, you're gonna get incredibly jerky motion, and we'll have that out in the demo area, so you can just literally play it for yourself. We've been working a lot with Google on optimizing Chrome. What traditionally was just Celeron, now we have multiple OEMs now launching Core i3 and even Core i5-based systems this year based on Chrome.
That'll help us scale the average selling prices on Chrome. On the right, this is just showing a Windows, a Windows version of that same thing with Core M, not even Core i3 being three times the performance relative to Qualcomm. We feel very good about how we're situated relative to Arm and the Core processor. The second area is RealSense. BK showed you a little bit of why he was excited about RealSense. As we get into the end of the year, we're gonna have this integrated. I'll show you the number of systems in a second. What I thought I'd do is just show you what happens to CPU utilization when you move to a world of 3D. A lot of people don't realize we've acquired a number of 3D camera companies.
Craig, why don't we show them? This is a system that'll be available in early 2015 from ASUS, in one of their notebooks. Why don't you show them what they have?
Yeah. Here with an integrated camera, and let me just show you just one basic experience of what we're looking at, combining the innovation that we're moving to the PC and our compute devices with RealSense 3D cameras, but also pairing that with the performance of our Core line. Let's go ahead and take a look at this experience. What we're gonna do, I figured we'd do something fun for the morning because we can do a bunch of gesture, we can do collaboration, but let's go ahead and we have a digital avatar application. Let's take a look at what that looks like here. As you can see, I'm just actually taking a window look at this. As you can see, the application itself has a baseline performance.
We're looking at about 30%-40% for me to just, you know, be ready to go ahead and make this thing work. What I'm gonna do is load on some of the calibration here, and that's gonna go ahead and start our scanning. As you can see now, it looks like it's actually doing all of the hundreds of points across my face, as well as to give us this experience right here. As you can see, all of the individual points on my face are being tracked at the same time, in addition to muscle structure. If I push out my cheeks, get something like that. You know, a lot of head movement, something like this. This is, you know, an experience that we're actually able to do in real-time for the first time.
If I wanna take this ugly demo guy mug and turn it into something that's a little bit more fleshed out than a wireframe, let's try something maybe a little bit cuter. Aw. That's a you know, something like that, right? What Okay. Well, I wanna go ahead and do an improvement on Okay. Let's see if I can make this demo mug look any prettier instead of cuter. Oh, well, okay. I'm gonna say kind of a definite improvement.
It pump up the stock price.
Yeah, I'll say I think we're up a little bit. You know, and here's the deal. Now, with this performance and innovation, for the first time, you know, that's a lot of numbers we said, thousands of points across your face and all the performance that we need. What we can do for the first time is actually convey real emotion on the digital avatar. You know, this guy is kinda like a sexy hey. Right? Or even to the point of excitement, like pump you up. You know, okay, that's all silly. You know, let's go ahead and bring back to the original CPU monitor. Now we've actually kicked it up to over almost twice with including some spikes there.
That's the real gist of this demo, everyone, is that we're looking at the combining of the innovation that we're getting in RealSense, and at the same time, all the performance that's required for a great experience like this.
All right. Thank you. Good stuff? All right. Next, I wanna talk about enthusiasts because this was honestly an area that this is our most loyal customer base. A lot of people don't realize it, but software revenue from PCs passed game consoles in 2012, and it's never gonna go back. It's $23 billion of revenue into PC gaming in 2013, going to $33 billion in PC gaming by 2018. These people, in some cases, refresh every one year, every two year, every three years, and we just weren't building products for them. This year, we actually went and built very specific products. We built the world's first 4 Core, 4 gigahertz on every Core, overclocking well over 6 gigahertz for these teams. This is like the most amazing audience.
You get in front of them with 3,000, 4,000, they're screaming about clock speeds and USB ports and PCI Express buses. We also went out and went one step further, and at PAX, to about 4,000 gamers, we launched our first 8-core, 16-thread desktop machine. I thought I'd just show you what the heck this thing does, as you connect it. This is the 8-core, 16-thread Haswell-E. What we're showing, you won't be able to see it necessarily on the screen, I don't think, but this is, we don't have a 4K projector, but this is full 4K, and you can run four discrete cards, multiple 4K monitors, and get just an absolutely outstanding gaming experience with this kind of product. Come out and check it out in the demo showcase.
Again, the world's first 8-core. This is why when you see our Core i7 business is growing so high because these people are just streaming in to win in this space. Okay. You got enough.
I could do this all day.
Pretty good. You wouldn't believe this picture. China just built its world's first PC gaming stadium with 20,000 seats. This is 20,000 people in Poland attending. If you're here in San Jose in a couple weeks, we're sponsoring Extreme Masters. We're expecting tens of thousands of people at the San Jose Convention Center to watch their best gamers compete for the basically gold medal of gaming. This is a big deal. We're focused on it, and it's a real reason why Core i7 is pumping up. Okay, it's not just about discrete. We have invested a huge amount of our die size, and one of the biggest benefits to Moore's Law is in graphics.
If you look at since 2006 when we launched 5th-generation Core in early next year, we're going to have a 100x increase in the graphics performance. Over that time, our share in graphics has gone from 40% to over 70%. That sounds good, there's still more. The reason I'm excited is we're getting an average selling price increase here of tens of dollars, and we can actually go after discrete. About 80% of the discrete cards in the market, we already technically outperform. Now this is just about building a brand around our new Iris and Iris Pro. You've seen Apple now bring this into a large percentage of their Macs, and it's not just, you know, these huge gaming machines. We can now put some of the almost discrete-level graphics into something as simple as this.
This is an MSI machine, less than 20 millimeters of Z height, 25K, 2560 resolution, only 1.3 kg , but discrete-level gaming performance. We're seeing more and more people go to gaming from a mobile perspective. You don't need to have these huge machines. This is the new OMEN system, we're seeing people like HP now enter the market. They had bought Voodoo. They had ignored it. They're now bringing a brand-new brand, their OMEN system, to market. This thing dances with LEDs. You gotta play around with it, purpose-built, high-end gaming machines for notebooks. If you can just look at the desktop, you don't need four huge discrete cards with these huge fans.
This is Iris Pro in a form factor that can support 80% of discrete performance today in a form factor like this. Lots and lots of enthusiasm here around the extreme gaming market that we will continue to work on. Quick roadmap update. Simply put, everything's on track that we've told you. Core M is ramping. 10 systems hitting the market as we transition over to holiday. Our first new PC brand in five years, primarily for detachables and new premium tablets. Second is we're launching 5th-generation Core. That's the Broadwell products. I'm expecting one of, if not the fastest ramp of a new process technology in our history. Core i7 down to Celeron launching in early spring.
Third is the next generation Braswell, Celeron and Pentium, taking over from the 180 or so designs we have on Bay Trail. That's on track for second half. Skylake, or 6th-generation Intel Core, is on track for the second half of the year. I feel really good about the transitions here, and especially the Core i5, 5th-generation Core ramp, because of the healthiness of that Bill showed of 14 NM. Lastly, new user experiences. What do we mean by this, and what are we doing? The reason we're focusing on user experience is we think it can help refresh those 600 million PCs that are out there. We think if we put things like 3D cameras and wireless charging into our tablets, they'll be differentiated, and they'll help gain market segment share for Intel tablets.
We're seeing some average selling price uplift. For example, if we put pro wireless display that's secure and managed, wireless display into conference rooms around the world, and that's secure. We can drive that with vPro, which is a sell up in business. Obviously, being an innovator, it helps grow the Intel brand as well. There's three areas that we've focused on. One is eliminating all wires from computing. The second is eliminating all passwords. The average person having about 18 passwords has to change them every 90 days. Our goal is starting by the end of this year, you'll be able to download a McAfee application that is you are your password, where your biometrics basically eliminate the need for you to enter passwords for Windows login and eventually all your websites ever again.
The third is move the world to 3D with these 3D RealSense cameras. When I say deliver a no-wires experience, you know, this is, as we do market research, this is one of the most frustrating things, is you got your power brick, you got your USB ports, you got your HDMI cables, your docking stations. What we're talking about is eliminating the wires for display, docking, charging, and data transfer between PCs. What I thought I'd do is just show you, again, something that's coming out in a few months, but I'll just show you what we mean by WiGig silicon, of how you dock in a future corporate environment, then how we do wireless charging. This is a reference design.
By the end of this year, for Skylake, we'll have full reference designs where you literally would not have to put a port on the side of a PC. You wouldn't have to have any USB ports, any power ports, anything like that. Craig, why don't you.
Yeah, let me go ahead. Let me start my entrance again here because I interrupted him early. Just as far as how the future's gonna look with our whole no-wires implementation, this is normally, if I just went up, I'm gonna go ahead and set my PC down, and this is normally when I get to work. I have all the adapters in my bag and all those other items. Here we're gonna go and actually automatically, just when I set it down and connect, I didn't press any buttons, I didn't configure any software, but just able to be able to set it down, not only my wireless display, but all the individual pieces. Your USB, your storage, your peripherals with keyboard, all of that's automatically connected just on the-.
It's about 3-10 times the performance, all wireless, going from an Intel chip in here, a WiGig chip and an Intel chip that's here in this WiGig dock.
Eventually, we'll actually have the WiGig connected directly into the back of the monitor. If you had a terabyte drive hooked up to it or anything like that, you could literally just do that as well.
Sure, that's gonna take away 90% of the wires that we have in our huge mass that we have over there. We have one more, and it seems to be the heaviest one that we all have in our bag, and that's that brick. We wanna ditch that out of all of our equations here, and here's how we're gonna do it with wireless charging. I've actually installed a mat on the top of this table, and what's great about resonance magnetic technology is that it can also go through 2 in of material, so we're looking easily to install underneath tables or inside furniture. What that allows me to do that I'm gonna illustrate with this board here is it creates this wireless hotspot that I have here on my desk. Now, what can I do with that type of technology?
Well, that's just easy. Kirk, I'm gonna go ahead. I can charge my Bluetooth earpiece. I have a couple of phones here. Can go ahead and charge that up. Again, multiple devices that we have at once. Here's actually one of my favorites is that we have it so small and simple, we don't actually have to have it built in. In a case like this, I can go ahead and retrofit my existing devices to be able to charge directly on the fly, which is pretty cool. Lastly, not only the multiple devices, but multiple power profiles as well. We're talking about being able to provide power all the way up to the laptop. When I go ahead and put this down, we can see that he's charging. He's not charging. I'm gonna go ahead and light up charging again. Not charging. Just that easy.
No buttons, no muss, no fuss. Just go ahead and put it down, and you're charging away. In 2015, we may never need to plug in again. Pretty cool. Okay, thank you. We're working with all the, you know, furniture office specs. We're working with all the major cafes around the world. We're working with all the major restaurants. We're working with the airline lounges so that the same technology will scale from wearable to tablet and to PC. A very scalable thing that, again, will waterfall through with the IP that Intel's building there. Okay. Last but not least on RealSense.
You know, there is, again, a set of cameras that are gonna come out that are gonna transform the world from thinking in 2D or with one eye to being able to look and see with, full 3D. There's both user-facing cameras, and then there's world-facing cameras for things like tablets and for detachables. This is gonna enable us to do gesture-based gaming. It's gonna enable us to actually do pulse detection and blink detection so you can securely log into your PC with very high fidelity because of being able to tell if there's actually blood flowing through your face, and it's not just something else. Capturing, sharing objects and being able to print them out with 3D printers, doing depth analysis and enhanced photos. I think Brian showed you one example.
We've got cameras going to tablets that are gonna be smaller than and thinner than even an iPad Air. This is the Dell Venue 8 that Brian was talking about with an integrated 3D camera to do autofocus. You've got it going into everything from this new HP Sprout, which was just announced about a week ago by Dion and the new HP Inc. This has a 3D camera actually built into this all-in-one, and it actually is doing projective touchscreen. I can basically, you know, open up applications and get two dimensions. I can scan objects here in 3D. For scrapbooking, there's a whole set of applications, and this is gonna be available for you to play with in the demo center out there.
There's a commercial to take a look at what the new Sprout is. This is just the beginning of a whole new range of projective computing where gesture can play a role in a multi screen workspace. What I wanted to show you here is the first kind of real demonstration of a world-facing camera that would go in a next-generation tablet or 2-in-1. I had the CIO of Emirates Airline, which is right now the largest international carrier in the world, believe it or not, on passengers. They're building an airport twice the size of anything else on the planet in Dubai.
What they do is they actually measure every single box that goes into every single airplane by hand, every single box for cargo, because they actually have to optimize it putting into the containers. They either have to buy a multimillion-dollar piece of equipment that scans and rotates the box, or they actually have to manually sit and measure it with a tape measure and log it. What you're basically seeing here is the airline industry and the whole shipping industry is going positively bananas because you can now literally just take a tablet, point it at this device, and you're getting the dimensions of that box with pretty damn good accuracy just by pointing it there. You could actually imagine we could waterfall this technology into a wearable device, so you could be hands-free in the future as well.
Just one example that has real benefit to driving up the average selling price of tablets. Everything doesn't have to go to a $99 or $79 tablet. We're seeing a lot of applicability. Things like insurance salesmen that go take 24 pictures of a, of a crash scene and a bumper. How long is the skid mark? Forensics in police departments wanting to do a 3D capture of a crime scene. There are so many business applications for 3D and tablets that are getting exciting here beyond what you can expect for consumer gaming. We have eight OEMs committed: Lenovo, HP, Dell, ASUS, Acer, Fujitsu, NEC, and Haier. We have 15 PC designs coming in the spring, more with Skylake.
We've seen Michael Dell up excited at our developer forum on putting it into tablets. This is just the beginning. Very excited about 3D, you saw what it did to CPU utilization, which means it's a sell up on top of the great usage models. What you should expect is right now we have more than 65 applications we're working with. Some of these are very confidential, but you know, you can see some of the biggest names out there from DreamWorks to Scholastic, who does Clifford the Big Red Dog, for those of you in the U.S. here, for kids gaming, from Crayola, you know, the Ability and Skype. A lot going on here on the applications front that really is gonna turn on this 3D capability. In conclusion, three things.
One is I do believe we're making the computer or the PC personal. It is a personal computer after all. I think the strategies and innovation that's happening in the industry with our OEMs will drive growth. I think we're obviously forecasting that. Performance requirements are growing, whether it's Chrome all the way up to RealSense. We think we're on a path to eliminate passwords, eliminate wires, and move the world to 3D over the next 12-18 months. Look forward to talking with you guys at lunch, and I hear your stomachs growling. Why don't we get Diane and I up here, and we'll do a quick Q&A and go from there.
All right. Thanks, Kirk. The procedure will be the same as the last Q&A session. I think in an effort to allow a few more people to ask questions, why don't we do just one question, and we'll skip the follow-up for this time around, again, just to allow more people to ask questions. It looks like we may have one over here. Please.
Great. Thank you. Chris Rolland, FBR. On PC, Kirk, maybe you could talk a little bit more here. Can you help us understand the guidance, sort of flat units, revenue down? How much do we think there is, for example, in XP left here? How much might be channel inventory that we're working through here? What's the sort of mix of enterprise to consumer here, and how much market share is there left for you to take in x86?
Okay. I'll do them in reverse order. I think the last time I stood up here as the data center guy, you guys were saying, "Hey, you're at 88% or 89% share. How could you possibly go higher?" What are you at now?
Ninety-four.
I think if we're sitting at some of our customers at 80% or 85% share, there's definitely room to increase market segment share. That's number one. Number two is we can compete head-to-head with ARM, I think very effectively, and we've embraced China tech ecosystem, and they're now excited about going not just at the $79 or $99 tablets, but going after Core M tablets. I think the next time I talk to you, we'll be talking about a bunch of Shenzhen designs on Core M, not just on Atom, which is another good sell up. The mix right now is about 50/50 business to consumer, give or take. I think we spend a lot of time talking about consumer but, you know, government, education.
We're bidding every week on million unit bids for one-to-one teachers on Chrome. Every time we win Chrome versus an iPad, it's Intel architecture, Intel Wi-Fi in the Chromebooks, et cetera. An iPad wins, it moves over to Chrome. We get the CPU in, the chipset, in some cases, sometimes the NAND, the Wi-Fi, and then that's a big shift, and education's a growing part. You know, the biggest issue I think that's a wild card is Brazil and Russia were our third and fifth largest markets, and given the Ukraine conflict and the Brazil financial situation, those have gone back. When I talk to the retailers there, people wanna buy PCs in those markets.
As soon as those macro issues come back, you know, two of our top five markets are gonna come back with a roar, you know, you guys can predict that as well, as well as I can. I have confidence there.
Why don't we move back over to this side of the room. Please, Ambrish.
Thank you. Ambrish with BMO. Diane, DCG, great job this year, but the last two years were not that great. Just looking out ahead Sorry. I did say great job this year.
You did. Thank you.
Looking out ahead, what's embedded in your assumptions on ASP, and then also in the enterprise? Thank you.
Yeah. You're right. Last year, we grew 8%. The year before that was 11%. I wouldn't say that was bad. We went from 11 to eight, heavily driven by the economic situation, and now, back up to 16% +, as Stacy said. Looking forward on enterprise, you're saying, he said the mix between Oh, ASP.
ASP.
You saw the trends are pretty consistent on ASP versus units, we don't expect any dramatic shift in that trend. As we continue to deliver greater and greater capacity, we expect to see the continued value proposition of buying up the stack. It's a very easy math equation, as James Hamilton, you know, did for the industry, that if you have, you know, fixed amount of data center capacity, the thing to do is to buy higher up in the stack and get the greatest amount of performance you can per system. That gives you the highest performance per TCO. It's a very simple equation. We don't expect any dramatic shift than we've seen historical between ASP and units driving that growth.
On enterprise now is, has fallen just below 50% of our total revenue now is going into enterprise IT. We've said historically we think the enterprise market will continue to grow around 8% and there's really no change there. The enterprise IT market will continue to grow thanks to the build-out of private clouds because of the efficiency that brings, deployment of big data analytics solutions because of the value that brings. Then we'll still continue to see a little bit of the refresh off of Windows Server 2003 end of life.
Great. Thank you, Diane. Scott Baker.
Hi. Harlan Sur, JPMorgan. Question for Diane. On the custom side, you know, obviously great traction here. How does the focus on IP and IP reuse enable you to drive more rapid design cycle times to support your customers? Maybe if you could just touch upon product gross margins, custom versus merchant on a relative basis.
IP reuse is absolutely the key for the data center business holistically, and I think both Brian and Stacy did a good job of showing the leverage between from an IP block perspective, the leverage between client and servers and even back from servers to clients and some of our Ethernet controllers we tend to lead. There is certain IP blocks that we tend to lead on. I think I will be bold and say the server group has led in the move to SoC capabilities, system-on-a-chip capabilities, where we can rapidly deploy a next generation product, you know, pulling different components together and taping that out rapidly. That move to system-on-a-chip capability is obviously important all the way down in the handhelds and the tablets and PCs, but equally as important for us.
It will give us across that spectrum you saw, the way we are delivering customized solutions. It will absolutely help us move further up that custom to the custom CPU side of that, of that, roadmap, where we can take our customer's IP block rapidly integrated into a system on a chip methodology with our own IP. The system on a chip move is absolutely part of that continuum, but it is evolving. It is evolving for Intel, and it is evolving for the Data Center Group as well.
Operating margins, yes.
Yeah, our operating margins.
Relative margins.
Relative margins. As I alluded to, with the James Hamilton AWS example, our customers are willing to pay a premium for a customized solution. The reason they want a customized solution, it's gonna bring them some value, right? It's going to, you know, with Oracle, it's gonna give their elastic compute solution a capability that it wouldn't have otherwise. For Amazon, it's gonna give them significant performance per TCO for a given workload. They're asking for those solutions because they're doing the math and saying they get value from it, and hence they're willing to pay for that value.
It's why you saw on the public cloud service provider chart, why we've been able to grow our revenue so substantially over the time as we've been working directly, engineer to engineer, with those customers to say, "What is your workload, and how can I accelerate it? How can I get you so much more value out of your data center that premium is a good thing to do." It's wonderful margins.
All right. Back up over here.
Diane, looking at your forecast next year for DCG to grow 50% or greater, have you embedded any benefit whatsoever from the expiration of Windows Server 2003 in that forecast?
The Windows Server 2003 end of life is obviously, it takes, as an old CIO, I know how long it takes to convert your applications onto the next generation operating system. It's not something you do overnight. It's a six-month to a 1-year process. Part of our growth this year in enterprise, we believe is because people have been buying, procuring new service and support of those applications moving. And we'll continue to see some of that, but the end of life deadline is July, I believe, right? It's June or July. It's July. We do have some of that baked into the first half of the year volume as well. It's all baked in.
I think we all knew that Windows would be end of life in 2003. It's all part of our model.
Great. Back up over here.
Thanks. Blayne Curtis of Barclays. Kirk, when you look at the 2-in-1 market, you said, I think the Broadwell-Y was 1 million units in Q1, not all 2-in-1s. Just any thoughts on the mix of 2-in-1s and PCs in 2015 and where that would come from, whether it be from the tablet market or notebook users moving to that form factor? I'm assuming the price points probably don't overlap with the 40 million tablets that you have, but is there any cannibalization there? Diane, I'll just sneak it in. Are you really getting four hours of sleep a night?
Yeah. I think on 2-in-1, I'm not sure you said I said it was 1 million Broadwell-Ys. We've been shipping millions of units of Core M into the market this year already that are now hitting the shelves. Again, between 7 and 10 designs will hit for holiday, and then significantly more as we go into next year. What we also said is we're shipping millions of units in preparation for a very early spring 5th-generation Core launch of our traditional Celeron, Pentium, Core i3, Core i5, Core i7, which will be on Broadwell-U. I think the form factors, the way we think about it is today detachable seems to be preferred if you're a tablet-first user, and that's traditionally a 10-inch, 11.6-inch or 12.5-inch screen size.
In that space, to BK's point, when you detach it, you want it to be the best tablet in the world, which means LTE attach rates are gonna be as good, I think, or better than tablets because you're buying a premium tablet. That's the small. The 13-15 in market, like a Yoga that I just showed you, is really about a PC replacement and accelerating the PC refresh cycle. I think, so far fold-overs and detachables are the two form factors that are winning, but you can still find swivels and flips and all these others. I think people are really starting to center more and more on the detachable and it's really a function of are you a tablet-first user or are you a notebook-first user?
If you're a 13 to 15 inch and you're buying a two-in-one, it has to be the world's best notebook. Then I'm basically giving you a tablet for free. It's just a large 13 to 15-inch tablet. If you're buying a 12-inch detachable, it better be the world's best tablet. Then you can dock it in and hopefully save weight and Z height and everything because you're buying a keyboard that was designed to work together. The other thing the OEMs are loving about two-in-ones is they can sell a lot of peripherals. You're getting multiple keyboards per screen now being sold. You'll have a keyboard with extra battery if you want 14 hours of battery life. You can have one that's just a cover.
By the end of next year, you'll have them with charging, wireless charging built into the keyboards. The OEMs like this. This is actually a reason why I think the OEMs are getting reinvigorated, is these peripherals are where they make a big chunk of their money.
Great. I think we can fit two more in. Let's do one back here, and then we'll come back over to the other side.
Oh. Thank you. Chris Caso from Susquehanna. A question about unit growth assumptions, and it's one of the things that came up on the earnings call as well. You know, given we're a little farther in, is there any incremental information you could provide that helps to reconcile what Intel's seeing with regard to unit growth rates as compared to the overall industry? And as you look into next year, what's embedded in the 2015 assumptions with respect to unit growth?
That was a server question? PC? PC question?
Yeah. The PC growth.
Yeah, I think what Stacy said is he's assuming, at least for the financials, flat units and slightly degrading ASPs. I think that's a function of mix because as you saw, the Celeron and Pentium is growing faster. Our Core i3, i5, i7, every time we come out, we basically keep telling you we have new records on vPro, new records on i5, new records on i7, and new records on our K SKU, which is our overclocking stuff. The high end is very, very healthy. The value space is very, very healthy. When you see an ASP decline, it's been primarily because there's we're gaining more share back from ARM against Chrome, like the Samsung conversion or against AMD in the value Windows and space.
Relative to unit growth, the reason I'm bullish is we have a new Microsoft operating system. I think security is more important than ever in business. I get a lot of questions on will Server 2003 affect corporate IT PC purchases. I've been talking to CIOs all over the world, and I think the answer is no, we don't think it will. Refresh rates are pretty locked on, and they're gonna just keep buying PCs. I think it'll be, as Stacy said, more consumer next year than business. This year was more business than consumer.
It'll still be more mature than emerging, and if we can get some of the emerging market I personally believe that as the same saturation in tablets happens in emerging market that's happening in the U.S. and other markets, that people will go back to the PC again because we have price points that are compelling for emerging markets. We've been saying that roughly that trails the mature markets by about 18 months or so. I think that's still a wild card. Then you have the macroeconomics and political instability of Russia, which has been, you know, Russia at one point was the 5th-largest PC market in the world. Brazil was the 3rd-largest PC market in the world, and both of them have fallen off to Japan and Germany.
When those come back, I think there's more tailwinds as well.
All right. We'll sneak one last one in. It'll have to be a quick one here in the back of the room.
Great. Thank you. My question's about Skylake. You mentioned it briefly, but when you had this kinda later than seasonal ramp of Broadwell, is Skylake gonna be kinda back to a back to school type of launch? You know, does that mean you have a shortened life for Broadwell?
Yeah, we haven't publicly disclosed the ordering of our products, but what I can tell you is, we'll have a very robust desktop ramp and a very robust mobile ramp for notebooks and two-in-ones. We're expecting to have both of those products ramping significantly in the second half of the year. As it sits today, you know, things can change, but it'll be a cleaner and more rapid ramp than you've seen with Broadwell this year. We have more headroom in the schedule to have like a, you know, a complete holiday kind of situation.
Okay.
Thank you. All right. Thank you all very much for your patience. I know we've run it a little bit long. We promise we'll get you out there to feed you right now. Again, seating is very informal for lunch. Grab any chair you like. There will be one or two executives at each and every table, and that'll be an opportunity for some informal networking. Thanks. We'll see you in about 45 minutes.
Ladies and gentlemen, please welcome Renee James.
Good afternoon. Welcome back after lunch. We can grab a seat. Thank you all for hanging in with us. We had a great morning, we have a wonderful afternoon planned for you. We're going to pick up on the theme that Brian started this morning and talk about what we do with the 75% of the $11 billion of investment, and how we use that in the adjacencies. We've picked a few of them to talk about in detail with you and give you examples of how those platform investments lead to differentiation and market leadership for us across our product lines. We talked a lot this morning. I'm using the same slide. I think Stacy did a good job of explaining it in detail, I don't need to do that for you.
A lot of our investments, as we've talked about, are utilized in different segments. We talked about the core of them, and this afternoon we're gonna talk about how they extend into platforms. I think many of you know that in the technology industry, platform investments are one of the key ways that IT companies create additional value by aggregating different assets together, creating leadership and differentiation at the platform level, and getting ahead the next generation to the next generation. That's really what we're trying to do with the investments we're gonna talk about this afternoon, extend the Intel Architecture into emerging and growth segments where we may or may not be, and not only find new growth for our business, but also extend our platform leadership.
I'm gonna go through several of them, and then we're gonna drill down on three of them, and I have the GMs of each one of the businesses to give you wonderful details and road maps and all the things that you would like to hear. I have no haikus. I have no demos, like Kirk. I know, right? I'm gonna go pretty quick. I am going to jump up if they run too long and say, "Okay, thank you, Hermann. We're done." I'm not picking on Hermann. I love Hermann. I'm your afternoon talk show host, is the way I would think about it. Let's talk first about mobile communications. I know you're all dying to hear Hermann. He's gonna give you a lot of detail on this.
I just want to reiterate a couple of key things that we should think about here. Communications in general is one of the emerging areas of key platform differentiation. As you know from the things that have happened in the market this year with the exits and some of the changes, Intel is emerging as one of the leaders in this area. I've talked to some of you at lunch, we talked about this, many of us have discussed this in other forums, it bears repeating. As you think about the investments that we're making, think about some of the R&D at the platform level like this, I think Brian started to talk about this this morning.
When a decade or two ago, a decade and a half ago, when everything was about graphics and Windows and streaming video and everything was brand-new and we were trying to do more and more, we made tremendous investments at the platform level in graphics. We were nowhere. Graphics was something that everyone said, "Why are you guys doing this? You're nowhere in graphics. Graphics isn't part of what you do." Over the course of a decade, we became the leader in integrated graphics as we are today. Graphics is just taken for granted as part of the platform, but it's a super high-value part of the platform.
I think you can, as we look forward, all the segments of computing from wearables, IoT, phones, tablets, PCs, and yes, even into the data center, as Diane described as she talked about what's going on in the network communications business, comms in general, multi-comms, is a strategic asset. It's a platform-level asset. Strategically, that element is something that if you have it, you're a have versus a have-not in the industry going forward. It's going to become a point of differentiation as it gets integrated, much like graphics did a decade ago. We think about it at the platform level as not only a BU investment, not only, of course, as part of getting into phones and tablets and all those things, but as a strategic investment for the long term. Security and privacy.
We're not having a drilldown today. Chris is here. I don't know where you are, Chris. I just saw you. Christopher Young, who just joined us, that Brian introduced this morning, is here, and I hope all of you will get a chance to talk to him. We are continuing our investments, and Stacy mentioned that this is a multibillion-dollar business. As a platform asset, those are leveraging the things that we've done in the core of the product line, both in Core and Xeon and extending into Atom, as well as the technologies we acquired in combination are becoming a platform asset. Very much like I just gave you the story on communications, over this next decade, security is becoming one of those next platform assets where you need to be able to integrate it into the platform at multiple different levels of integration.
We continue to do our work. The thing about security that's important for this next year is how it plays into our core product line, and it's end to end. Security is from the client through the edge, through the cloud network into the data center. You'll find that many of these platform technologies are end to end. One of the unique differentiators about Intel that Brian talked about this morning is that you can design one product with multiple operating systems, one product with multiple points of differentiation at the platform level end to end. This is important for creating solutions, and solutions is the way that we think we are gonna generate, from software to silicon, more value on our platforms. Internet of Things. We're gonna have a drilldown on this, and you'll hear from Doug.
Internet of Things is very exciting for many reasons for us. Brian talked about how we are already in this business, and we've been in embedded for a long time. That's great news. It leverages connectivity, it leverages software integration capability, more importantly, it is sold as a solution end to end. One of the things I'm very, very excited about is that this group is the furthest along, from my opinion and all the decades we've worked on end-to-end solutions, in really realizing the value of being able to create vertical solutions for different segments of the market. It's a multibillion-dollar business, it's growing rapidly, we're able to sell solutions here, that's beyond the silicon, so silicon and software. It's a very exciting emerging segment for us. In wearables, Diane demonstrated. She's still wearing it. She has her bracelet.
One of the things that at the platform level, we think about wearables, and people say, "What are you doing in wearables?" You know what I think about wearables? We're getting the next generation of developers excited about the Intel platform. We're not missing the next big thing. We're out there with development kits and ideas and prototypes and partners, getting developers around the Intel architecture. This is critical, as many of you talked to us about getting into mobile and all the work we had to do to get Android developers caught up on our platform. Wearable developers are being born on the Intel platform. I think, you know, Brian's enthusiasm in embracing the Maker community is a huge piece of this as well.
These are the people who are designing the products of the future, when you are talking about platform investments and proliferating the Intel architecture for the future, this is a big part of why we would do that. Finally, certainly not last or least, is storage and memory. The industry always, you know, goes around and around about storage and memory. At the end of the day, one thing is very, very clear. As you build bigger and bigger systems that are more and more complex, memory and storage become even more strategic. I think over the horizon at the platform level, not to say it's not a good business, and Rob will talk about it, yet again, another multibillion-dollar business.
At the platform level, this is a very strategic area of investment because it becomes either the opportunity to accelerate or the bottleneck. It becomes another high-value portion of the platform that we want aggregated around Intel. Strategically, it is an opportunity for growth for us, so Rob will talk about that. When we look at, you know, looking forward for the next N number of years, this is kind of the list of what I would call key platform technologies that we're investing in. When you look at that stack of the shared investment that we talked about this morning, inside of that are these investments. They're not that visible to you, so we thought we'd take a minute and just call them out.
As you listen to the afternoon presentations, listen to them from the frame of mind of, okay, they're laying tracks for the future here. They're also building new multibillion-dollar businesses, but as Stacy said, a $2 billion business on the scale of Intel doesn't always show up. That's really kind of why we aggregated them together this afternoon, so that we could look at them and think about the future strategic position that Intel will be in as a result of having made these investments and the adjacent growth opportunities that go with them. With that, we've got three that we're drilling down on. The general managers for the others are here. We have demos on everything else, so we're prepared to talk about everything. In the interest of time, we picked these three.
We're gonna start with communications and mobile technology. Just quickly just to talk about what the strategy is at the platform level, it's leadership in connectivity. By that, I mean multi-coms, so not just cellular, right? It includes Wi-Fi, Bluetooth, NFC, et cetera, et cetera, et cetera. Leadership in multi-coms is a strategic asset for the corporation across all platforms. That's the number 1 strategy from a platform perspective. Second, of course, the platform, mobile platform leads in the low power, and we talked about some of that this morning, and Hermann's gonna talk about it more. It is the point of innovation, and you saw in Brian's color chart, the green arrow coming back. Innovations that come back into the core from the low power team, very important from a platform.
Of course, not talked about much, but something that is absolutely critical and many of you have asked about is the development methodology around SoC and rapid derivatives. This team, one of their platform jobs, right, because we're all one big R&D team, is to really innovate on new design methodologies around rapid derivatives. With that, I'd like to ask Hermann to come up, and we won't give you the hook, Hermann, and talk in detail about the mobile strategy. Thanks.
Thank you, Renee, for the introduction. A warm welcome from my side as well. I go a little bit deeper. First of all, I think I want to share with you a few market trends that we see and that are leading us in our decisions on where we orient our focuses. First is the smart mobile devices continue to grow. We continue to invest, has been said before, and that stays as that is. The next thing what we observe is that the low end is actually growing much, much faster than the high end. Our strategy is also to participate in the entry segments and in the value segments. Another trend that we observe is that the LTE growth is actually much, much faster than any generation had that before.
It's very obvious why that is. That makes us doubling down on our investments in communication technology and push the envelope here even faster than we did ever before. Last but not least in that slide, we see a growing trend on phablets. 60% maybe already counting in this. Also here we take our tribute, and we will see later. What we do is we develop platforms that can go in any form factor and any shape. Let's take a look back for 2014. We are still in '14. We took a very focused approach in this year, and that focus was pretty much anchoring around three points. One is we wanted to grow Intel Architecture and the footprint in mobile. We did that first on tablets, and we are focused on this one, working on cost reduction and differentiation.
Next pillar, grow our leadership in communication and connectivity. Develop a full portfolio, as Renee said before, has been one of our focus points for today. Capture and lead in the low end and entry segments where the biggest growth is. Build partnerships here and see that we get a foothold in the China tech ecosystem where a lot of that takes place and build partnerships. Before I go into the details, I think we can go and score and see how did we actually do against them. The year's not fully finished, so we have a few weeks still to run, so don't count all the chicken before they are hatched. I think we can at least speak about where the trajectory is and what we think it's going to happen.
We believe we are on a good track towards the 40 million tablet volume that we had given us as a target. Last quarter, the quarter before, if you look into those numbers, we see that that scored us actually to be the number one merchant silicon supplier into the tablet market. If you take Apple's captive volume into this calculation, it leaves us on a number two spot. I think that is a reasonable achievement. Nobody of all of us, maybe even including me and our CEO, might have believed that we can achieve that audacious goal that we had given ourselves. We are on track to this. We delivered also on world-class advanced LTE. XMM 7260 generation carrier aggregation, CAT6, 300 Mb per second is shipping. It's a globally competitive platform. We have accelerated the rate on how we innovate.
Look at the generations in 2012, not present. In 2013, maybe half a year behind the market leader, and in 2014, at time to market with the market leader. I think we have delivered on this as well in this year. Last but not least on that first integrated Intel architecture, integrated SoC communication technology and application processor. We are on track here. Will PRQ, means production release, have for this product by the end of the year and then ramp it with customers in the next year. Let me go into more detail. That is pretty much saying everything on what we want to talk about. Here's a few more details. There are 40 million tablets.
If we drill a little bit deeper into this, we see on the statistics, first, we made big great inroads in entry and value segments. You might have said a year before, "Yeah, maybe they get a little bit on the high end and the performance segment, but will they ever be able to get into other segments?" Yes, we are. You see price points $99 and below and stretching up to $400. We can cover the entire price stack in the market. That is done. You see on how that proportion is, how many designs we have in those sensitive segments, and I think that is a big, big step forward. The company is changing, and we know how to maneuver in this market. Next thing what we can see from that statistics is, you might have said Intel is a Windows house.
They cannot do Android, what is about the compatibility, and so on. Already, those numbers of designs show that we put our mind there where the market is, we focus there what the market demand is. Sure, we all know there's a lot of demand in the Android space, so we have many, many more designs in that space with our customers. All in all, we count about 100 OEMs and about 350 ODM design SKUs that our customers build on our platforms. They may not see all the light of the market, we have enabled so many designs that customers have something to choose from. Our customers can go and offer different SKUs and different designs in their market. That is the heavy lift that we have done this year. We have long-term partnerships developed.
They are shipping, they are growing. What we see today is we have about 20% of wireless WAN attached, so cellular attached. That part is growing every day, so to speak. Last but not least, it's not yet the last, what we have in this market and brought to this market is a completely new way on how to engage in this ecosystem. With channel partners and helping our customers in marketing, helping our customers also with how the way they can sell those products, and opening up channels for our customers so that this circle gets, so to speak, closed again. Good platforms, chips from our side, then in the end, when it goes and sell them out, find their ways, we have a network around the entire planet of resellers, retailers, and so on.
We can help with this whole ecosystem that we have created around tablets with our partners. Last but not least, of course, we are working hard on the platform, EBOM cost reduction, and that brought us step by step further in this market as well. I have a few showcase designs here. First of all, I would like to take the left one here. That is the thinnest tablet on this planet. 6 mm thickness. See height. Has a wonderful display. What we have done in this tablet here is we have integrated more cameras. With this, we can take depth photography. Easily speaking, those cameras do what our eyes do, and they memorize their picture in different levels of depth. That helps to do completely new stuff.
A complete new ecosystem can build around this. We can show you one example, what we can do with this. As that picture, as we know, or the technology, so to speak, knows the depths of the picture, we can make variations that depend on how far the objects are away. You see this here on how the color goes through that picture. That means depending on the distance that gets colored or stays black and white, we can do this from back and forth. It's a nice game, but it gives an indication on what could developers actually do with this technology. We are just at the beginning of those capabilities. Our picture of next year is there will be tens of designs out there with that technology, maybe hundreds.
There will be thousands of apps developed to that technology. There will be million of people taking pictures with this and billion of people making these kind of modifications. The modifications, by the way, you can do on any device. Just if you have taken that picture with this wonderful technology, you can have everybody in your family play with it. Another example where we brought innovation to the market is, for example, the Fuhu DreamTab for kids. Very nice design, very robust design. It's a powerful design. It's based on Bay Trail. If kids are in bed, it's also good for adult to have a powerful performance tablet. Another innovative design, sleek, unique, two-in-ones have come out of this portfolio. I have here the Lenovo Yoga 2.
That device, even here, this one, that has a camera, a projector in it. You can take this and project what you want to show, not only on that display, rather also against the wall or whatever. A screen. Let me change gears on you, let's step over to on how can we actually make this being so fast. How do we get the customers so fast on 350 different SKUs? How do we do this even faster the more and more we do that? This is our turnkey program.
The turnkey program means we have a master reference design, so customers can, if they want, take that right away, or they take any kind of combinations of components that we have pre-validated and verified that this works, and we give them tools that they can board very fast, that they can make modifications. We give them support, whatever is needed here, and with this, we can bring a customer up within weeks. After platform is finished, customers can start build on it, and that works within weeks. What we can give around this and on top of this, that is the Intel brand, our channels, and our matchmaking. We even run trade fairs with our customers and help them find other partners on totally different places on that planet where they can make business with based on our designs.
They know once we are with our marketing in this, that helps them sell their products. With this, I spoke to the 350 global designs, 30 ODMs already on this design methodology in China, printed circuit boards, assemblies, as people doing intermediate steps. One thing is important, that is what I wanted to mention, this is already this year why we were ramping up that capability. 45% of all the tablet devices that we are shipping comes out of that capability. Already while ramping those capabilities, that makes up for 45% of the numbers. How powerful this is, maybe I can give you somebody who can speak to this much more authentic than I can.
I'm Awen Wang, the President of E Fun. We're an established tablet and mobile device manufacturer in China. We have been building products in our Shenzhen factory for more than 20 years. We have shipped almost 15 million tablets to market in the past five years. Intel's turnkey program has allowed us to keep costs down and deliver our first Intel product with exceptional efficiency and speed. It took just two months to go from RFP to shipping our new Notebook tablet that will be available at Walmart on Black Friday for $99. This is our first product with Intel, and now we enjoy a great partnership. I'm very impressed with the dramatic changes that Intel has made in such a short time to support our new product.
A year ago, we couldn't do business together, but now I believe we will have a long-lasting partnership to deliver high-quality Intel architecture and brand to our customers.
I couldn't do this any better, so that is authentic. Let me change gears on you. Performance and experience matters. We heard from him that these platforms have a performance, and they distinguish in performance, not necessarily always in core count, but that is an architectural choice we take on how many cores do we put into and how powerful are they and what is the best choice in terms of numbers of cores and power of individual cores. With this, we achieve good scores in the benchmarks. If that was a trade show, I would speak much, much longer to that slide here.
For this auditorium, it should be I think I want to say we have those comparisons, and we have those spaces where we focus on and what is important for the user experience, where we invest into those platforms and make them better in delivering you a better user experience with this. Another topic I want to address is the applications. They, of course, run on that performance, sure, and you would always have trusted that Intel can do the best for a Windows environment. Sure. They know that for centuries. Android, you have always been asking, "Is that compatible? How does that work? Can you really do this?" That is all designed around ARM, and so on and so forth. You know all of that because you have asked me those questions. I can just tell you it just works.
We have made it just work. Whatever that is, whether what kind of app that is, we wait every day to customers to call us and say, "Here, there is an app with a problem." That is what we are waiting for. We have a team for this, and we wait every day if somebody comes, and then we jump on this and fix that problem. In most cases, it turns out it was pretty much bad, poorly programmed. After we help those app developers to program it right, it even ran better on ARM devices as well. That is what we do for this market, and that is what we provide. Once those people are in communication with us, they start to think about, hey, what is actually if I go and do this natively?
If I compile this right away natively on Intel architecture, then I can unleash the power of that architecture for me. Step by step, one after the other, they go and do their apps natively. That is, of course, what we prefer, because that unleashes the power of our architecture. That comes with the full support for developers in that ecosystem with software development kits, with development tools what they need completely for Windows as well as for Android. That is the way on how we have made our way into the Android ecosystem. I want to give you a few examples on what that led to. Here's one example. This is a Nexus Player. This Nexus Player does not only play Netflix or Hulu, you can have access to the complete Android ecosystem with that player.
It's also optimized for gaming. It's based on Moorefield. We came within one year from you asking us on, how are you doing with Google? Is that really compatible? Would they ever look at something else to have a Nexus in the market? Google has chosen Intel architecture for that Nexus device. That was actually a good piece of work because that brought a lot of engineers together and optimized the Google software and the apps in order to run on that device. With this, we get into this market step by step and get the credit of that whole ecosystem. What can deliver more credit than a Nexus device? Another example is our reference design. We call that IRDA, Intel Reference Design for Android.
That is a design which is very narrow in the component list that we have agreed with Google on, but is broad enough for our customers to differentiate. If customers stay in that envelope, they get a quick access to software because we work with Google to make that software always concurrent and work, and it can be distributed with the lightning speed. That helps customers, and it helps Google to have always a concurrent software load on the devices out there. You see on our shipments on Android, those numbers are constantly growing as a result of that what we do for that ecosystem. Now I want to change over, shift gears on you to smartphones. You have seen the new family that came from ASUS, Zenfones, the ASUS PadFone X Mini. Brian already spoke about that before. It is really a good deal.
$199. It's an LTE device here in the U.S., available in multiple stores. $199. Everybody asked me when that came, is that really without a contract? Yes, that is really without a contract. Go and buy one of them, two of them, three of them for your family. It is a good deal. Another cornerstone for smartphones are, of course, modems, communication technology. Intel LTE Cat 4 shipping in Samsung Galaxy Note 4, shipping in the Alpha, and also shipping on the LG G3. Towards the end of the year, you will see more devices in the market using that technology, including TDD in China and TD-SCDMA. Last but not least, for that slide here, we have also new partners with whom we work.
For example, Lenovo will come up soon with new products, but also partnerships like Rockchip and Spreadtrum, with which we go in a completely unprecedented approach into this market. Complete new engagement and way on how to go to a market, how to broaden a footprint, and being fast and nimble in the market. Innovation does not only happen in labs or in research institutes. Innovation also happens in business models and engagements and in processes and procedures. The LTE RAM is always something that comes up, and Renee spoke about this. Communication technology is an anchor point. We are shipping today in the U.S., in Latin America, in Europe, in Southeast Asia, so to speak globally. I have here a few logos of all those operators at which we are certified. That is a global certification. This is a true 5-mode solution and globally certified.
That is what we want to have in the market, that is what we believe what customers eventually all will need. We have more than 25 designs in the market. ASUS, Samsung, Dell, Lenovo, these are the brand names in which they will show up. We are also pursuing other legs. With modules. Modules help us to be in the market faster, they help us also to be in the market for designs or products where it wouldn't pay off to right away do a sold out on development. For those markets, we have modules which can be equipped or not equipped, put on or not. Our module partners are Huawei, Sierra Wireless or Telit. Those modules go also into the wide open market of machine to machine and Internet of Things. We are present in that with all the technologies we have at Intel.
If I sum this up, this means we have a very solid foundation to fuel growth for the future. We have built strong partners. With customers and with partners, we have a design pipeline, customers taking our platforms. We support ecosystems, and we have built an ecosystem as we have shown in the tablet space, and we have a solid roadmap of products. Moving in 2015, we have, sure, of course, focus points for our expansion. One is the SoFIA family. I said before, it's on track to be ready by the end of the year. It's the first Intel integrated architecture with communication SoC. I want to give you back again the numbers. It's a 3G version. It's a SoFIA 3GR. It's an up, a performance up version of this, and SoFIA LTE.
All those platforms are designed that our customers can build any product they want out of this. Here I have a few examples. This is the MRD, the Master Reference Design. That is a 5-in one, and that is a 7-in one. Out of the 5-in, customers can build anything from 3.5 in, 4 in, up to whatever they want. The 7 in is optimized for if you have more space, you can do a cost-optimized design. For that reason, we have built that 7-in design as well to show customers here. If you want to have a cost-optimized one, 7 in and above, you may want to take this because that can be even more cost optimized than the space and volume optimized design here.
Whatever customers want to build, whatever kind of market they want to serve, screen size, change, go and do. They are ready and prepared for this. Those platforms will also work with our partners. Those partners will also go on those platforms and help us proliferating this and bringing that footprint into the market ever faster than you ever could imagine. One click deeper on the communication and connectivity. That is an advantage for us because we have all the technologies we need for that market. We have 2G, 3G, 4G, aka LTE. We have NFC, we have positioning systems, we have Wi-Fi, we have WiGig, we have Bluetooth, we have FM radio, everything what we need to build those products. We put that together as we think we need this. Our capabilities allow us to integrate everything what we want to integrate.
We are maybe not integrating everything. You have never seen us integrating a 3G RF with a 3G baseband because we are smart enough to touch that, rather go to the market as the market needs it. You have seen us doing that in 2G, and you have also seen us integrating even a 3G power amplifier into a 3G transmitter. We have the secret sauce, we have all those co-building blocks, and we have the secret sauce of system know-how on what goes together well, best and what the market needs. That is how we integrate this. One example is Sofia 3GR. The RF chip comes with an integrated 3G RF, with an integrated Bluetooth, with an integrated FM radio, with an integrated Wi-Fi, and with an integrated global positioning. All integrated, all our own technology and it's all low power. Also Wi-Fi.
Intel is a very known, very well-known Wi-Fi expert that has been all brought down to low power, and that is in all in one chip. That is what we can do. We can break it out and build a discrete LTE modem as we do for the high-end roadmap, or we build modules and bring this in a different form and shape. You say, "Okay, that is engineering. What do you do in order to protect that? What do you do for the ecosystem to push that forward?" We are present in all standardizations that are necessary for this, and we contribute heavily and actively in those standardizations. When it comes to connecting all those devices, we even have driven this convergence with the Open Interconnect Consortium that we are driving. It's an open platform that can connect all those devices seamlessly.
That is the philosophy with which we drive this market. Before I come to the end, I have a few more boring slides for you, maybe. It's the roadmap. It's obligatory now that a roadmap comes. For 2015, in the performance and in the mid-range, our working horses are our LTE solutions, the XMM 7160, XMM 7260. Shipping today, working horse. Towards the second half of the year, we will have XMM 7360. 3 carrier aggregation. It's CAT9, 450 Mb per second. Coming to the market, customer sampling next quarter. Bay Trail is our working horse today. It's shipping in high volume. Moorefield is a working horse today, shipping in high volume. Towards the end of the year, we will qualify for production Cherry Trail. Cherry Trail will then go into volume in the next year.
I have an example here, if you run around later, you can also play with the Cherry Trail technology. For 2016, you see Broxton as the next generation, and you see SoFIA Mid. This is important. You see that we bring the SoFIA architecture in the higher market segment, into the mid-range of that market segment. We broaden this architecture, and we broaden those capabilities and bring it there. Last roadmap slide for the value and entry. Bay Trail is currently the working horse here. That will be phased over to the SoFIA 3G, SoFIA 3GR, and the SoFIA LTE. In 2016, you will see a 14 NM version of SoFIA out of our own manufacturing.
Now I want to preempt a few questions, and I already had this at my lunch table today. What are you doing exactly with Rockchip and Spreadtrum? If I knew really what we are doing exactly, no. What we do is we have with Rockchip, we started with the SoFIA 3G. That was the first SoFIA platform we had. We said, "We go and engage with this." We built together with Rockchip, the SoFIA 3GR. Both of us can sell that product, we own this product, and we have one more product on our platform. That's the whole philosophy. This is no big secret sauce. On Spreadtrum, we start first on the SoFIA LTE, go with an agreement with them on the SoFIA LTE, and then we build from there.
There is more in the drawer, but nothing more to release. That is how we started. These are the platforms, the products that you will see us collaborating on. With this, I come to our next year, the focused strategy to win for 2015. It's not very much different than this year, but actually a good company usually has continuity. The Intel architecture mobile footprint. I think that's very obvious on how we want to do this, the MID, our partnerships, our customers. The global leadership in communication and connectivity with all those IPs in-house and the capabilities to integrate them, I think that is very straightforward. We want to capture more of the growing value and entry segment. Very simple. SoFIA is the anchor point for this.
China tech ecosystem is the hotspot where we have to be for this. We are there, and we have partners there. You have seen one of them before. Thank you.
Thank you. I should have mentioned this up front. We're gonna have Q&A with everybody at the end, so, 'cause there's more in the drawer from Hermann, he'll be coming back, and so get ready to question him to find out what's in the drawer. Equally exciting is the Internet of Things. Doug knows this 'cause I'm, like, the biggest cheerleader for him 'cause I'm very excited about this segment for a bunch of reasons. You know, Brian talked about silicon, software to silicon this morning. I told you in my preamble comments that I think Doug, not to put pressure on him, has the best opportunity to really build end-to-end solutions out of his business, and he's gonna talk about that.
We are able to use our leadership in connectivity that you just heard about, as well as the low power investments, and really turn them into something to follow on to a great embedded business that we've had for a long time. I think all of you know that we put the Wind River assets in with the group. That's been very helpful. Doug's gonna talk about it a little bit, but I think that allowed us to even get more out of that investment. As Brian said, on the operating system side, software in general has become a bigger capability and differentiator for us. When we start to go into vertical solutions like IoT, software is a very big deal because the customers expect you to deliver the entire stack with the silicon.
This is part of the reason that we did that integration a little over a year ago and is part of the reason that I'm very excited that I'm seeing Doug being able to do so much with that. Without further ado, I will like to bring up Doug Davis to talk about a wonderful opportunity for growth and a big platform opportunity for Intel.
Great.
Thank you.
Thanks.
There you go.
I've realized through the course of the day that it's kinda great to go a little bit later in the day because all the pieces are laid out, now I just get the task of pulling it all together for you and pulling that whole story together. There are really three things I wanna talk about today. The first is around how we have this opportunity. Today, we have a big business. Already, we have a big footprint. That we're evolving from that footprint that we've been building over the last 30 years. This, you know, number of devices, these things that are out there that have integrated Intel architecture already.
The second is how we're utilizing the other technologies and capabilities from across Intel in many of these different market segments, and how we can deliver those technologies and bring value into other market segments. Lastly, I'm gonna shift gears a little bit and go from talking specifically about the business to talking more broadly about the end-to-end capabilities that Intel has when we talk about the Internet of Things. It's not just about the endpoint devices, but as you've heard throughout the day, the capabilities to create these end-to-end solutions and the value that can come from those. I wanna start out and talk a little bit about a definition as well, because as I've been out talking to customers and analysts and many of you, the description of the Internet of Things is a little bit different.
We had a great conversation even at lunch today about what is the Internet of Things. I thought, you know, this picture might help. You may have seen different versions of this from different people around Intel or even other companies. For us, it really starts with those sensors. The sensors are the eyes and ears, you know, the temperature, the vibration, the rotational speed of a shaft in a motor. Even cameras are considered sensors that are helping us understand what's happening in the physical world. That sensor data then comes into these computing devices where we can do something with that data. We put these things into three different categories. We talk about the home Internet of Things.
These are devices such as PCs and tablets, game consoles, but increasingly things like home automation systems, home security systems, that where we need to be able to aggregate different protocols and aggregate data from these different types of devices, and then to be able to feed that data back into the data center or cloud. That's typically done through a home gateway device that serves that function of aggregating data and creating that connection back to the Internet. The other area is around what we call the Industrial Internet of Things. It's machine tools, it's medical instrumentation devices, it's even that windmill out in the North Sea, right? It's aggregating data from sensors in these kind of industrial applications into typically an industrial gateway.
That gateway can do local analytics, feed data back into the data center or cloud. The third area we describe is the mobile Internet of Things, and these are devices that connect directly to the broadband wireless infrastructure. Of course, smartphones and wearables like Diane's wearing. Things like cars and trains, those things that need to be able to connect to a broadband wireless connection in order to feed that data up into the data center or cloud. These devices need to be able to use secure, interconnected intelligence to deliver on the capabilities that I'm describing, and then use data analytics to be able to extract information from all that data that these things are creating. Connect devices through the network infrastructure and then into the data center or cloud.
Now the other thing about the Internet of Things is there are numbers that are getting thrown around that are huge. You've probably heard the numbers from John Chambers in terms of the economic impact of the Internet of Things. There are predictions from analysts that by the year 2020 there'll be about 50 billion devices connected to the Internet. Those 50 billion devices will generate about 44 zettabytes of data. Yeah, I had to go look it up. 10 to the power of 21. That's a zettabyte. Then once those things are connected, you know, that gives us the ability with all that data to do analytics. Some people have looked at this and said, "Well, isn't this just kind of a different name for the embedded business, right?
This has kinda been going on for a while. I really see some changes as I'm out talking to customers, we're seeing these dynamics from an economic standpoint beginning to shift things. The economics are driven by the fact that you see at the bottom of the slide. The cost of sensors have come down 2x in the last 10 years. These things need to be connected and that connectivity is becoming more and more pervasive. Why? The cost of connectivity has come down 40x in the last 10 years. As we gather that 44 zettabytes of data, we have to be able to do the analytics, and that requires more and more computing.
Computing at the edge of the network, computing within the network in some cases, and certainly big data analytics and computing of all that data that goes into the data center or cloud. The cost of processing has come down 60x over the last 10 years, thanks to Moore's Law. You know, how does that translate then into our business? As we've looked at this space now and how we wanted to focus, we focused this business into a number of different market sectors. I wanna give you some insight into some of the numbers and conversations I've had with many of you. You've been asking about, you know, what this market looks like. We've structured around market sectors.
Renee said earlier that's one of the things that we've done is really put a focus on these sectors. We have a focus around retail, transportation, manufacturing, industrial, and we include energy or utilities in that space because they have similar characteristics. We've always kinda had a broad market business, and this is where we do things like digital security and surveillance, we do medical instrumentation devices. We've created a new sector around smart buildings, and I'll show you an example a little bit later around why we think that's so important and the promise that we think is that's associated with that. You can see across the top how we are positioned in terms of market segment share. We've had a focus in the retail space for several years.
We have about a 40% market segment share there. I won't read them all to you, but you can see how we're positioned in each of these sectors and where we're deriving the greatest growth. Some of these sectors are growing fast. Others may be growing a little bit slower, but it's a characteristic of those markets. Manufacturing, industrial space, moves a little bit more deliberately than some of these others. Our growth in transportation is driven largely with the efforts that we've put into the infotainment space and building our presence in that space in the automotive market. You can see in the upper left, the overall size of the market, and we describe this in terms of SAM based on a number of industry analyst forecasts.
We're giving a bit of a range here as well, $11 billion-$13 billion, and you can see how it breaks out by these sectors. You've seen the revenue numbers associated with the business. Many of you asked about market segment share. Based on that SAM definition, we have about a 17% market segment share, and you've heard the growth numbers. Renee mentioned that Wind River is part of this business, and I wanna kind of expand on two things. Wind River gives us the ability to engage in a broader set of applications. They have a big embedded footprint, they have a legacy in VxWorks, and that gives us a big footprint, where it creates more opportunities for Intel architecture over time, and it gives us the ability to create more integrated solutions.
We're creating a capability that we call application-ready platforms by taking our processor, an operating system from Wind River, an application stack that gives this thing a personality, and even integrating technologies from McAfee to create an integrated secure solution, and a customer then can order that SKU and get all those elements combined together. It doesn't mean that we only work with Wind River. The majority of the business has historically been based on Windows, continues to be an important part of our business in terms of overall size, but certainly Linux and Android, Chrome in some areas, as well as many proprietary operating systems are important for us in terms of this business. Then I'll give you a little bit more insight as to how processors and technologies come to play. You've heard a lot about the adjacent nature of this business.
We describe our product line and how we use products in the business according to this chart. We've used this kind of notion of a pyramid for some time now. We describe the TAM as everything from very high performance computing all the way down to 32-bit microcontrollers. That's the TAM. We're looking at 32-bit processing when we think about the overall size of the available market. Our SAM really goes down into what we call an entry performance 32-bit MPU space. That's the space that we participate in today. We utilize the subset of the product portfolio that Intel has to be able to meet the needs of our customers in each of these different areas.
We'll pick a subset of products that enable us to give customers a range of performance that they need, to offer long life capabilities, and even in some cases for these products, depending on the market sectors that we're in, we'll offer extended temp capabilities as well. We'll take this subset of these products to be able to support these different market segments and bring the domain expertise from those sectors into how we support the products and how we support customers doing development. You can see now with the addition of Atom over the past few years, it's one of our fastest growing product lines today in terms of overall volume.
Quark now gives us the ability to go even deeper into the pyramid, to be able to support another layer of processing where customers see value in what we can deliver from an Intel architecture standpoint and the kind of integration that we can provide as well. We take advantage of all of these technologies across Intel and bring them into these different market segments and take advantage of that scaling and the adjacency of the overall business. Stacy talked a little bit earlier about the importance of us looking at the business and how the future looks in terms of the design win pipeline. This is really important for us because as you look across those different market segments, they have different characteristics. You can see in retail typically moves pretty quickly.
We introduce new products, they're able to adopt those products, get them integrated in their products, and deploy them fairly quickly. In some of the industrial or manufacturing spaces, that process takes quite a bit longer. They go through a much more rigorous qualification process, and that timeline from design win into volume production takes a bit longer. We need to be able to have this robust pipeline of design wins to continue to fuel the business over the next several years. With that, I wanna shift gears a little bit. I wanna open it up now beyond the specific business here within Intel and talk a bit more broadly about the Internet of Things. We worked with an analyst.
We went out and talked to a number of companies that are doing smart factory implementation, smart buildings. They're doing, you know, medical type of applications and really trying to create end-to-end solutions as defined by that Internet of Things definition, right? You can see the sample size is 37, and you go, "Well, that's a relatively small sample size." These were 37 face-to-face interviews going into a deep discussion about the challenges that they face in creating these kinds of end-to-end Internet of Things implementations. You can see from the green bars, it describes the performance of today's solutions, how capable are the things that they have available to them now. That yellow line describes really how important it is for them to be able to complete these kinds of implementations.
You can see, you know, over on the, over on the right-hand side, their requirements and how important they are are pretty well matched up. Well, that's good. We're delivering great products for them, right? Where they have challenges are around security, device interoperability. You have different sensors oftentimes speaking different languages in terms of physical protocols, and that device interoperability becomes really important to them. The IT/OT integration. There's a lot of talk in the industry around this, around that infrastructure from a data center standpoint that directly supports factories versus the data center infrastructure that supports the other operations and the desire to converge those. Advanced analytics, right? The ability to now extract information from all that data and then ease of use.
It's how complex is it to bring all this together and to make it all work. Based on this data, we've put together five tenets, five elements that we think are important in order to create these kinds of end-to-end solutions, and this is really guiding the development work that we're doing around Intel today. Of course, it starts with security. Security is absolutely the foundation, we all read the newspapers, key focus, and it's security both in hardware and software. It's integrating security capabilities into our processors and combining that with the software capabilities that we can deliver as well. It's having those two things interoperate that's important. The next is around connectivity. By definition, these things are connected to the data center or cloud, so connectivity.
But then once you start connecting billions of devices, it becomes a challenge around discovery and provisioning and the management of billions of endpoint devices. I said earlier, you have all these sensors and these devices that need to talk to each other, so data normalization is important, right? The ability to have these things share data and then to be able to look at the effects that these different data sources have on each other. Actionable analytics, right? Being able to extract information from all that data, but more importantly, to be able to do something with it, to create new products, new services, create greater productivity in your operations. Finally, the ability to monetize all this. You've created this end-to-end capability. How do you get paid for it, right?
We hear the term a lot, data is the new oil, right? The ability to not only, as the owner of that data, to be able to create revenue from it, but maybe even to give access to that data to another party who may be able to use that data to create a new capability and then pay you accordingly. Those are gonna be important as well. We're also very conscious that creating these kinds of end-to-end solutions, to be able to look at different data types, to be able to do analytics across a number of different systems.
Maybe it's the, in the retail space, my customer loyalty program and looking at my inventory levels and looking at real-time store sales information, requires a number of different partners that we need to be able to work with to help our customers really implement those and make it easy to deploy. Working with industry consortiums are going to be increasingly important as we go forward. Intel is one of the five founders of the Industrial Internet Consortium. You can see the founders, AT&T, Cisco, GE, IBM, and Intel.
We formed this consortium to be able to define what it takes to create these kinds of Internet of Things implementations, then to give direction to the existing standards bodies in the industries around what needs to change in order to support the evolution of the Internet of Things, and to then create work groups. These work groups will take this technology and actually go build proof of concepts to do pilots to demonstrate that these things can be delivered. The other one that we think is important, you saw Hermann mention this one as well, is the Open Interconnect Consortium. Again, Intel is one of the founders of this.
You can see Samsung, MediaTek, again, Cisco and Intel, forming this consortium to create a framework to enable interoperability amongst these different endpoint devices, not only in consumer applications, but across these other areas that I described as well. Much broader than consumer, we want it to span industrial, commercial types of applications. To be created in such a way that we can have a truly open ecosystem to be able to create new applications, new technologies at the pace that innovation can move, and to do so in a way that is unencumbered. When I started into this role, I'd probably been in the job a couple of months, and I was meeting with a reporter, and he was kind of pounding on the table saying, "Hey, this all sounds really great, Doug. Where's the economic benefit?
I want some real examples that are generating real value. Where are those? Please show me those." I thought I would show you a few examples. I'm gonna click through these fairly quickly. The first one, you've heard Kirk mention it, you heard a few references, that we use vPro technology in the retail sector to be able to provide manageability and security in endpoint devices in the retail environment, point of sale, digital signage, kiosks, ATMs, those types of applications. We worked with NCR on a technology that we developed, a software technology called the Intel Data Protection Technology. And this technology works on financial transactions. We announced this October 15th at, in London at an IoT day event that we conducted there.
What this enables us to do is to be able to encrypt the transactions that occur in the retail environment from the card swipe device all the way back to the server. What's important about this is not only the encryption, but the fact that that data no longer resides in the point-of-sale system itself. It provides that secure level of transaction, and we see the challenges that the industry faces today, and we believe that this type of an end-to-end implementation can help alleviate some of those challenges. The second one actually comes out of our own factories. We worked together with Mitsubishi to create a solution where we put a gateway device into tools in our assembly test factories.
By the way, if you go into our factories, our factories are already pretty amazing Internet of Things implementations in a factory environment. This gateway is giving us access to even more sensor data than we had previously, and the engineers are now using that data to be able to further optimize how those tools perform, to be able to further optimize yields. They've been able to pretty significantly reduce yield loss that's just caused by variations in the equipment. I had the opportunity to be in Penang last Friday and met with the team and saw the results that they're delivering, and you can see it's delivering about $9 million of savings per year, both in terms of improvement in yields as well as improvement in equipment uptime.
The next one is a customer that we've been working with called Vnomics. Vnomics builds a gateway device that goes in 18-wheel trucks that are running up and down the highways, and they built this device for a company called Saia Trucking. Saia's a large trucking company, and they've implemented these in their trucks, and I thought it would be good maybe to have them tell you the story about these devices. If we can run that video.
A company called Vnomics is providing real-time individualized driver coaching, comprehensive vehicle analysis, and safety and compliance tools through the Internet of Things.
[Presentation]
When I talked to Saia's COO, he told me that they're saving about 6% on their fuel costs on an annual basis, that translates to $15 million in fuel savings. I was conscious of my audience and went and looked up their earnings report for 2013. 2012 to 2013, they improved their net income by about $11.5 million. They said it was because of focus on efficiency and implementation of technologies. I don't know if there's a direct correlation there, it seemed to hold up, it's good. The last one is a company called Rudin that we've been working with. Rudin is not necessarily a household name. They're a company that's been in existence for 100 years.
Some of you from New York may recognize the Samuel Rudin Trophy as handed out for the New York City Marathon, which was run just earlier this month, that's the association with Rudin. We worked together with Rudin and with Cisco on a product that Cisco has called EnergyWise to be able to attach to a large building that Rudin manages in New York City. They manage lots of buildings in many cities, and they use this technology to really instrument one building, that one building, in the last year, they've saved $1 million in energy costs. That includes, you know, living through one of the coldest winters in New York history. I thought it'd be kinda cool to show this to you. They gave me access to their system.
They call it DBOS, and they describe it as an operating system for buildings. This is live data if we can pull it up on screen. This picture is actually their energy usage from a little over a year ago, before they implemented this system. There are some important things to realize is they would start to heat up or cool the building about 6:30 in the morning, and they would keep the building at a pretty constant temperature throughout the day and begin to ramp down around 6:00, 7:00 in the evening. That was their typical profile. They implemented this system, and they started looking at other data sources. They connected to data at the turnstiles in the building, and they started looking at building occupancy.
Interestingly enough, 8 A.M., people start, you know, coming into the building. By, you know what, 10 o'clock, the building full. Let's give them credit. Let's assume there are customers coming into the building, so they come in a little bit later in the day. There's this lull in the middle of the day, people going to lunch. There's about a two-hour window there where the population of the building starts to fall off. Then later in the afternoon, you can see people start leaving the customers start leaving the building, then later, the employees start leaving the building. What they did was they took this data, and they combined it with their energy consumption. By the way, you see this blue line isn't complete 'cause this is live data in that building right now.
What they started discovering was that they could begin to modulate how they managed energy consumption in the building based on the occupants that were in the building. You can see now that they can ramp up, but they begin to ramp down much earlier in the day, and they can ramp down in the middle of the day when there are fewer people in the building. Important capability, but what was even more fascinating, I thought, as we were talking to them was they said this double hump camel has now given them an opportunity to implement battery storage capability in the building.
You couldn't do it when you had that kinda constant curve. Now they can start to implement just the right amount of cost-effective battery backup to be able to clip the peak demand usage, and that's one of the ways in which they get a lot of the efficiency. The other interesting things is these algorithms continue to improve as they're doing the big data analytics back in the data center. This is a profile of a week. You can see that their predictions adjust based on the week. Now, interestingly enough, it appears that on Friday, they must have a lot of people who work from home. Just saying. "They're out visiting customers," Brian says. I agree. There you go. I wanna wrap.
If, if you look at end-to-end, the capabilities that Intel has to deliver on these Internet of Things implementations, I contend we have some of the strongest assets of anybody in the industry. Processors for things. Gateways, both for the home and industrial, I described those earlier. The ability to connect. Herman talked about the assets that we have from a connectivity standpoint, and we need to connect Internet of Things applications. Diane talked about the transformation in network infrastructure and the cost that we saw, down 40X in 10 years, as SDN and NFV begin to be implemented more and more, it gives us more capabilities there. Of course, rich assets in the data center to be able to aggregate, store, and analyze data. The ability to deliver security solutions and the ability to deliver new services to extract value.
Hopefully you can see that we have a big opportunity that we're building off of. We're leading with solutions and technologies that we can bring into these other market sectors, and we're well-positioned to deliver on end-to-end Internet of Things implementations. Thanks, and I'll hand it back to Renee.
Well, don't go far away. All right. Last is not least in this case. This is an equally exciting opportunity for us. Rob's gonna come up and talk about memory and storage. This is an area that really is a platform play. Rob's gonna talk about how important memory is to the platform and throughput and other things. As you heard about the data center from Diane, and you heard from Brian and Stacy how big a part of the growth of the future it's gonna be, as you know, right behind communications has to be memory and storage as a technology that's critical to the future evolution for throughput, for performance, for the health of the platform. With that, I'm gonna ask Rob, and then when Rob's done, we'll have time for Q&A. Thanks, Rob.
Thank you, Renee.
Yeah.
Thank you. Happy to have just and actually I'm excited to have a few minutes to talk to you about our memory strategy. I would hope that you would take away a few things from the conversation that we have here. One is that although we've been in memory for a very long time as a company, the more recent sustained innovation and focus has delivered profitable growth for Intel and cash contribution back to the company. That when you look at memory and storage and you include it more aggressively into the platform, as Renee was indicating, you get a bigger canvas and a bigger palette of colors on which to paint solutions to solve IT problems, both at the computing, fundamental computing level and bigger problems for the IT industry.
Then finally, our computing insight into how computers work, the advancement of computer architectures, the use of computers in IT industry, along with the ability to optimize the platform at the CPU and memory controller and storage interface level, and the ability to innovate in memory technology, gives us a unique advantage, both in terms of advantaging the platform and as a storage technology vendor. Underneath this, our strategy is effectively one of Moore's Law. We're gonna be technology driven. We're gonna take advantage of Moore's Law. We will advance semiconductor process technologies and innovate there. We'll be customer inspired. We're gonna use that insightful information about how computers are used and the advancements of computers and computer architecture to build better products.
Finally, we'll be platform connected, in that we will use that understanding and the ability to optimize at the platform level to make Diane's servers better, to make Kirk's client platforms better, and to make our memory storage products better. I thought I would start on the lighter side. Memory and storage have always meant to be together. Data and the, sorry, CPU and data have always been meant to be together, and like a kid from the other side of the tracks, economics has kept us apart. I put on my tie. That's my avatar there. I put on my tie and we're coming across. You know, seriously, what did CPUs do?
They operate on data, and you want that data to be as close to the CPU as possible. You're doing comparisons, you're doing calculations, you're doing searches on that data. Any time there's delay in access to that data, it's a problem. In a perfect world, all the data the CPU would ever be working on would be right there in a register immediately available to the CPU. Those of us in storage are really all about getting that data closer to the CPU from a time standpoint, getting a lower delay to the CPU and getting more data to the CPU closer. You can see that in a more scientific curve than my avatar there about the advancements of technology.
You can see the impact of Moore's Law versus what was happening in mechanical storage. As Moore's Law advances, particularly, as we get to multi-core CPUs, the performance of our processors is going up very, very fast. The mechanical media, although it's gotten higher capacity, more bits per disk, has actually not been improving in performance, and that creates tension. That creates a gap in computer architecture, and filling that gap becomes critically important. We've known this for a long time. We've known that storage, that semiconductor storage, is more reliable, it's much, much faster, thousands of times faster, than mechanical storage, and that it will scale well with Moore's Law and we've been working on it for a long time.
That solid-state drive there is from more than two decades ago. It was our first solid-state drive in the marketplace. You know, like the kid from the other side of the tracks, it was not economically viable in computers. It didn't have high enough density, didn't hit the right cost, performance, density curve. We've kept at it and we've looked at multiple different technologies, and today our solid-state drives are nearly 200,000 times more dense than that initial solid-state drive. That is the magic of Moore's Law compounded over multiple generations. In the sort of more recent view of technology advancement, we broke through at the same time we were struggling with some consumer sales into non-volatile memory.
We had a breakthrough in technology around using this for solid-state drives, we've been evolving that technology. We've started to hit the right cost point and the right density point, we've evolved that technology very aggressively over the last six to eight years and, you know, began with an investment in leveraging our deep understanding of semiconductor process technology together with Micron. Delivering pitch doubling before anyone else in the industry led us to be able to leapfrog the industry and get to 34 NM before anyone else. The rest of the industry later came along with us in pitch doubling. We did wireline air gap isolation for bit cells. Bill talked about that earlier in terms of logic.
But in memory, we had to do that earlier because we were seeing interference between cells, and it enabled us to get to the next click on Moore's Law in memory. The rest of the industry came along later, and that had enabled us to get to 25 NM before anyone else. Then finally, you guys are familiar with the High-K metal gate technology that's happened at Intel. We used High-K metal gate technology in our last generation of planar cell technology, and that enabled us to get to planar, to 20 NM process technology before anyone else in the industry in a way that is very, very useful for computing. It wasn't just for consumer electronics, it wasn't just for low-cost storage. It was in a performance way with the right level of reliability for solid-state storage.
In fact, that year we won Process Technology of the Year. Now we're in the middle of an important transition in solid state. We're in the technology. It's about 3D NAND. 3D NAND is really all about getting to low cost and getting to higher density. Doing 3D NAND just as a direct replacement of 2D NAND is really not that interesting. What we really want to do is we want to expand the TAM, get to a bigger market, with memory technology, with 3D. I'm gonna talk a little bit more about that later. I want to go back to my story around fixing the gap in process in computing. We want to make Diane's servers faster, we want to make them scale faster. What we did was we created solid-state drives that fill that gap.
They are the thing that fixes that big latency problem and enables the processor to scale more effectively. In 2008, we brought what I would call the first modern-day solid-state drive that hit the right cost performance trigger in the industry. It was at the right level of reliability, and it began a whole industry. In 2008, when we launched that, the SSD business was a zero billion dollar industry, effectively. It was a couple hundred thousand dollars. Since then, in the last six years, that industry has gone from zero to $13 billion in six years based on that gap that we were talking about before. It was a computer architecture problem that we understood and were able to fill the gap, and it enables servers to scale. That's the so what.
It's cool technology, but so what? It created a whole industry. Why is that? Why, you know, you can see the computer architecture problem, but it delivers real results for our customers. It solves a real IT business problem. Diane talked about the importance of big data and scaling big data, and when you do that with hard disks, you get a certain level of performance. Really what we want to do is we want to do real-time analytics. We want to get time to that business solution, that answer much, much more quickly. When you implement that with solid-state drives, you see tremendous performance speed up. You don't just see 10% faster or 30% faster or 50% faster or 100% faster. You see hundreds of percent faster.
3x in this particular example of the modern day business analytics with Hadoop. The reason for that is that they're much, much faster. At the physics level, we know semiconductors are much, much faster than mechanical drives. We demoed at IDF a couple months back, a new level of IO performance. When you look at storage, you look at it in terms of the capacity, how many bits can you fit in a form factor, but also, how fast is it? You take something like this. This is sort of a modern-day storage device that looks like a hard disk, right?
Today, because we're a disruptive technology that we're bringing into an existing market, we build them in so that they physically fit in the same hole in the box as the predecessor technology and they connect to them. If you take a look at that and you want to get the same level of performance, if you want to get super high performance and get that analytics done, with just 28 solid-state drives, so it's about 50 TB, 60 TB of data, you can fit that in 4 in of space, right? You can perform these very fast analytics. If you want to do that same thing with hard disks, this rack would need to be 500 ft tall in order to get that level of performance.
If you want high performance, you would move to a solid-state drive. Breakthrough, this isn't just a little bit faster, it's not just a little bit denser. It's way faster, and it's game-changing. Like Diane was talking about, it changed the way people use computing, it's much more efficient, and it drives more use of computing at the end of the day. I want to also talk about this canvas, this idea that we can address IT pain points. We always knew that when we brought solid-state drives into computing, that it would cause them to be much, much faster, that it would cause them to scale much more effectively.
What we didn't realize as completely is that when we went more aggressively to embrace that as part of our solution space, that we'd see that much bigger of a canvas on which to paint solutions and that much bigger of a color palette in which to paint those solutions. What we found was, we had a new way to address scaling, bringing caching into servers. We've got a lot of activity going on in that area. I wanted to use just one as an example for you, and it's a security one in vPro. I noticed that a number of people use that today. When we look at security in a client environment, we use a hospital as an example. We've got laws in the country around protection of personal information.
We've got hospital workers that are trying to be productive with new devices. They're using personal information. On the other hand, we need policies that enforce the protection of that data. We want to make it easy and highly secure. We can make solid-state drives with encryption in them that encrypts the data and then we can have passwords on them. In isolation, that's a good solution. When we engage the vPro team and we create a vPro-based SSD, we have additional information available to us. We know where the drive is, we know where the platform is, and we can change the level of authentication we need. We can make it easier for the nurse or the doctor to get access to the data when they're in a safe zone.
When the notebook or the tablet or the phone moves out of that safe zone, we then have stronger authentication and better restriction and access to less data available to them. We call that geo-fencing, so that the platform behaves differently, and the access to your data is different based on where you are. We put that in our Pro SSDs, and they make the vPro platform better, more secure, and they create differentiation for our solid-state drives. What, so what? The implications of that platform technology that the integration into the platform overall and the technology-driven new levels of capability results in market leadership. In one of our focus markets is the data center, of course. We're trying to improve that. We are seeing we have had tremendous success there.
We've had 30% plus market segment share directly. We have a partnership with HGST, who's now part of Western Digital, for the SAS market. We have 2x, more than 2x, the market segment share of our next nearest competitor. Even if you were to stack a SanDisk and Fusion-io have become one, more recently, even if you were to stack them together, we would have more than 2x the market segment share. The business results of that technology-driven platform connected strategy have generated market segment share. Now I wanna go back and talk about technology again. I said I would talk more about 3D technology. Today in 2D, and actually some of the new 3D technologies, people are putting 128 Gb of storage on a given die.
The reason we wanna do 3D NAND is because we want to get higher capacity, we wanna get lower costs. In order to do that, we need to get more bits per die. With our 3D NAND, we're putting twice the number of bits on a die that compared to anyone else out there in the industry, why do we wanna do that? We wanna do that because we wanna get breakthrough levels of cost. How do we do that? What we do is 3D NAND is what it sounds like. We store cubes of transistors instead of a checkerboard of transistors.
In traditional 2D NAND, what you do is you lay out a checkerboard of transistors, you have 128 billion of them in a memory array on a device. That's a lot. That's a lot of transistors. In 3D NAND, you're trying to advance Moore's Law, but you're doing it in a more creative way. What you do is you put the checkerboard a little further apart, right? You relax how close they are together, but then you go vertical with them into a cube of transistors. In this particular example, we have four billion pillars or holes that we're etching down into that die, into a stack of materials that are 32 layers deep.
What we do is we create a, you know, pinhole, if you will, right, 4 billion of them, that have 32 layers stacked in them. You can see here is a cross-section of that. That aspect ratio of that, think of the tallest skyscraper you might know, we'll use the Taipei 101 example. You have to stack multiple Taipei 101s on top of each other to get that kind of aspect ratio, at each floor if you will, in that tower, you have a memory cell. At every layer in there, you end up with a memory cell. We have on one die cubes of transistors where we end up with 256 billion bits stored there. I thought I would come up with a three-dimensional example. That's a lot, right?
$256 billion, however you wanna look at it. At MLC and actually at TLC, we're at $384 billion. There are about 300 billion stars in the Milky Way galaxy, right? That's your perspective on size in three dimensions. What? I mean, it's cool. It's really interesting technology, but what does that mean? What's the benefit of that to computing? The benefit to that is we can put about a terabyte of storage in a form factor that's about 2 mm thick. See if you can see that. You see how thin that is? You get about a terabyte of storage in that. What?
You know, that's actually kind of cool and technologically interesting, but what that enables us to do is put no compromise storage that's high performance, very reliable, and high density into all of these devices over here. We're no longer having to make a compromise on how much storage we can put in there, and it's still very reliable, it's still very fast. When we talk about market growth in a minute, I think it's even interesting with what you do is you can put good enough storage, so something less than a terabyte, at a much, much lower cost. That enables us to expand the market for this type of thing much, much more broadly. There's no reason to not put in Kirk's best 2-in-1s, not to make them super thin, because you can get as much storage as you want.
You also, if we can get it cost effective, then we get it at the lowest price points as well. For Diane's business, what we wanna do is we wanna get as much storage as close to the CPU as we possibly can. What we're able to do in these kind of form factors that slide into Diane's racks, sideway here, we can get over 10 TB of storage into this. We plan on building, you know, 10-plus terabyte storage devices within the next couple of years based on this. It's a breakthrough. It is an impact to how we make semiconductor process technologies, and it's a significant impact on how we make computers as well. I did wanna say one thing.
BK's into using technology, you know, the early versions of the technology in his day-to-day life. We did have an early 3D NAND SSD, and actually so we're using it to present this presentation. I got a question from a number of you guys about where are you in the technology health. We're planning on launching this in the second half of the next year. This presentation here is being done from a 3D NAND SSD, and so far so good on that. Hey, I did want to say this is a disruptive technology. We're really at the early stages of solid-state drives. Although it's gone from 0 to $13 billion, we're really in the early stages of technology adoption.
We're in this traditional technology adoption life cycle where we go from working with the innovators to early adopters to the majority, and as we do that, we see tremendous volume growth along with it. Today, although we're at $13 billion as an industry and had, you know, $2 billion inside of Intel, we're really at the very early stages of this. We're at about 20% market penetration, depending on which segment you're looking at. About 20 probably in servers, less than that in desktop, a little bit more than that in notebooks. Our growth, in terms of the solid-state drive business is not limited by overall market growth of either the desktop or the notebook or the server market. It's actually limited by the penetration.
Our goal, the reason we're worried about solutions and costs of 3D NAND is about driving that technology adoption curve, and we're about to get at the steep part of that technology adoption curve. In order to supply that, many of you guys know we have a big wafer factory up in Lehi, Utah. We have about 20,000 wafer starts a week up there. We have a special relationship with Micron, where we take that same technology, and we use their other factories. We also have a relationship with SK hynix and are using their technology someplace, 'cause we need a broad set of supply because this market is somewhat unpredictable, and we need to be able to expand quickly for our customers.
Finally, of course, if it makes sense, we can bring the technology inside and manufacture it inside of Intel. For you guys, knowing my audience, right? Oh, sorry. I do want to make one more thing. The current projection is that we'll be at 50% by 2018. That seems a bit conservative. The current market projections are that there'll be about 50% market penetrated. In the notebook market, they're probably thinking about 50%. In desktop, something less than. In server, they're still saying about 30%-35% penetrate. There's an enormous opportunity for sustained growth in this market for some time to come.
Now for this group, Stacy showed you a view of our P&L over the last six years. I want to show you underneath that, we've been transitioning the business from being a NAND business to being an SSD business. When you look at this business from a SSD market perspective, it looks much, much more interesting than as a pure NAND business. We see sustained growth opportunity because the market is under-penetrated. We see an opportunity for us to be differentiated and grow aggressively with that market. I'd like to leave you with our fundamental strategy is one that's technology-driven. We're gonna take advantage of Moore's Law. It's about scaling. That's a unique advantage we have.
It's about taking advantage of the unique insight that Diane and Kirk's teams bring to us on how computers work and how our customers are using them. Finally, we're gonna be platform connected. We're Intel, right? We're gonna make Diane and Kirk's platforms better, and our interaction with that around vPro or around servers is gonna make our products together, better together. There's tremendous growth opportunity, and with that, I'd love to take any questions you have, and we'll bring the rest of the gang up here.
Thanks, Rob. All right. I'd like to invite Renee and Doug and Hermann. We got Hermann? Great. Up here. We'll bring some chairs out. Then we'll do another Q&A session. Again, I'll ask you to ask just one question, if you would, and if you could hold off asking a second round of questions until we've had a chance to get around the room and hit anyone who hasn't had an opportunity to ask a first question. All right. Okay. What do we have here?
I'm gonna stand.
You're gonna stand?
Yeah.
Okay.
I don't feel like sitting. Let everybody get seated. I'm standing. All right. Wonderful. Gary. Okay, over here, please. Stacy.
Thanks. Stacy Rasgon, Sanford Bernstein. Had a question on the, I guess, for Hermann maybe on the mobile strategy.
Where's Stacy? Where are you? There. Okay.
Right over here.
Yeah.
I know you said you were cutting down your internal SoC development, and I know you said you were sort of putting that onto your partners to do the SoC development and the derivatives. I was just had a question. What is it about what you're doing with partners that makes them willing and able to take on the extra effort and the expense of doing their own sort of derivatives? It's not a service that you're providing anymore. Why do they wanna do that instead of buying products from a competitor that are fully baked where all of that is done, all of the certification is done, the performance of the radio and even the process of this one is probably better than anything you're able to deliver, at least with SoFIA, which is made on lagging edge.
What is it about that business model that's gonna help you drive volume in smartphones? I guess as a corollary to that, to drive economics in the business, I presume you need volume. To get volume, I presume you need smartphones. Just, I guess I'm unclear how the strategy to build volume in smartphones to drive economics in the business works given you're cutting your own internal investments.
There's different kind of partners. We have one group of partners which are partners in the sense that we build together with them products. They are, so to speak, our customers. They wouldn't go and modify a platform, or they wouldn't go and modify a chip. They buy the chips as they is, and hopefully we have built them smartly enough and with a lot of enough of foresight that it exactly hits what their market is expecting them to deliver. We have other partners with whom we engage on a much earlier stage, on the chip level stage, and those partners are not customers. These are partners in the same level of the market. They have their own customers.
They have their own, kind of market niches or market areas where they wanna go, which is complementary maybe to that what we have done. We are, for example, in the case of the 2 3G versions we have from SoFIA, they have different kind of feature set. We, with that, we can address different kind of market segments, and, can address more customers because they have other customers inroads that we may not have, and they have also engineers that go to customers, it's brings a much, much better, and larger workforce and a portfolio together.
Could you give me a more concrete example of that?
SoFIA 3G is one of those examples. SoFIA 3G was the platform that we did to be in that market of the single chip integrated parts. We built that as a dual core. We had a particular graphics in it, and so on and so forth. When we sat down with that partner, in that case Rockchip, we came to the conclusion there are other segments in the market where we might be beneficial to have more horsepowers on it, so have more four cores in it, have different memory interfaces or have a more powerful graphics in it because that market segment would need that. We built that together with that partner.
All right. Let's come back over to this side of the room. Gary, please.
I've got a question on 3D NAND. Hynix and Micron, I think all the 3D NAND guys have different technologies. How does your solution encompass the different 3D NAND technologies, and what do you mean by disruptive cost?
First of all, the Micron Technology is through a JV with us, right? The Micron and us are collaborating together very effectively, I think, on new technology development. If you look at what's out there in the market today, the initial 3D NAND devices are still 128 Gb devices. You know, our goal is not to simply have the same cost structure but a different manufacturing methodology, but to have a breakthrough in cost. In order to have a breakthrough in cost, we need more bits per die. We already have a 128 GB die on 2D NAND. We'll be delivering 256 Gb die in 3D NAND MLC and then 384 in TLC.
Generally, if you can build a higher density die, it means you're lower cost, right? If you could build a 256 gigabit die, you would.
Just to clarify, It's 3D NAND technology agnostic in your SSDs?
Yeah, good question. There's two things that are going on here. One is we're doing the material science and process technology development around a specific implementation of 3D NAND, right? That's what I was showing in those electron microscope pictures there. The early version and then what we were running up here was an early version of that 3D NAND technology that's 32 tiers deep, and has 256 Gb per cell. Now, in addition to that, on 2D and 3D, we have controllers that enable us to work with that technology. We also have a very fast-growing market, so we're gonna wanna be able to have the ability to get supply from multiple places.
We figure out how to make that controller also work with other people's technology.
Okay. Let's come back up to the top center here. Let's go maybe C.J.
Great. Thanks. C.J. Muse with Evercore ISI. I guess, Rob, follow up on the 3D side. There's clearly challenges there with the new technology. I'm curious if there are delays there, where you would look for increased bit supply, whether you would look for an increased planar shrink, whether you would add new capacity. Then if I could ask a quick question to Hermann Eul. Following on Stacy's question, could you provide just a big picture level of how you think about, you know, handset business from here? Is it really all about, you know, partnerships with Spreadtrum or is there something internal that I should be thinking about as well? Thank you.
I didn't quite catch the 3D NAND one. C.J., you wanna hit 3D NAND again?
Yeah, sorry about that. If challenges persist and you're not able to get the pilot line next year and ramp more meaningfully in 2016, where will you look to source your bits?
Yeah. The short answer is, we have 2 D NAND in our factories today. We have multiple options for that. You know, the shortest answer is, I just showed you an SSD working with 3D NAND in it. We're confident it's gonna work.
Great.
The other question, I think that's a quick one. We have customers, our partners have customers, so you can expect to see a mixture.
Maybe, Hermann, we can give a little bit. We have customers that I think we talked about, OEM customers who do phones today. We also have a lot of modem customers who build phones, which we have been talked about through the day. There's kind of a combination of three different ways that we would serve the phone market.
Great. Let's get to Betsy here in the middle.
Thanks. Thanks, Mark. Betsy Van Hees, Wedbush Securities. Thank you so much for putting on a great day, Mays. Really appreciate it. Just wanted to go back to 3D NAND versus planar NAND. Just curious as to why you guys aren't putting focus on instead of going to 3D NAND, 3 bit per cell in SSDs, and why can't you get the cost breakthrough there before making the huge capital investment into 3D NAND?
Yeah, the clear long-term strategy is 3D NAND, right? What you're doing is you're backing off the density, if you will, from an XY dimension and going vertical. It is, you know, the right long-term technology strategy, where you have multiple process generations that you can do off of that. TLC and the next, you know, level of compaction on 2D NAND is a band-aid for the short term. What we really wanna do is get on the right long-term technology strategy.
Let's come back over here.
Ed Snyder, Charter Equity Research. Hermann, it sounds like you've kind of got a bifurcated model here. I know there's three sets of customers, but you've got this SoC business that's moving along, and you're gonna push that off to a lot of your partners. Yet, at the same time, you're still developing these thin modems, 7260, 7360's coming out, and you've got the win with Samsung there. Is the investment in that gonna continue? That has been a very expensive project for everybody. Broadcom just exited, and we know the struggles that many other OEMs have faced. Ericsson is out. I mean, they're in the Samsung model, but they've already abandoned their efforts there. That seems to be where most of the R&D was spent at Infineon and Intel and subsequent companies.
Is that roadmap gonna continue thin modems, and you're gonna keep moving into the LTE Advanced? Or is there some shift going on that you're looking at attacking the low end of the market, and then maybe if you don't get success on the big picture smartphone guys, you back off there? And is there any chance that you're moving away from ARM so that some of this traction you're getting, especially in the thin modems, has some benefit for the factory because you're not fabbing that internally now? Thanks.
The thin modem technology or standalone modem technology is something that we have traditionally done. There was always a significant market for this, and we believe there is a significant market for that. That directs the leading edge of the devices. They need the highest data rate. They need the most advanced feature set. By itself, needless to say, that is a wonderful vehicle to develop the technology and to push the envelope on the capabilities. Why others have left that, I think that has nothing to do with this or that. That has just to do with those investments are really high. Not everybody on that planet can sustain this. If your products are not good enough, you have not enough customers, and then the business case doesn't work out anymore.
We believe we have that technology. You see we are shipping this, and we continue to push the envelope here forward. That is one part of the answer. The other part of the answer is how does that translate or how does that map onto the SoCs. Of course, this technology also goes into SoCs. Those SoCs want to have LTE technology as well, so we bring the 72 IP that is currently shipping into the SoFIA LTE that I have announced for next year. That goes into this as well. We can make that decision whether we launch this as a standalone modem or as an integrated version for every generation again and again, new.
There is no need to do first the slim modem, and then an integrated part, or do first an integrated part and then do a slim modem. It depends on what we have customers for and what we deem is the right thing to do.
Yeah, I just wanna add on. That's why I started at the very beginning with my comments about this being a platform-level investment. Leadership in comms is a platform-level strategy for us, the answer to the question is no, our strategy's not changing, I just wanna make sure, you know, we're super clear. Yes, we know people have exited, but I think that's an opportunity for us long term.
Great. Let's go over here.
Hi, two quick questions on NAND. The first is you mentioned the breakthrough cross-performance. Is that unique to Intel? Like, does Micron get to share that with you or? The second question is, you talked about bringing the manufacturing in-house, so would that be a separate fab for memory that Intel would build or how do you do that?
Yeah, the for the 1st question, the answer is we jointly develop this technology together, just like we have all the previous technologies, and Micron has the ability to bring that technology to market as well. As far as bringing manufacturing in-house, what I actually said was, if we have the ability to bring that technology in-house if it makes sense. Let's come back over here. Trey, looks like we have a mic.
It would be, if it were all, you know, if we wanted to bring in an Intel factor, we can. It depends on a lot of factors whether we would do that or not.
Thanks. Gus.
Gus here. Hermann, I'm thinking about your go-to-market strategy with Spreadtrum and Rockchip. What's to prevent you, the two partners from undercutting each other on price if you're both gonna sell the same parts and/or prevent Spreadtrum from, like, just taking your IP and the LTE modem, and then integrating with ARM and going off in their own direction?
Every project that we do with them is a project on which we mutually agree. That speaks to can they take that IP and go? No, they can't, because we own that IP, and they work in our environment. The other part of the question was, do they ship the same? That would give you insight into what kind of contract do we have with whom. As you could see from that what we have revealed today, we are working with Rockchip on the Sofia 3GR, and we are working with Spreadtrum on the Sofia LTE, there is no overlap. Even if there was one, it's always an open market, and people compete in an open market.
Let's come back over here. Good. All right. Up into the back. Thanks, Gary.
Thank you. Hi, it's Brett Simpson at Arete. Hermann, just a quick question on the modem roadmap. We've heard from some of your competitors like MediaTek that they feel having EVDO is quite important to address the China market. What's your strategy for Intel to address EVDO, and will you be launching six mode modems next year? Thank you.
Our strategy has been that we integrate the technologies that we thought have a real great market. We have integrated TD-SCDMA, that is a very, very large market in China. We see on the other hand that, for example, American operators are speaking about going LTE only, without any fallback solution. It's a pretty mixed bag on what you're looking here, and you have to pick your choices on what you invest.
Great. All right, I think we'll try to sneak one more in. Trey, do we have one? Great.
Thank you.
Right in the middle.
Rob, throughout today, we've heard a lot about Moore's Law and how important it is. I was just wondering, how do you justify the, or how do you look at NAND long term in that, you know, Gordon Moore exited DRAM.
Yeah
35 years ago because I don't know why, but he's a lot smarter than I am.
Me too, so well, go on.
Why is that a good use of capital? What's different about NAND versus DRAM long term?
The question we have to ask ourselves is, can we have a competitive advantage, and can we sustain that competitive advantage? Is there something unique about, you know, our position related to the other people in the industry? My view is that, one, we're very good at Moore's Law. We know how to advance Moore's Law in these kinds of technologies, and we've demonstrated that over the last six years. So that, you know, the methodologies, how we do process technology development, et cetera, is an advantage that is, that Intel owns. Secondarily, our unique insight into where computing is going for NAND in compute markets, not for NAND in, you know, in its generic purposes, but for the markets that Intel serves.
You know, and then finally, a unique insight into how platforms work and being able to optimize across platforms is an advantage for us as well. That if we can demonstrate and sustain that advantage, it's an excellent investment because it's adjacent to the other platform technology that we're working on and our customers, et cetera.
Great. All right. That wraps up the formal presentations for the day. Thank you to everyone for joining us here. Thank you to everyone on the web for joining us virtually. I'd like to invite all of you who are here in the room to join us for a reception. That's held in the same area where we did lunch earlier today, and the executive team will be there for informal networking. Thank you all.