Morning. Along with Andy and Mark, I would like to welcome you to 2013 Investor Day, but also just Intel's headquarters. This is the heart of the company. Before I get into my foils, though, in my talk this morning, I thought I would like to make two points to the group here this morning. First is the importance of Investor Day to me and to this organization. You, as the investor and the industry analysts, provide exactly what Andy just talked about, an outside-in point of view. You give us a perspective and a view that only you can provide. As you go through today, not only do I want you asking questions of us when you are in the break room or at lunch or in the reception at the end of the day, not only asking questions, but also giving us your point of view.
We want that outside-in point of view. We would like to hear from you how you see Intel, how you see the ecosystem evolving, what you see from a competitive landscape. Please, that is important to us, and that is of value to both sides today, in my opinion. Second, as Andy said, long history at Intel of shareholder return, valuing it, focusing on it. There is a fundamental underlying precept I want to make sure you understand. That shareholder return is based on a relentless pursuit of innovation. Not only pursuing that innovation, but then bringing that innovation to the market, and that is where the shareholder return comes from. As we go through today, I want you to really take a look. Are we driving that innovation? Are we driving that to the market? Do you see the shareholder return? That is how we view the world in this space.
With those two introductory remarks, I will get into my talk this morning. My job this morning is to give you an overview, an overview of the strategy, and an overview of what people are going to talk about today. Oh, wait, can we go back? I am sorry. I picked the title of my presentation on purpose. If it computes, it does it best with Intel. There is a reason for this title. It is all-inclusive. It means that from our Intel Xeon processors, through the Core family, through Intel Atom, through Quark, however the market moves, wherever the compute need is, we want our products to do it best. It is also inclusive of our foundry philosophy. If it is somebody else's compute, we want it to do best with Intel. In this case, it may be through our Intel silicon, but it is still computing best with Intel.
It is a statement that underlies everything we do. It gives us a chance to talk about it from one set of perspectives, whether it is the foundry or whether it is our own products and our own systems. That is why we title it this morning. That is something that we have been driving into the company since we took control. The second thing that I often get asked, and I thought, before I get into talking about the various segments of the business, is what is going to be different? What is different about Intel moving forward now that you have become the CEO? I thought rather than talk about it in generalities, we just go right to the topic. I put together this foil of you really cannot talk about what is going to change unless you also talk about what is not going to change.
We'll start on the what's not going to change moving forward. As Andy said, this fundamental belief that Moore's Law, more than anything, drives this corporation. For those who study Moore's Law, they understand that Moore's Law is actually a law of economics, and that you obtain scale through scaling. It's a simple way to look at it. That scale can be scale of volume, that scale can be scale of economic profit or economic cost reduction, but that scale is achieved through the relentless pursuit of scaling. Bill will talk to you about it this morning. Stacy will talk to you about it this morning from an economic standpoint. I'm talking about it. That's not going to change. We believe it continues to pay benefits. It continues to pay that fundamental foundation of what Moore's Law is, the economic value.
The second thing that won't change is that we will continue to develop products that make the best PCs. This continues to be a foundation of the company. It generates more than $12 billion of profit. It also allows us a platform from which to innovate from. It's a platform from which we can bring new technologies and new capabilities to other areas of computing, and it's a foundation from which to build up and down from. Those products will continue. As shareholders, that's a core of our business that generates quite a bit of revenue, and it's something you want us to continue to focus on. As Andy said, everybody should be running out buying a Haswell system, so you have all-day battery, and you don't worry about those plugs. What's going to be different?
The first thing is exactly what you've heard us talk about so far, fairly often this morning, is that we're going to become more and more market driven. We're going to have an outside-in view. We're going to be sensing what's going on in the ecosystem, what's going on in the industry, and getting ahead of that. Whether it's in the data center or whether it's in the far other end of the Internet of Things or in between with what's happening between the client PC and what would be considered the traditional PC and tablets, mobile products like the phones. We want an outside-in view. We want to be sensing. We'd become insular. We'd become focused on what was our best product versus where the market wanted to move. We'll embrace where the market moves. That's one of the big changes that's occurred. We've talked about it.
It was the title of my topic this morning. If it computes, it computes best with Intel, and that's from edge to edge. You've seen us introduce new products like Quark. You've seen us introduce products on the other end, like some of the new Xeon E5 and so on. This is from edge to edge methodology. You're going to hear us talk about the foundry, and you're going to hear us talk about it a little bit different, that we'll open the foundry to any company capable to utilize our leading-edge silicon. We're going to go much broader. It goes with this, if it computes, it computes best with Intel, which means if we can utilize our silicon to provide the best computing, we'll do that. We'll create platforms for enterprise, not just silicon. We have a great footprint in the data center.
I'm going to talk to you about it. Diane's going to talk to you about it. Stacy will talk to you about the economic benefits. We believe we have a great platform to build from. The last one is really drive a focus on bringing the innovation to market quickly. You will see a pragmatism in several of these presentations this morning that'll feel a little bit different from Intel. We'll try and call those out. We'll try and talk to you about them. I'll try and talk about some of them this morning. You'll see some in Hermann's, in Kirk's, and various other places. You'll see some unfold over the next 6 to 12 months as well. When you look back, look at it as they're being pragmatic, they're driving innovation to the market quickly, they're doing what it takes to get shareholder value.
With that's kind of a good synopsis, I think, of how we're trying to change what's going on inside of Intel. What I'm going to do now is give you kind of an overview of the strategy in each one of the segments. The segments I'll work through is first, the data center, secondly, the PC experience. As I said, PC experience, we'll talk a little bit about that. It's just not the PC. We really want to talk about how we're going to change the experience of the PC. We'll talk about mobility, then we'll have kind of a broader discussion about technology, because again, that's an underpinning of everything we do here. With that, let's go into the data center. First thing is, I want to reinforce that we're continuing to commit to this 15% compound annual growth rate through 2016.
We believe that when you look across the four segments, the enterprise, the cloud, the high-performance computing, telecom, you break it down below that into the Internet of Things, big data, the infrastructure. As infrastructure becomes the demand for computing becomes higher and higher in infrastructure and software-defined networking, software-defined storage, you'll find that as Diane talks, those are some of our fastest-growing areas in this data center. That gives us confidence that we can continue to grow this segment at 15%. We also believe that we're the only company who can provide this complete solution, it's something we can build on, that we can build from, to build above and below with things like the Internet of Things. You saw us create a new organization around the Internet of Things, building above with big data solutions and big data platforms.
We believe we're the only company who can provide all of that. What also gives us increasing confidence about that 15% is the diversity of the revenue. What I'm showing you here is the projected revenue as segmented by those segments I just talked about over time. What you see is the enterprise, which I think we all talked about, grew a little slower than we wanted this year. Diane will talk to you a little bit about what the drivers for that were. Even moving forward, as we continue to grow, more and more percentage of our business is from those segments that are both growing quickly, emerging, and in some cases, we have really nascent footprints, growing footprints. There's both growth in those segments and growth in our ability to gain share. You see a diverse revenue stream really coming out of this.
That gives us a confidence in that ability to grow at 15%. I'm not going to do Diane's presentation. What I want you to walk away with for the key messages in the data center and our strategy around that is it's a strong growth segment today. It's becoming increasingly diversified, and we're working to diversify even further with more of a platform solution and not just selling silicon, but selling complete solutions. We're forecasting that we'll continue this 15% growth through at least 2016. Okay. We've talked about the data center. Let's move to the next segment, which is the PC. Remember I said we're going to talk about the PC experience, and then Kirk's brought a bunch of examples on stage here, and he'll give you some good discussion about what's really meaningful about them.
What we're really talking about here is the PC experience. The PC experience changing, not just the PC. I think if you went back in the years, we talked about, oh, we're going to make PCs thinner and lighter, or we're going to make battery life longer. It was still about the PC. What you're going to hear today, and moving forward, is how we're going to change the experience. Before we get into that, I thought it was important that we first give you our view of the PC market, because the PC market is important to us. Our view of the PC market, and we believe, although the numbers change almost nightly from third parties, believe we're aligned with third parties. Our view now is that the PC market is beginning to see signs of stabilization.
We continue to forecast low single-digit declines, so it's still declining, but that decline has slowed down. What's driven the decline has mostly been that we're seeing enterprise business start to pick up and especially mature market, both business and consumer. You can start to see that today even as you go out into the various marketplaces in the U.S. Our view is that it's declining, but it's beginning to show signs of stabilization. What we need to watch is the emerging market. That is the place that we need to really continue to watch over as to what's going to happen with this market as we go through the year. Let's now shift from that back to that experience here around the PC.
The first thing we're going to talk about is this 2-in-1 with convertibles, detachables. The real difference between convertible and detachable is, and Kirk will show you many of these, a convertible tends to either flip around 180 degrees and allow you to go into a tablet mode just by flipping it around, or you'll see some kind of a swivel mechanism. There's a variety of those mechanisms. It's interesting to us, there seems to be some level setting now between demands of convertibles versus detachables and then the kinds of convertibles. The real focus that we have is not just which hinge or what hinge mechanism, it's how do we create an experience that truly gives the person, the user, the best of both worlds.
How do we give them an experience that says when you're in tablet mode for whatever you're doing, it truly is a great tablet experience? When you're in PC mode, the classic clamshell, how do we give you the best experience for what you're trying to do, the usage models you're trying to do in that mode? That's really what we're focusing on here. Some of this is making them thin and lighter. Some of this is making them fanless, and as we go through next year, you'll start to see fanless systems. Battery life, longer battery life. As we move to Broadwell, you'll see longer battery life than even at Haswell. But already at Haswell, you see great battery life, MacBook Air at 13 hours. It's pretty regular. You can go out and find a Haswell-based system that's eight, nine, 10-hour battery life.
It's really become an all-day battery life type of event, with the strong computing capability of both a PC and a tablet. The other thing is that people want choice. We see it. We are going to provide PCs with all OSs. We already provide systems with Mac. We already provide systems with Windows. You'll also see systems, especially already in some cases in the market, but as we go through the holidays and into next year, Android and Chrome. Kirk will talk to you about the number of Chrome systems, but we actually have more than 50% of the Chrome systems out there now. We believe that we need to become. This will be a theme across both PCs and tablets, is really provide operating system of choice.
What's unique about this is that we're also the only people who can do this in such a way that an OEM or an ODM can build one system, one set of hardware, and choose which operating system they load at the last minute, at shipment. They can even delay it and let it be loaded out in the field. As we move into Q1, we will provide hardware and firmware that allows an OEM or an ODM to make that choice at the last second or ship without the choice and let the field choose. That gives them a flexibility, a BOM cost reduction, an inventory reduction that nobody else can provide. Operating system of choice is critical. If you're going to provide the best of both worlds, you have to provide those options. The next thing is you've seen great reviews about Bay Trail.
You see us put more and more effort on Atom. Both Bay Trail and Haswell allow us to get into price points that you've never seen before for these kinds of devices. You'll see two-in-ones, convertibles, and detachables as low as $299. I think at holiday, you'll probably see something in the low 300s, 320, 340, something like that. As we head into Q1 and into the first half of next year, you'll see those price points drop below $300. Majority of those systems will be Bay Trail. You've seen the performance on Bay Trail. Kirk will talk to you some more about it. These are compelling systems. Two-in-ones, detachables that are providing the best of both worlds of a PC for really what becomes tablet-like prices.
For a PC, the key messages I want you to walk away with, and as you listen to Kirk, make sure you really understand and capture, is the market is stabilizing. We're really focusing on innovation here and form factors and capabilities. We'll talk to you about perceptual computing and some of the innovations we're going to drive in perceptual computing. The best experience on every OS. I'll talk to you a little bit more about that as we go into the mobility section next. Amazing affordable prices that provide this best of both worlds experience at very affordable prices. That's what I wanted you to capture on PC. I'd like to move the next section to mobility. For us, 2013 was a year of establishing a footprint. We were, as Andy said, nascent in tablets.
In fact, we'd let that market grow up around us without a strong presence. We had seen the netbook, and we had actually created, one might say, the leader into the tablet world with the netbook, and we stopped innovating in this space, and we let it go. 2013 was about establishing a footprint, and that footprint included getting the OEMs and ODMs aligned so that they would build tablets with Intel, getting our products aligned so that our products and our own BOMs were aligned so that we knew how to build tablets. Just basically getting the ecosystem understanding that we were going to be in the tablet market for good. I'll be honest, at the beginning, as we tried to go in and enter it, a lot of the response was, "Well, wait, this thing grew up without you guys here.
We didn't think you wanted to be a part of this. Are you here to stay?" We had to spend some time talking to them, convincing them, going out and visiting them and saying, "Yes, we're here to stay." In some cases, we made long-term agreements with OEMs and ODMs to convince them that, yes, we're here to stay. This is a market we're going to play in, and this is a market we're going to lead in. That's what '13 was, establishing that footprint. '14 is going to be a year of growth. We're going to invest to scale. Our goal for '14 is to more than quadruple our volume. That means more than 40 million units or tablets sold.
We're still, remember, taking parts that were pretty much built for those two-in-ones and some of those other PC-like devices and bringing them down into this tablet environment. There is still, and Stacy will talk to you about this, some BOM disadvantages we have at the very low end, because in order to quadruple our volume, we've got to go from top to bottom with tablets. We'll talk to you about the investments we're making there. The result of this, the result of targeting to quadruple our volume is that you'll see tablets with Intel Inside as early as this holiday at price points below $100. We will span from the upper end, let's call it $350, $400, down to sub-$100. Sub-$100 systems will tend to be kind of spot market individual products, but you're going to see a lot of systems, $99, $100, $129, $149.
To make a proof point, I gave every one of our board members $149 tablets at one of the board meetings a few weeks ago, just to show we're in, we're here, and these products are good. If we just do that, if we just go into the tablet market and become another player in the tablet market, that's not enough. You're not really using that leadership, if it computes, it computes best with Intel. You are because you're driving the cost, and we can do that as we design our products more and more, and I'll talk to you a little bit about that in a second, about what we're doing with our product roadmap and our SOCs. We'll use Moore's Law to drive these lower price points at very profitable stages. That's not enough.
What we really want to do, I talked about this a bit, is take a lot of the innovation and differentiation that we are able to do up on the PC and bring that down into tablets. We believe we can do that cost effectively, we believe we can do that across all of the OSs, talk to you about, we are able to go span all OSs, and we believe we can do this uniquely because we have that PC experience. We'll be able to do performance segmentation. I'll show you some products that we're going to talk about today that segment from the low end feature phone and entry tablet up through performance tablets with high-end Bay Trail and Haswell even on some tablets. We believe we can scale Android. We'll talk a little bit later on about what that really means.
What it means at a high level is that we'll bring 64-bit to Android. We'll scale it so that we can bring it into the enterprise. Multi-windows. We have the ability to do that nobody else can. We have that experience. Our cores, Bay Trail, already is 64-bit capable. We've been doing 64-bit on PCs for years. You'll see it in systems in next year, starting on Windows, probably, moving to Android. We'll bring perceptual computing. We've talked about it. Kirk's going to talk about, a lot about perceptual computing, I think, during his discussion. We can use the same innovation, we can use the same hardware, bring it down into the tablet. In the tablet, it really provides another set of features and usages that can create a great experience for users. That's another thing we'll use to differentiate.
Multiple OS. Again, somebody can build a single platform and decide to ship it with any OS the last minute. The last one is around security. Ren é e's going to talk to you a bit about our security strategy. At a high level, it's about making security ubiquitous. It's about driving security across all of our platforms, all of our silicon, across all kinds of products. We have a unique ability because we have the assets, both within a silicon, but also things like McAfee, to provide security solutions in these devices that nobody else can. We'll innovate and differentiate in those ways, not just cost, not just driving down the price points. We'll do that as well. I want to move from the tablets to the mobility or phones. You've seen we are shipping our 7160, our LTE product.
We continue to have a strong commitment through our acquisition of the Infineon to continue to drive our comms capability. 7160 is both data and VoLTE. It's shipping today in systems. Hermann will talk to you about the 7260, which will come out next year, which will provide advanced LTE and data aggregation or carrier aggregation. Those products, we believe, and Hermann will talk to you about it, are world-class. Great performance, great capability. Those are fundamental. They're fundamental to phones, they're also fundamental to tablets. More and more tablets are shipping with either 3G or LTE. This is another place where we can scale. We can use these products across multiple form factors. One of the things that is changing is our focus on phones.
We're moving from being broad, building our own form factors, targeting carriers and multiple sites, to really focusing down on the big players. Working with them to develop silicon, develop capabilities, and really focusing on the large players in this portion of the industry. You'll see our focus, our spending, everything narrowed down and focused towards these key suppliers. I told you that we were going to continue to drive our Atom roadmap. I also told you that we were going to be more pragmatic to drive these solutions into the market quicker. This is one of the places that I want to show you is very different. This is one of the differences. The first one is a product we call Broxton. Broxton is targeted towards the performance segment of phones and tablets. Think of it as the next generation Atom.
What's different about this, and what we've done to accelerate this, is that this is a core that has a complete new, what we call chassis, or basically connectivity that allows us to do iterations and derivatives of this core at a very fast clip. Think of it as a core with a common chassis that allows connection of IP, both internal and external IP, at a very fast rate. The kind of rates that you'd see with external products. The kind of rates you see our competitors do this kind of capability. You've said, "Why can't Intel?" This product is a product that brings that capability really to reality. It's targeted towards the high-end. It's targeted towards mid 2015. The second one is one we call SoFIA, and this is a product that's very pragmatic.
What we did is we looked across our assets and said, "Okay, we've got Broxton. We know how to do this. We know how to build these cores that are more synthesizable and you're able to add IP," but this is a performance segment. At that level, there's still not the level of integration we were looking for. How do we go down towards the entry level, the entry-level phones, the entry-level smartphones, the entry-level tablets, and build something very quickly that is highly integrated? When we did that, we looked across and we said, "Well, with that Infineon asset, they actually had some very good products. Been sitting out there with 3G already integrated. We can integrate in LTE fairly quickly, and they're Arm-based." They're running it outside foundries.
We said, "Okay, step one, take the Arm core, put in an IA core." That's the first instantiation of SoFIA. At the end of 2014, our plan is to introduce a SoFIA that will be foundried on the outside, that is basically one of the original TSMC integrated products, and we've replaced the Arm core with an IA core, basically an Atom core. The first one will be 3G, the next one will be LTE. Step two will be to really get the advantage because remember, we're pragmatic, but we're also driven to if it computes, it computes best with Intel. The next thing, what do you want to do? You want to move that inside. As we exit 2015, you'll see SoFIA move from the outside foundry to inside at 14 nanometers.
You have an entry-level part, fully integrated 3G, LTE with IA on 14 nanometers, something nobody else in the world will be able to compete against. That's a pragmatic approach to get products quickly to the market and use your assets wisely, return shareholder value. That was an example of this pragmatism and this quick moving that we wanted to talk to you about. Three months ago, this wasn't on the roadmap. This was a brainchild of a person in this group that came up with this and said, "If we want to think creatively, if we want to think outside the box, I have an idea." We put this on the roadmap. We have to execute, which is another one of my agendas. The messages I want you to get from mobility.
First, we're investing to make a 4x footprint, north of 40 million units in tablets. We've got to have that footprint. We've got to build that scale. It allows us to introduce these new products. It allows us to introduce these innovations. As we exit next year, you'll see these innovations start to enter the market in big way. We're going to use the PC innovations and move them down to accelerate the differentiation and innovation that happens in tablets. We're going to drive leadership in mobile SoCs and communications with two products that we've just identified with Broxton and SoFIA. We're accelerating Atom to be on par with Core, just from a general standpoint of how we drive our graphics roadmap and how we drive our comms and connectivity roadmaps. Atom is an equal partner to Core. Doesn't mean we slow Core down.
We still expect Core to be 2x performance, but we're not going to slow Atom down to get there. That's what I want you to walk away with from mobility, strong, aggressive roadmap that has pragmatic solutions. Last section, bringing innovation to the market, manufacturing. Bill's going to talk to you next. What he's going to talk to you about is our continued drive and relentless pursuit of Moore's Law. He's going to talk to you about 14-nanometer. He's going to talk to you about 10-nanometer. We will not take the foot off the pedal here. We believe, as I said, that we can provide that fundamental feature, the fundamental reason Moore's Law exists, to obtain scale through scaling to get that economic benefit.
I also told you what's going to change is that we're going to drive that and utilize that now to open the foundry service. You will see us focusing on a much broader set of customers. People who can utilize this. Again, if it computes, it computes best with Intel. If people can use our leading edge and build computing capabilities that are better than anybody else's, then those are good candidates for our foundry service. The next place I want to talk about bringing innovation to the market is this perceptual computing. Kirk will talk to you about it. I think Hermann talks to you a little bit about it. This is really a chance to differentiate ourselves. This is a place that you can use our computing capability. You can use the fact that we're multi OS. Everything that Intel is great about.
We've been working on this for a few years now. What our goal is to bring it to market in 2014. What you'll first see is probably voice. You'll see voice systems this holiday. I was at Best Buy the other night, you'll see a little sticker on there that says, "System has voice." You can already start talking to your PC. As we move through next year, you'll see gesture. You'll see the ability to do facial recognition for security. You'll see a variety of what I'll call making the PC experience three-dimensional, that you're bringing the PC out of that flat screen towards you in a variety of ways. The tablet as well will utilize this down into the tablet. The last way I want to talk to you about it is security.
With security, what we'll focus on is, as I said, using our assets. We have silicon, we have things like McAfee, we have our own design capabilities. We have the ability to really integrate security and make it ubiquitous. Renée is going to talk to you about this. We believe that we have our footprint in the data center, our footprint in the clients, our silicon, and our architecture that uniquely allows us to provide a security solution. The key is to pull it all together and to integrate it. Renée will talk to you about that, but that's what we're changing. We're driving that, pulling it together, and integrating it. The key messages I want you to walk away with, accelerating the manufacturing leadership. We're not going to take our foot off the pedal of Moore's Law.
We're going to not only just advance the perceptual computing technologies, but we're going to bring them to market in 2014. 2014 and 2015, you guys will look at the PC experience and the tablet experience, your whole computing experience, so to say, and say, "It's different." It's now three-dimensional. It comes out of the screen. I'm gesturing to it. It's seeing me. I'm talking to it. That's a three-dimensional world. It's no longer a flat screen. You'll see integrated security, Renée is going to talk to you a bit about that. It's a unique way we can differentiate. I want to close up with how I started, with two foils. First, what will not change and what we're going to do differently. Part of what you guys are here to do today is to grade me and this organization of are we doing what we said?
Are we following this set of rules and guidelines that we've laid out here? Are we saying what we're doing? Things that aren't changing, driving Moore's Law, driving the best products. Probably as, or more importantly is, are we changing? Are we changing fast enough? These are how we're driving the organization, and I think it's important that you take a look at this discussion today in reference to these guidelines. The last one I wanted to leave you with is just how I opened, the title of my presentation or discussion this morning. It's important that you understand it's inclusive. At 2:00 A.M. last night, I woke up, and I wrote on my speaker notes, "It's inclusive." Because it dawned on me that was the key thing for you guys to grab. If it computes, it does it best with Intel.
If you think about that as being all-inclusive, spanning everything that has to do with what we do, then you can start to make sense of our strategy for Quark. We're going down there. We're going to put computing capabilities that nobody else has at the very low end, at the edge of the internet. At the upper end, with our Xeon family, we're going to put computing capabilities that nobody else has with integration and customization that Diane will talk to you about. Lastly, it lets you understand the foundry strategy. If somebody can utilize our silicon and make computing better, if it computes, then we want it to compute best on Intel. We'll look for and open up and talk to customers that can really utilize that capability. Pretty simple sentence.
You have to include it's inclusive, it's all-inclusive, then it starts to make sense. It's what's driving everything we do here now. With that, I'm going to finish. I'm done. I'm going to introduce the one person in this room who is a bigger believer and a more knowledgeable person about Moore's Law than me. For sure. Actually, there's probably a couple of you guys out there. For sure, the one I know of, and I'd like to introduce Bill Holt, our senior vice president of technology and manufacturing, and a good friend of mine.
Good morning. I'm here to talk to you, of course, about Moore's Law, I will do that, and I will cover why we're so bullish on continuing this. Today, I also want to spend a little more time on how this affects all the products across our product portfolio. Those are my two main focus areas. If you look at the agenda that I want to cover today, I have three main focus areas. First, I want to refresh our minds on what's really fundamental in Moore's Law, and that's cost. That's the most important, but clearly providing additional capabilities for our products. Then one that is becoming increasingly important in today's market is providing a real reduction in power that enables us to use all those cheaper transistors.
The second area that I want to focus on is reducing cost in a capital-intensive environment. You're all very much aware of the capital that we're spending, and that certainly is a challenge. I want to show you that in spite of the fact that it's getting more expensive to build these wafers, we can in fact provide real cost reduction at the most fundamental level, and that's what drives our ability to continue investing in Moore's Law. Finally, I want to talk about how these benefits can be applied across our entire product line. Hopefully when you're done, I will have convinced you of those three points. Before I go into my agenda, let me take a brief aside and deal with the status of our 14-nanometer technology.
As you're probably all aware, at our last earnings call, Brian announced that we would be delaying the start of our 14-nanometer ramp by about three months. That wasn't a very pleasant activity for us. It certainly wasn't very pleasant for me and the people that develop these technologies. It was an honest assessment of where we were at. If you look at the graph in the upper left-hand corner, you can clearly see what the problem is. Here, what we've plotted is the yield, relatively similar metric for our 22-nanometer technology on the upper line in the yellow, and for our 14-nanometer technology at the bottom line, and what we've done is we've offset them by two years so that they are at the same point in time.
If we were actually executing to our two-year beat rate, then those lines should have been on top of each other, and clearly they're not. What you will also see is that we're making extremely good progress and speeding up our yield learning, and this has been by a focused effort because this is not something that we were not aware of for a considerable period of time, and you'll see that we are, in fact, on track to close this gap. If we're going to close the gap, and this will be closed by early in Q1 of next year, you might ask, well, why did you have to delay this? The fact is, we don't just develop technology, we also have to develop a product, and that product needs units.
The yield that was low for the last year from our projection really reduced the amount of availability of good units that our product people could use to validate their parts, test their parts, develop the software that would go with them. The effect was that our slow learning on yield, in fact, delayed the entire product development cycle. That's the bad news. It's also the good news because if you look at the rest of this chart, the chart below is a measure of how many of the critical parameters in our factory, our lead 14-nanometer factory, are actually achieving their three sigma control limits. This again is offset in time so that in this case, the two lines are, in fact, right on top of each other.
What that says is that the fundamental ability to build these wafers is in fact not where the problem is. The control that we have, even though we're building smaller geometries and providing additional scaling, we're actually very capable of doing that. If you look at the lower right, you'll see the status, again, offset two years of the reliability status of this technology. While each individual category that we characterize here may not have the same grading at this point in time, what you'll basically see is we're on track. We have the same kind of improvements needed. There are some areas that are not done, the ones that are not green. The fact is, if you look at the one that is high risk in our assessment on 14-nanometer, it's related to yield.
The good news is we have fundamentally a challenging yield that we're closing, but everything else about the technology is extremely encouraging. We expect to see this ramp next year and in fact, deliver really outstanding products. With that, let me get back to talking about my agenda. First, let me talk a little bit about the fundamentals of Moore's Law. We all know, as Brian already mentioned, Moore's Law is fundamentally about economics, and that is the upper left-hand corner. What you see is that we're projecting, and in this case, the projections are very sound. For 14-nanometers, we have the technology completed as far as projecting what the cost will be. We're on track to deliver traditional kinds of cost reductions in our technology.
This graph is a measure of cost per transistor, and what that does is it enables our products to, in fact, add considerable new features without having to increase the cost. This is the fundamental engine of Moore's Law. This is what has allowed the expansion in the electronics industry. We're on track. In fact, when we look out to 10-nanometers, and for 10-nanometers, we've closed the definition. The products inside of Intel are beginning to be designed for 10-nanometers, and we, in fact, have test chips running on 10-nanometers. It's not as though we're guessing what this is going to look like. We're projecting similar kinds of improvement in cost, even out to 10-nanometers. The next critical thing in today's environment of power-constrained products is on the upper right-hand corner.
This is reducing the active power that it takes to perform a given function. Without doing this, you can't really use those cheaper transistors because if you have a cell phone, for example, and you want the same battery life, which all of us clearly do or better, you can't go throwing more transistors in there unless they're actually using less power. While without the cost, the fundamental driver of Moore's Law, you can't afford it, without the power reduction, no one would want it. These two are the fundamental graphs that really demonstrate that we are able to continue to provide these benefits. On the bottom two graphs, we've updated them. In the past, when we showed this, we would have showed 22-nanometer as our best here. We've updated this to show the improvement that we're projecting going from 22-nanometer to our 14-nanometer.
In our case, these are both Tri-Gate devices. Others call them FinFETs. In spite of the fact that we're not changing fundamentally the architecture, there are still substantial improvements that can be extracted as you learn how to build these devices better. What you see, whether you choose to apply that as performance improvement on the one side or in power reduction on the other side, there is substantial benefit that's available from this advanced technology. Let's move on and talk about cost. I'm sure that you all will have questions for Stacy this afternoon on capital. That seems to be one of your favorite topics. I do have to admit that I'm probably the biggest cause of Stacy's challenge in that I'm the one that spends all that capital. I used to be able to say I spent the development and Brian spent the manufacturing.
I can't even use that out. I just spend a lot of capital. Let's take a step back. As Brian said, Moore's Law is about economics. If you look at one of the fundamental graphs that Gordon had in his paper, it's this one on the bottom, and this is really about density. This is showing that the number of components that you could make economically was increasing. Well, that's the inverse of saying the amount of area it takes to build a transistor is getting smaller. In the time that Gordon wrote this paper, wafers really didn't get much more expensive. The advances in equipment and other aspects of it said that wafers basically cost the same. Craig Barrett, one of our former CEOs, used to like to say that the cost of a square centimeter of silicon was constant.
At one point, he compared it to real estate in Tokyo, that's probably not all that relevant. It was basically constant. Today, that's not the case, back when Gordon was writing, what was driving Moore's Law was density, that's still the most important factor that we have to remember, that's really what I want to emphasize today. Let's take a look at what's happening in the area of density. On the right-hand side here, I've replicated the curve that we've shown before on how we plan to scale the area of a given chip as we move from technologies. On the right-hand side, I've showed what our foundry competitors have been very public about saying, that is that as they move to the next-generation technology, they are not planning to do any scaling.
They are planning on introducing the FinFET device that we introduced at 22-nanometer, that will provide them a substantial benefit in transistor performance, they're not doing any area scaling. If you don't do any area scaling, you can't project any cost reduction, in fact, there is in fact an increase because the extra complexity of improving the transistor will in fact make a small increase in cost. It is this focus and importance of scaling the area, which is what's enabling us to continue to deliver cost reduction in spite of a more expensive technology. You might ask, "Oh, yeah, that's wonderful." We have seen the graph on the left. People have said that. How do we know you're going to do what you say you're doing?
one of the benefits of being in the foundry business now is that we do get direct input from non-Intel customers. On the left-hand side here, you see a graph from one of our key customers in the foundry, and they basically illustrated this same point. On the bottom, you see a projection of what will happen when one of their competitors moves from TSMC 20-nanometer technology to what they're calling 16-nanometer technology. There is the addition of the FinFET device, but as you notice, that die is the same size. There is no area scaling. It will result in increase in cost. On the top is Altera's projection of what will happen when they migrate from that same external technology to our 14-nanometer technology, and as you can graphically see here, the die is smaller. Fundamentally, the area is reduced.
On the other side, they're also illustrating that moving to our FinFET transistor technology, our Tri-Gate, provides them a substantial opportunity to improve performance. They're not implying that all this comes from Intel. If you look at their chart, a lot of this comes from architectural, and they are in fact moving two generations in this chart. Even they attribute a substantial amount of that performance improvement to those transistors that are in fact going to be better. Let me try to take that and quantify that a little more. If I take those first two graphs and I now align them based on our assessment of the density that we'll be able to provide at 14-nanometers and what our competitors are indicating that they will provide based on what they're doing at 20, you can align these two graphs.
what you see is that as we introduce 14-nanometer technology, we will be providing a substantial reduction in area relative to what our competitors are able to do. In fact, as we move on to 10, where we're projecting even an increase in scaling, if we project what, again, what people have said they will be doing, that increase will even grow. You also notice from this graph that at 32 and 22, we did not have a lead in density. That's not any news to anybody. You probably have heard that quite a bit over the last few years. To a very large degree, that's intentional. Those products were optimized, those technologies were optimized for products that primarily were delivering performance. The interesting thing about density is it largely depends on the size of the metal wires that you put on your die.
while transistors get better as you make them smaller, that's one of the wonders of silicon technology, wires get worse, and they get worse in that they can carry less current, and they run slower. As you scale, you actually have a substantial challenge to deal with, and one way is to not scale the wires as much. The reality is that in many of our previous generation technologies, we didn't scale the wires as much as we could have because the products that we were building didn't really demand that. There is no doubt that density as a measure was not one of the parameters which we focused on in the past.
As you can see from this chart, we're making a substantial improvement, and that substantial improvement is going to provide us with a leadership position, not just in transistors, but also in density. Why am I so concerned about density? Well, the other side of cost is wafer cost, and this chart here shows that in fact, the cost of a wafer or the cost per square millimeter of silicon is in fact increasing. Now, it's increased for a long time, but if you look carefully from the 90 nanometer to the 22-nanometer generation, largely this was a constant rate of increase. The line is basically straight, and on this scale that says that we saw about the same kinds of increase in cost. What you're seeing in the last two data points is an acceleration in this increase in cost.
This is the problem of capital. This is the problem that everybody's talking about in terms of increased cost. This is largely driven because we are not seeing fundamental improvements in the equipment that we use to pattern lines. In the past, the equipment suppliers would provide us with a better tool. That tool would allow scaling. That tool might be a little more expensive, but not substantially, and for that, you would get a substantial reduction in area. Today, however, the challenge is that there isn't a substantial improvement. The way that you scale is you use more masks, and that implies more process steps and implies a substantial increase in the basic cost of the wafer.
You do have an advantage if you decide to scale and you go from one mask to two, you no longer get about a 0.7 reduction in the pitch, but you actually have the potential to get substantially greater reduction because you're basically printing it twice. The other part of transistor cost is, in fact, the amount of area per transistor that you require, and this is why density is such an important parameter going forward. In order to offset that increase in wafer cost, you have to scale. We've always scaled. You can see that the old data points look about the same, and those two graphs combined give you the cost per transistor curve. In spite of a substantial increase in the wafer cost, we're projecting that just like everybody else is.
We're projecting an improvement in scaling due to the intense cooperation that has gone on in terms of defining this technology, not just from the technology point of view, but also from the design point of view, understanding exactly what could provide the most benefit and how we can extract the most value out of that. When you combine those two, you can see that we are very confident that we will continue to provide this cost per transistor improvement. If you take that last graph and expand it, you can see that fundamentally, we don't build wafers. We build wafers, but what we really build are transistors. The cost parameter you have to be most concerned about is the cost of a given transistor, because that's the basic building block for all of our products. Let me just give you just an aside.
When we began looking at the increasing wafer cost, we had a somewhat substantial debate as to what should be that point for 14 nanometers. When confronted with a substantial increase in wafer cost and a historical trend of density scaling, we had considered whether we would have to begin to bend this curve. There was, in fact, a substantial opinion that it was now time to start bending the curve. We all talk about the end of Moore's Law not necessarily being a law, but being a bending of the curves. We stepped back and said, "Well, what does Moore's Law do for us?" It largely predicts the future. That's what Brian said. It doesn't really predict the future. It tells us where the future should be so that we can then go and execute to it.
We had this debate, and we decided that, well, why would we predict anything but the point on the line? That's what we set. We set that for a goal for the definition team, so we're going to define the technology and define the products. What we actually got was a point substantially better than what we had asked for, and that point is below the line. It's not an exaggeration to say that we're not just continuing, but we are, in fact, surprising to many of us, we are accelerating the rate of cost reduction that we're projecting going forward. You can say that 14 nanometer point is because I showed you that we were actually behind at 22 and at 32. All we're doing is catching up. To some degree, there would be some truth to that.
At 10, there's no question. We'll have demonstrated our leadership capability at 14, and that line is on the same slope, the same kind of improvement. Yesterday, I was in Oregon, and we were having our semiannual technology and product review meetings for two days. While I'm not putting it on this graph, and I wouldn't, we did talk about seven, and they talked about the goals that they're going to be taking, and those goals are to continue this trend. We are really serious about this. It's not to say it's easy. Obviously, our 14 nanometer performance demonstrates this is very hard, but we believe it's extremely possible, and we are extremely committed to continuing to deliver this kind of cost reduction. Let me move on to my last topic then and talk about applying this benefit across the portfolio of products.
You've also seen this chart, and there's a number of points I want to make. First, I have made the 14 nanometer line solid. That's not saying a lot. This is a picture. It is, in fact, saying that we're not projecting anymore, okay? We know where this is going to be, and we can, with confidence, put that line on the chart and say we are going to deliver that kind of performance improvement. It is our second generation of building our Tri-Gate transistor, and that allows us to apply a lot of learnings that we gained from the first time around, and those learnings enable us to provide a substantial improvement in performance. The other thing I want to point out here is that once you have that fundamental improvement in the transistor, there are a lot of different ways that you can apply it.
I'm not implying that all servers are only concerned about performance, but they have a much bigger benefit if they provide true performance improvement, even if they don't see a substantial power reduction. However, as you move into the mobile space, a significantly larger portion of that transistor improvement has to be applied to power reduction in order to allow those extra transistors to exist inside the same battery life envelope. At a very high level, a basic improvement in transistors enables different products to apply that differently. If I take and actually plot the horizontal improvement, which is the transistor improvement or the reduction in delay or the performance improvement of the product versus generation, I also plot the vertical as separate charts, you get a slightly different view of what's going on.
What you see here is If I look at performance improvement provided by the technology per generation, you see that servers in general will choose to take more of that and apply it to performance. In the mobile space, more of that will be applied to power reduction. Now, if I'm at least self-consistent and I apply that same optimization point to power, you'll see that the servers see less improvement in power reduction, because in fact, they applied more of it to providing increased performance. Again, if you combine those two, you get a graph that is basically back to the fundamental performance per watt improvement. All of the segments need this improvement in basic power, and it turns out that they're almost taking exactly the same advantage. That shouldn't be any real surprise because in fact, that's what moves the line.
That line is fundamentally a movement of the performance per watt that you can get from a given transistor technology. The same fundamental improvement in the transistors or the technology is what is providing a wide degree of optimization potential. That is also why it is so important to move to this new architecture of transistors. Without this, there is no real movement. At 22-nanometer, frankly, without FinFET, there would have been no substantial movement in the line. That is also why you're seeing everybody moving to this transistor structure in spite of the fact that in many cases, they're doing it without any improvement in cost. Without that, there really is no benefit to migrating to that new technology. Now, transistors aren't the only thing, again, I talked about metal systems.
As we expand the portfolio and try to provide benefits across the product line, what you can see here is a cross-section, and there's really only one point to take away. The lines on the left are bigger. That's the technology optimized for our CPUs. Those lines on the top are faster. They transmit signals faster, they allow those parts to be faster. The lines on the right, at varying choices and optimizations, are smaller. Those lines provide better density. Those provide parts that are smaller, and they are not going to be as fast as the absolute best part, but that's the right choice to provide for those product lines. As we move into these markets, we're expanding these choices. Another way to look at this is just to take a look at the breadth of different technology flavors that we are now offering.
Not that many years ago, only the left two columns would have existed at Intel. Clearly, the first one is the primary driver of our basic server client business, and that's where all of our focus would have been. For many years now, we've been making chipsets, and those differed slightly from the basic capabilities of our CPU technology, but not substantially. Today, we've expanded this portfolio with processes that are targeted specifically at the tablet market. We even have an embedded DRAM process that allows us to build memory that goes into our high-end graphics parts, as well as a wide variety of parts that are focused in the mobile space. Some of these are used by our foundry, and some of them are only used by our foundry. Clearly, our FPGA customers have a unique set of requirements, and we have technology specifically targeted towards them.
As we expand the foundry business, we'll expect to be seeing more and more overlap in the capabilities that we provide. The fundamental improvement in transistors enables a wide breadth of technology offerings. We can tailor those to meet specific product needs, but again, the basic scaling of density and scaling of transistors is what derives the benefits of Moore's Law. Let me take just a moment and go back to time. All of these graphs today have been, to this point, based on named generation, and as I've shown, those names are also getting a little mixed up, but there's also the factor of time, and while most people have the same transistor technology on their roadmaps, they don't have it at the same point in time. We've showed this for a number of years.
These are the fundamental improvements over the last decade in transistors, and in each case, Intel has introduced them first and provided them to our customers, initially internally, but now to across our foundry spectrum before anyone else. Everybody's talking about improving enabling FinFETs. We ship these out as parts that you could buy at 22 nanometers. The talk about FinFET is clearly very vocal, but even today, those are talking about technologies which will be available for prototyping next year, but they won't ship to customers until 2015. The lead that we have is still substantial. You might say, "Well, that's just about transistors.
What about this density and delivering technologies?" Again, I've taken a foil from another one of our foundry customers and given their assessment of, with some industry analyst help, but looking at how far ahead they believe that we are introducing the fundamental technology. What you see is, I'm not going to stand up and say that there's any easy way to actually compare. You have to look at product releases, customer product releases, volume shipment. There's all kinds of different ways to measure. By any measure, we provide these technologies prior to anyone else. In summary, I hope, we continue to deliver the benefits of Moore's Law. These benefits are not just in cost, they're also in capability. We're committed, and we have a roadmap for the future where we believe this is something that is not only possible, but actually very probable.
We can, in fact, provide true cost reduction in spite of the increase in capital. We're going to do this by taking advantage of our internal design customers, our internal design partners, our internal CAD tool developers, to enable an even increase Amount of area scaling, which then provides that transistor improvement, that cost reduction, in spite of the increasing cost of the wafer. Finally, these benefits are not unique just to our high-end products, but they can be applied across the entire product portfolio. With that, it's my privilege to introduce Diane Bryant, and she'll talk to you a little bit about that high end, that business that makes a lot of money for us, our server business.
Thanks, Bill. We're super excited, of course, to talk to you about the transformation of the IT industry and what it means to the growth that we're seeing in the data center business. We are in the midst of a pretty significant transformation in the use of ICT, information and communication technology. IT has moved from a support role focused on delivering greater value and benefit to the business to actually being the business, IT-based services. This is being coined the digital service economy, where IT is actually fulfilling business, public sector, consumer needs. In the manufacturing segment, for example, it's being called the third Industrial Revolution, where you have now product development being done on devices that then are moved onto the cloud for online services. You have worldwide collaboration on product development. You have printing of prototypes down to 3D printers.
This is being called now the virtual factory. Fundamental changes in the use of ICT. A couple quick examples that highlight the opportunity and that this creates. The first example here is WeChat. You guys may know about WeChat. It was launched by Tencent back in Q1 of 2011. It started as a free mobile texting solution, text over IP. It has since been expanded to include walkie-talkie voice, photo and video sharing. They now have online gaming integrated into it, digital stickers, payment systems. In just two short years, it has grown to over 400 million users and has gone international. Another fun example, just from last week, you may have participated in this, but November 11th is 1111, it's China's Singles Day. It's the digital version of a Hallmark holiday here in the U.S.
This was invented by Taobao five years ago to encourage folks to go online and make purchases for yourself. This is when you splurge for your own benefit. In the 24 hours of November 11th, $5.7 billion were spent online, an 85% growth year-over-year for Alibaba. That reflects then three times the size of last year's U.S. Cyber Monday. Significant. If you just think about the data center infrastructure that is required to deliver this service. At peak, there were 17 million concurrent online shoppers. Alibaba handled 188 million transactions in 24 hours. Then you think about the logistics to deliver 152 million packages. I was thinking about this. I think Alibaba could lend a strong helping hand to Obamacare in how to deploy a full-scale solution.
I am also happy to say that Alibaba's infrastructure is running on Intel Xeon servers and Intel Xeon storage. These are examples of the foundation of this virtuous cycle of computing that we talk about. It is what fundamentally is driving the data center growth. You have new services that are getting launched. They attract more devices, those devices then increase the load on the data center, that drives the build-out of the data center in a cloud computing model that allows rapid deployment of new services, and the spiral just continues. Today, if you think about the Alibaba and the WeChat example, today, China is bandwidth limited, network bandwidth limited. Next year, they will release the fourth generation of LTE licenses.
The expectation is that with that increased bandwidth, it will generate lots of new services and demand, which then will drive infrastructure build-out to support it. We just see this cycle continuing and continuing. We have talked before to you about the correlation between devices, both consumer devices and machine-to-machine devices, and then the data center build-out. Now it is 400 phones. A couple of years ago, it was 600 phones per server. Now it is 400 phones per server. The reason is the average phone is consuming a lot more data than it was two years ago. 660 megabytes of data per month on average per phone versus 150 meg just two years ago. This is obviously a combination of better connectivity, you got higher bandwidth, and you have just got more interesting apps to access.
Below it, the wireless patient monitoring wearable is continuously streaming patient information like blood pressure and temperature and heart rate. It enables clinicians to remotely manage the patient. It is 100 devices to one server, but it also requires significant storage. There is a gigabyte of data per patient per day being stored. Then in manufacturing at the top, we did a proof of concept with an industrial control and automation supplier. They deployed 300,000 devices for control and management of the manufacturing environment that required 7,000 servers, a 1 to 40 ratio, and it also required one storage system for every 10 servers. The last one, digital signs. The ability to real time adapt to customers, dynamic targeting of content, that drives one server for every 20 digital signs.
It is significant If you take all that math and then apply it to a real-life deployment that we did in a Tier 1 province in China. We worked with one of China's leading digital surveillance providers, we created an end-to-end traffic monitoring system. There were 30,000 sensors deployed across the roadways throughout this Tier 1 province, another 20,000 surveillance cameras. The data feeding off of those sensors and surveillance cameras went into a network video recorder system that was running on Xeon E3 processors, that did all the video analytics to extract metadata like license plates and timestamps. That data is fed to the data center, where the solution runs on Xeon E5 servers with Intel's Hadoop distribution, Big Data Hadoop distribution, where the further advanced data mining occurs.
That Hadoop distribution was selected by the customer, the Intel distribution of Hadoop, because our solution provides an additional 40% performance over other available Hadoop distributions. That's because, as BK alluded to, we're investing in making sure the complete solution runs optimally on Intel. The result is reduced traffic congestion and improved safety. The system can accurately identify a car on any roadway by license plate or make, model, and year in under 300 milliseconds. That is real-time big data analytics. That's the solution. This is 50,000 Internet of Things devices. The forecast is, you've seen many forecasts, is that there'll be 19 billion connected devices by 2016, all of which are generating data, all of which then drive data center growth. With that as the context now, I'll go into our business plans and our forecast.
We look at our data center business from two different lenses. We look at it primarily by the market segment, the people that are actually procuring and deploying infrastructure and managing that infrastructure inside of their data center. That's either enterprise IT, it's the telco service providers, the public cloud service providers, or high-performance computing solutions. We also look at the business by systems category, our server business, storage business, and networking. With the expanded growth, as we talked about, just greater and greater usages of ICT, we see tremendous growth in the different types of applications and workloads that are running on infrastructure. We have everything from compute-intensive applications to IO-intensive applications to lightweight workloads, and we have workloads across server, storage, and network.
If you're going to deliver an efficient solution for these IT services, you need to have the infrastructure that is targeted and optimized for each of those workloads. As you know, we have a very broad range of products and technologies. That is a core strength of ours. At any point in time, we have over 100 processors in production covering that full space. This year alone, we launched 40 products targeted outside what you would normally think of the Xeon target of applications on servers. We had 40 different products launched targeted into high-performance computing or into the storage market and into networking. We also delivered 15 custom processors. These are processors that we developed in conjunction with our customers, both end users and direct customers, targeted at their unique workload, their unique application, or their unique data center environment.
We also launched our second generation of the Atom SoC to really target those lightweight workloads across microservers, low-end networking, as well as cold storage. We also announced that we're going to include Xeon into that SoC product line so that we can deliver the same density of compute through high integration, but with a higher-performing Xeon processor. Our job is to make sure that all workloads across the data center run best on Intel. Before I talk to you about future growth, I want to start by going back in time and looking at what we told you before and grade ourselves against that. In 2011, we made the statement that we would double the business to $20 billion by 2016. Back then, we did not break out how that revenue growth would occur by segment, but I'm doing that for you now here.
If you look at the green boxes, that reflects our revenue model at the time, back in 2011, of where that revenue comes by segment. The gold boxes next to it reflect actual growth. That's actual growth from 2011 to 2013. As you can see, we've done very well against our original commitment. We've exceeded in cloud growth. We've exceeded in high-performance computing growth. We've hit the growth in the telco market. We've hit those assumptions. Where we didn't hit is pretty clear. The enterprise IT segment just did not grow as expected. In fact, the market has actually contracted over that period, and you've heard BK and Stacy talk about that in earnings calls over the prior quarters about the enterprise IT market being soft. The biggest driver of enterprise IT growth is GDP.
IT spend is heavily correlated to GDP, and as Intel CIO for four years, I know that as a fact. Andy Bryant never hesitated to pull my budget when the economy was hitting a low. That correlation does tie very nicely. The other driver of enterprise IT growth in the prior days was, or I'm sorry, the slowdown in the most recent years is the fact that enterprise IT has basically completed the virtualization of the data center. IT organizations around the world invested primarily in 2011 and 2010, '10 and '11, in virtualizing their server environment. No one would virtualize on old servers. You're just not going to. You need the hardware assistance of virtualization technology that we included in the processors in order to virtualize. That virtualization process created a growth spurt in servers in the 2010 to 2011 era.
Now that virtualization growth spurt is pretty much over, as well as GDP, and that's why we believe we see the results that we see here. If you look in total, we had forecasted a 15% CAGR. The actual then from 2011 to 2013 is 8%. Okay? Now looking forward, going from 2013 to 2017, we do expect continued robust growth in cloud service provider markets, in high-performance computing, and in telco, and I'll tell you why we believe that that growth will continue. We do believe that enterprise IT will recover. We are seeing signs of that in the second half of this year, that enterprise IT is starting to spend again. We've set our model for enterprise IT at 8%.
That's anchored off of analyst projections on volume, taking into account what we believe the uplift will be from an ASP perspective, as well as our growth beyond just processors. We'll talk more about our expansion beyond the CPU. In addition, prior to this virtualization growth spurt that we saw in 2010 and '11, enterprise IT was generally growing at about 8%. We think 8% is a reasonable number for enterprise IT. As BK's model clearly showed you, we're becoming less and less dependent upon that slowest-growing segment of enterprise IT as we move forward. Obvious math that it becomes a smaller and smaller portion of our total revenue. Okay, now going into each of the segments, I want to talk about what we see as the big growth drivers and what our Intel advantage is in those segments.
We've talked a lot about the high growth of the cloud, public cloud service provider market, the Amazon and Facebooks and Alibabas. It's good to do a click down into where that growth is actually coming from because there's a general perception that the growth of cloud service providers is a cannibalization of enterprise IT, and that's what IT trend. The vast majority, 75% of the cloud service provider volume is going into consumer services. It is incremental. It is TAM expanding. Things like Facebook and iTunes and Instagram, right? We know all these services. 25% of it are solutions targeted at enterprise, whether it's Azure or Google Apps for business analytics, Salesforce.com. 25% of it is targeted enterprise, but the bulk of it, 75%, is consumer-directed. The second point I want to make is that the cloud service providers value technology like nothing else.
They truly value technology. Their data centers, the infrastructure, and those applications, that is their business. They see the value of riding Moore's Law, very direct value. Their business metric is performance per TCO. As you heard Bill talk about, that's what Moore's Law does. It gives you greater and greater performance at lower and lower power. Power is the biggest contributor to operational expense. You have a very significant improvement in your primary business metric as you move to next-generation technology. We see that. The top four cloud service providers had deployed next-generation Intel processors, Ivy Bridge, before we had even launched them. They were taking them before we had actually achieved full production quality, full production launch, because it meant that much to their business to get those processors into the data center.
We also see a continuous maturing amongst the cloud service providers in how they model the data center and how they manage that metric of performance per TCO. As their models mature, they continue to recognize that there's value in not just buying the latest technology but buying high in the stack. Of the top seven cloud service providers, six of them have opted to pay for more to buy higher up in the CPU stack and get higher performance for the past two CPU generations. Direct value to them. In addition to them valuing leading technology and at the high end, they are also early adopters of other technology that we are providing. They are the leading adopters. We're the leading adopters of 10 Gig Ethernet. They're the leading adopters of our next-generation solid-state drive products.
In fact, they lead the general data center population by 30% in the adoption of additional technology beyond the CPU. Our advantages with the cloud service providers are great, and it starts with the technical collaboration that we make in direct engagement with the cloud service providers. Certainly deep technical engagements with the top seven, but direct engagements with the top 40. The top 40 cloud service providers cover 70% of the total volume. Through these engagements, we get a very clear understanding of their workloads. We get a very clear understanding of how we can benefit them, how we can help them with that performance per TCO model. We obviously have a broad range of products to address those needs.
Although we do have a very broad range of products, I will say that 95% of all volume that is going into the cloud service providers is two-socket Xeon E5 processors. Again, a very clear statement that this is a segment of the market that truly values performance. Where also though, in some situations, a customer may want a very targeted solution, targeted at their particular workload where they think an incremental performance boost or an incremental accelerator will give them a competitive advantage against their peers. So we are now able to meet those needs. We are able to deliver custom solutions. A couple examples up here. One is eBay. eBay has a very innovative proprietary cooling solution in their data center, so we were able to deliver to them a unique processor that actually raised the top performance up by another 50%.
Dramatic performance improvement, given the environment that they were running the processors in, the servers in. For Facebook, as you know, they store lots and lots of your pictures that, I am sorry to say, are never looked at again. So 300 million per day, in fact. They needed a very low power, low compute, it is not going to be accessed very often, storage solution. So we delivered two unique processors for their cold storage solution. Then the last example is that some of the cloud service providers are looking to us to embed unique IP, unique accelerators into the CPU, targeted at whatever their particular service is, and one example of that is voice recognition.
We are working with Nuance to develop an accelerator around voice recognition, so pattern matching, to integrate into our processors targeted at a service provider that is delivering that type of service. Lots and lots of innovative investments we are making, thanks to very tight collaborative engagements that we have. We also have a breadth of expertise at Intel that I will say is unmatched, and we are obviously a silicon company, but we have great strength in systems design, in data center architecture, in software design, software development. We leverage that breadth of talent to engage with the cloud service providers on the next generation of data centers. You see here the rack scale architecture.
This is a new way of thinking about the rack, taking the rack and breaking down those individual servers into pools of IO, pools of memory, pools of compute to deliver greater efficiency to the cloud service providers. One instantiation of the rack scale architecture has been contributed to Open Compute Project, the Facebook-driven consortium. Another version has been contributed to Scorpio, a similar initiative out of China. By having this direct engagement at a solutions level, as BK was mentioning, we are able to obviously better target our products to target where they are going in the future.
We do believe that through efforts such as rack scale architecture, that we will increase our share of that total rack solution by 15%-40%, as well as we will be able to increase the refresh rate of our processors back to a clear value proposition to the cloud service providers. Moving now to enterprise, and the growth drivers of the enterprise market, I said, it has been a tough couple of years. The first two quarters of 2013 were also quarters of contraction of this segment. I also said, IT spend is highly correlated to GDP, and you can see that on the left. The yellow line is GDP. I'm sorry. The yellow line is enterprise volume. The bars are GDP. If you look there, GDP is now forecasted to pick up by about a point next year, 3.4%-3.6%.
That will obviously on the enterprise IT spend. There are also other growth drivers for enterprise IT that will obviously contribute to that 8% that we are now forecasting. One of those drivers that we firmly believe in is the deployment of private cloud solutions. The value of the cloud is improved total cost of ownership, lower total cost of ownership for the IT organization. It is a very clear statement. The cloud service providers are obviously benefiting from that lower total cost of ownership, but enterprise IT sees the value, and they are motivated to move to a private cloud deployment. What is also true is that deploying a private cloud on-premise is going to cost less than utilization of a public cloud solution. As much as half of the cost per virtual machine per year.
We see this play out, we see this trend where companies will start out in a public cloud solution. A startup like Zynga will go to the public cloud solution because that gives them very rapid time to market, very rapid scale. But then once that company gets to a certain scale, a certain size, and they have the economies of scale behind them, then they pull their applications and solutions out of the public cloud and deploy it on-premise in a private cloud environment. We see this Zynga as an example here. There is many other examples that we cannot give specific names, but there is a clear trend because of that lower cost of owning and managing your own cloud locally versus as a service. The first step, if you are going to develop a cloud environment, the first step is virtualizing the servers.
As I said, we saw that happen a few years back. When it happened, it increased our server growth rate. That was, again, because enterprise IT needs to buy the latest generation technology in order to deploy virtualization in an optimal manner. We do expect that same phenomena will occur with the cloud. We do expect next-generation hardware will be deployed as enterprise IT moves to a private cloud environment. Our next-generation hardware will do more and more and more about exposing the underlying attributes of the silicon up to the cloud orchestration layer. Attributes such as utilization levels, security level of the infrastructure, the power level of the infrastructure.
By exposing those fundamental attributes up to the cloud orchestration, the cloud orchestration level can make sure that the service level agreements for that application are met, that the application isn't going to be bottlenecked, and that the operational costs are actually minimized. That's the first thing that we believe will happen in the move to private cloud. The other is security is always top of mind when you talk about cloud environments. Multi-tenant environments, it tends to raise concerns around security. With each generation of our technology, of our microprocessors, we deploy greater and greater levels of security, new solutions and new features, new improvements in the crypto algorithms, secure keys, trusted pool environments, so you know that your private cloud is in a trust attestation mode. Security will also be a driver for refreshing the environment.
Third, a true complete cloud solution is one that exposes all of the infrastructure up to software. Not just servers, but servers, storage, and network, and that would drive a refresh of the storage and network solutions as well. The move to software-defined networks and software-defined storage. The combination of those three, we believe, will drive a refresh of the infrastructure in support of the move to private cloud. The other driver is around big data, and there is a lot of talk these days about Internet of Things and big data. It's all the buzz. It's all the buzz for a good reason. There's a lot of value in it. Data management analysis has been the job of IT forever, but the data is changing significantly, right?
The data is moving from structured to unstructured, fueled by all the social media sources, blogs, and videos. The data is also growing dramatically in volume, driven by the billions and billions of connected devices. Thanks to Moore's Law, thanks Bill. With Moore's Law, we've been able to drive down the cost of both compute and storage. Now it is affordable to amass all that storage and actually compute on it. Also thanks to new data management solutions like Hadoop, an open-source solution, as I said, that we're a distributor of. It's now possible to analyze and store these massive sets of data and extract business results from them. Intel recently, earlier this year, we did a survey of CIOs. 82% of the CIOs agreed that they could deliver better business results to their company through big data solutions.
Only 6% said that they are actually today making decisions based on big data analytics. That gap is a clear opportunity. The good news is, as you can see on the slide, we have a very compelling and probably the only, I think we are the only, that have a true end-to-end architecturally consistent solution to deploy big data solutions. From the intelligent devices out at the edge that are secured with McAfee running on whether it be Quark or Atom or even Core, depending on the solution, to the mid-tier solution, the gateway solution that aggregates the data and filters the data that runs on Xeon or Atom, running with our embedded OS from Wind River and obviously the McAfee security solution.
Back up into the data center to the data-intensive analytics that run on Xeon for compute and storage with our optimized Hadoop distribution. End-to-end, all running on Intel, consistent architecture, consistent instruction set, consistent security solution, end-to-end big data solutions. We have seen that once an enterprise deploys one big data solution, it will lead to many more big data solutions because the value back to the corporation can be so great. For instance, there's a large telco provider that deployed a big data solution that was originally targeted at network optimization. Once they had amassed that data set, they then used it to mine call records for demand trends, they've layered on top of it now marketing to optimize the billing plans. This continuous evolution and growth of big data solutions within a given enterprise IT.
The combination then of GDP recovery and new solutions that will be deployed, such as big data, Internet of Things solutions, and private cloud, this is what gives us the confidence in the 8% growth number that we're now projecting. On the network side. There is a major transition that is accelerating and has been accelerating over the past five years, this is the transition of the telco industry onto general purpose computing. This is occurring across all of the compute workloads inside of telco, from signal processing in the base station at the edge to the control function in the gateways to the data plane processing in the core to the applications like security running in the back office.
This transformation of telco is happening to offset the capital, the ever-growing capital and operational expense of running the network while accelerating the ability to deploy new services out to the customers. This transformation is very similar to what we saw in the '90s with servers and the move from proprietary RISC architectures onto open Intel platforms. To this goal, as you can see, AT&T just announced what they call Supplier Domain 2.0, this is their program to accelerate their transformation of the network onto a new architecture. It mandates that all of the OEMs and TEMs that want to do business with AT&T need to move from proprietary closed appliances and architectures to open hardware and software solutions. As AT&T said, they are restructuring to make the network a revenue-generating machine. This move is very clear.
We've been engaged in this move of the compute workloads onto Intel architecture for the past few years, we just now see that trend accelerating. The move to software-defined networking in the middle and network function virtualization, these are also big industry trends. Software-defined networking is the industry phenomena that's occurring, recognizing that the network today is very difficult to run and manage. It is still heavily manual. Everyone has network administrators, that when a new application or new service needs to be deployed, it's a manual process of reconfiguring the network in support of that. Software-defined networking says, take the control of the network and raise it up to the data center level and manage your network holistically in an automated fashion.
Intel IT has moved to software-defined network solution. They are a proof point that you go from weeks to deploy a new network architecture in support of a new service to now just minutes, fully automated. This is a big trend. It's a trend that is not just telco by any means. It's a trend that benefits all networks. It's a trend that we've seen strongly adopted by the cloud service providers. In fact, you may remember in 2012, Google actually did the keynote at Open Network Summit and talked about how they had moved their network to software-defined networking. Network function virtualization, that traction is also very evident. Over 25 of the carriers have come together, joined ETSI, so the telco industry working group, and they, through the course of just one year, have released the specifications around network function virtualization.
If any of you have worked with the telco industry, one year is warp speed. I mean, that is incredibly fast to get that number of parties together, agree on a specification, and launch it out to the industry. We now have over 15 carriers, folks like BT, Telefónica, SK Telecom, China Mobile. All of them have regional deployments of software-defined networks and network function virtualization running on Intel architecture, so they're in qualification. As you know, the process of deploying new infrastructure into telco is a very measured, deliberate process. They have regional deployments and looking forward to moving into production. Let's talk about the clear Intel advantage then in this massive transformation that we see in networking.
First of all, just like we drove and enabled and benefited from the consolidation of applications in a server virtualized environment, that same opportunity is emerging, and will emerge, in the network appliance side. In the networking world, all of those proprietary fixed function appliances, whether it's VPNs, gateways, all of that is moving now to an application running on a virtualized, or a hypervisor virtualized layer running on standard Intel hardware. That's the trend of network virtualization. The benefit from capital expense and operational expense is that that Intel server solution is an open industry standard. As Brocade points out at the bottom now, with their firewall example, that solution now transforms the economics of the network dramatically.
Today, we have only 5% share of the total network market, and that's looking at all compute inside of the network, whether it's switches, routers, base stations, line cards. We have just 5% of that very large market. You probably know we now have 94% of the server market. We have over 80% of enterprise storage, but just 5% of networking. A clear opportunity. It's also very clear why we only have 5% of that market is historically, it has been a market that has been driven by highly customized, highly tuned ASIC solutions targeted at the different workloads inside of the network. Thanks to Moore's Law, we just have so much to thank Bill for. Thanks to Moore's Law, we've just been able to continuously increase the number of cores per chip.
Thanks also to a significant investment that we made in building a carrier-class Linux solution, so an open operating system alternative to the proprietary operating system. Also thanks to our investment in workload accelerators, so data packet acceleration that we've embedded into the platform. The combination of those has allowed us to increase the performance by 6x over the past 5 years, it is well accepted now in the telco industry that Xeon is a compelling alternative solution, a compelling solution to the historical custom ASIC world. We have launched developer reference platforms to help make it easier for the transition off of proprietary onto Intel architecture. Those are out in the industry, with the telcos, for both software-defined networking and network function virtualization.
Another segment that is growing rapidly that we're always very excited to talk about, a segment that truly values the highest performance, is high-performance computing. As you can see, the worldwide investment in supercomputing just continues to grow as countries recognize that if they're going to compete at a global level, they need to be investing in scientific computing. Over the past 10 years, the performance of the top 500 supercomputers has increased very consistently, 50% every single year. If you just look at the number one supercomputer, it doubles in performance every single year, so moving at twice the pace of Moore's Law. Today, we hold the number one supercomputer. The new list just came out 2 days ago. Obviously, we're still number one. That's a system that has over 80,000 CPUs, a combination of Xeon and Xeon Phi in a single system.
It's a very large and powerful system. I do want to say it's not all about government and academics, where our growth is coming from. Obviously, the supercomputing space is a big part of it. Commercial adoption of high-performance computing is also on the rise. For instance, the middle there, genomics example, TGen. They took their platform, they refreshed their platform, moving from a 4-year-old Intel processor to the latest generation. They added 10 Gigabit Ethernet or 10 Gigabit Ethernet solution and solid-state drive solutions, they were able to take the period for a full genome sequencing from 7 days down to just 4 hours. Dramatic reduction. When you have that kind of dramatic change, that really just completely transforms the whole diagnostic process. There's so much you can do that you couldn't do before.
Last year, we launched Xeon Phi, so it's 1 year old now in the market. It's our co-processor for highly parallel workloads. We see the interest in co-processing solutions, heterogeneous solutions. We have the processor and the co-processor. The interest in the HPC community is growing significantly, as you can see there, over just a 2-year period when we survey HPC both the build and deploy HPC solutions, the interest level has gone from 28% to 77% interest and intent, I should say, in just those 2 years. Six months after we launched Xeon Phi, we took the number one position in co-processing flops across the top 500 list. In 6 months, we surpassed Nvidia, who Nvidia has had product in the market now for over 6 years. We've moved into the number one position and are doing very well.
The reason for that, the reason for our acceleration to number one is very clear. It's the IA programming model. We have a common instruction set from Xeon to Xeon Phi. The Intel instruction set is a strong industry asset. It's an asset that consistency of the asset is critical for maintaining the economics behind building applications to run on these massive supercomputers. It's a very well-understood programming model. It's got very well-established tools. There's a large knowledge base behind it. That common architecture between Xeon and Xeon Phi has tremendous end user value. We're now at an all-time high share of high-performance computing. We have 90% share of the total HPC market. That's a record for us.
If you just look at the top 500 list, we watch new systems coming onto the list because systems will stay on the list for multiple years. 96% of all new systems coming onto the list this year are running on Intel architecture. The alternative architectures continue to age and decline. I do want to spend a minute to talk about our Atom SoC product line. This is a great example where we identified an opportunity, got out ahead, and led. That opportunity is in this emerging class of lightweight workloads. Whether it is entry-level networking, branch routers or cold storage solutions or microservers, this is a place now that we can participate in and win very well at. We just launched our second generation Atom, as I noted. At launch, we announced 50 design wins on that product.
27 of them were in the networking space, of those 27, 10 of them were architectural conversions, so conversions off of MIPS and Arm onto Intel architecture. We're winning share in this space at the entry level in networking with our Atom SoC. There is a lot of focus in the industry around the microserver segment. I recognize that. Today, the microserver segment is still less than 1% of our total volume, it remains a small segment. As Stacy will reinforce later, we have very good margins on our Atom SoC product line, thanks to the fact that it is a very low product cost. We have absolutely no reservation about competing in this space of lightweight workloads with our product line.
The benefit, as I mentioned, has been that we are able to address workloads that the Xeon processor just wasn't able to address efficiently before. Where we have microserver design wins, and we look at where we're winning them from, we are winning them from the very low-end AMD single-socket solutions, we are winning them honestly off of desktop-on-a-side solutions. These are dedicated hoster environments where you want a single website per node, you don't need a lot of compute, you just want a lot of nodes. In some instances, it's desktop-on-a-side solutions that are now being sold as real servers with the Atom SoC. Then, of course, because the Atom is instruction set compatible with Xeon, the ability to introduce Atom into the data center is very seamless. It's very efficient. It runs all the same software.
Cloud service providers are very, very clear to us that a homogeneous environment is the lowest cost solution for them. They like the consistency of architecture throughout their data center. The investments that we are making in ecosystem enabling and the value of that continues to be a very clear differentiator. Investments that we make in things that seem as simple as firmware and driver developments, providing on-site technical support. There was a design win early on, earlier last year, that got a lot of press around a cold storage solution running on Arm into one of the big cloud service providers in China. That solution has now been pulled out of the data center due to instability issues in the software stack.
That enabling that we do of not just providing the CPU, but enabling all of the elements around the silicon is incredibly critical and valued by our customers. We're committed to winning in this low-end space, leveraging those same wonderful assets that have allowed us to win as we have in the data center with Xeon. I've covered the four market segments. Now I want to talk about. Well, I should say, if you look historically, we were very focused as a server company at one point in time, and then we have obviously moved to all the other systems in the data center, so server, storage, and network. We are also expanding from CPUs to additional products and technologies in the data center.
I want to talk about three of those investment areas that we're making that you should be aware of as we talk about our future business, and that's fabrics, Silicon Photonics, and non-volatile memory. Today, we hold a very nice share of the discrete Ethernet market. We're number 1 share across 10 gig and 1 gig, we have 56% share of the market. All data centers run on Ethernet. That is the standard. It will be the standard forever. What we are doing is we're integrating Ethernet into our CPUs. We did that for the first time with the Atom SoC solution that we launched just back in September. This provides the end users with all the benefit of integration, all the benefits of our continued transistor density growth. You get lower power, you get smaller footprints, higher density solutions, and lower cost.
It's all the goodness of integration. Our plan, our strategy, is we will continue to integrate Ethernet into our product line. We'll launch it into Xeon next year. We will continue to grow our footprint of integrated Ethernet into the processors, as well as continue to deliver our discrete Ethernet product line. If you look at the high-performance computing side of the world, it is much different than the Ethernet data center world from a fabric perspective. The demands on the fabric are intense. Much higher expectations on low latency and high reliability. In that space, we are currently leveraging the acquisition, the assets that we acquired through QLogic, their discrete InfiniBand product line. That is our solution today.
What we have done is we've taken that technology, we've advanced it, and are integrating it now both into Xeon as well as Xeon Phi. Those products will start coming out in 2015. Our customers agree that the integration of the fabric into the processor for high-performance computing is highly valued for all the reasons we just talked about of lower power, higher density, higher performance. Some will even say it's actually required if high-performance computing is going to continue to grow in performance at the pace it's been growing. We do believe this will be a clear competitive advantage for us going forward. A new product line that we're super excited about, and that is going to be demoed here today, so I do hope you have time to go visit the demos, and that is Silicon Photonics.
Thanks to Moore's Law, compute density continues to increase, so we get greater and greater compute per socket, and more and more sockets per rack. With that ever-increasing compute density, you create a bottleneck, and it's how quickly you can feed data to and from that compute environment. The network rapidly becomes the bottleneck for performance. Copper cabling, as you know, is the current data center solution, but copper breaks down at over 20 gigabits per second. In general, the data centers have moved to 10 gig now, and in some places, you will see deployments at 40 gig, and then of course, 100 gig after that. Copper starts to fail at 25 gig, or the copper cabling becomes prohibitively expensive and high power-consuming.
The general manager of Microsoft's cloud services, he was at one of our events, and he said that cabling has become now one of the biggest challenges to the data center for scale. The masses of cabling inside the data center are actually blocking airflow, so you have problems cooling the data center, which as we talked about, it is the highest operational expense in the data center. His quote was, "Wires are really hard. Really hard." Bill would say wires are hard, but he would say wires are even harder. The move to optical is the logical next step. Getting photons across fiber instead of electrons across copper. It delivers much higher bandwidth at longer reach. However, current optical solutions also hit a limitation. Current optical solutions use esoteric materials that are expensive and that also have limitations above 100 meters at 20 gig.
In steps Intel Silicon Photonics solution, enabling all the density, the higher density, higher bandwidth, very lightweight cables. It solves the data center problem. It delivers very long reach, and you get all the benefits of silicon, of lower cost and lower power. We have already publicly demonstrated 25 Gig running at over 800 meters, a clear record-breaker, and we have, as you can imagine, incredible interest across the segments, high-performance computing, where you've got very dense compute, the telco market where you've got very long reach, so you need the reach of Silicon Photonics, and then the cloud service providers, as I mentioned. We have started sampling. We will begin revenue shipments next year, and Silicon Photonics becomes a meaningful portion of our revenue in the latter end of the forecast horizon that we're talking about today. Next is the storage market.
The storage market is going through tremendous transformation between the rapid growth of volume and the variety of the data that's being stored, combined with the fact that the utilization of storage today is pretty low, it's a very unoptimized environment. All of those conditions are driving a rethink of the storage environment and the different storage tiers, different tiering from very high performance, frequently accessed hot data, down to the very high-capacity, low-cost cold data solutions. The overarching solution for the storage challenge that the market is seeing is increasing the intelligence of the storage solution. You can see that transformation in our numbers. You can see it in the ever-increasing proportion of our shipments into the storage market that are Xeon versus Atom and Core. We've gone from 38% Xeon to 54% Xeon in just two years.
We're obviously investing to make our solutions run best in a storage environment on storage workloads. That's a silicon statement as well as a software statement. The other big trend on the right here in the storage world is the use of non-volatile memory and solid-state drives. Intel's solid-state drive business, that's run by Rob Crooke, has seen tremendous growth in server adoption. It was led by the cloud service providers, who are very compelled to adopt, thanks to the tremendous application performance benefit that solid-state drives deliver. We will see 50% growth this year, and we expect the growth to just continue as it moves from solid-state drives adopted by cloud service providers into other portions of the business. As we know, the cloud service providers, they're generally the early adopters. They prove out the technology, the rest of the market follows.
I will say the innovation, the tomorrow statement there, the innovation for us in this space continues. We are working on multiple non-volatile memory technologies that are going to further blur the lines between what is storage and what is memory. We believe we are in a unique position to facilitate a technology transformation given our presence in the data center and our non-volatile memory position, which, as I might have missed, is we have number 1 share in data center solid-state drive. We're not talking a lot about that today. Further details will come on our future technologies, we're super excited about the business opportunity that this is going to bring that you will see impacting the second half of this horizon that we're talking about today from a business revenue perspective.
We are, I hope you see, we are leveraging all of Intel's assets to cover the needs of the data center, our assets are many. We have leadership process technology. Bill clearly showed you architectural consistency in the data center is a huge advantage for us. As I said, we hear from the cloud service providers that the way to drive down cost to improve total cost of ownership is to have a homogeneous data center environment. If you have to add a second architecture into the environment, it means you've got a second operating system, a second middleware stack, management solutions, developer tools, all of that just adds cost and complexity. The big win is when you can deliver a consistent architecture top to bottom across all workloads, we can do that.
Software compatibility, there are millions and millions of applications that are around the world running on Intel architecture. It's a huge collective investment by the industry, not one that would be disrupted easily, for sure. You've also hopefully heard that we've moved from a focus on general purpose computing, to also augmenting that product line with custom solutions targeted at particular customer needs, as well as deploying our SoC capability that allows us to generate additional processors, different SKUs off of the base, targeted at the different workloads in a very rapid and cost-effective manner. We are obviously known as the CPU company, but our technology is extending well beyond CPUs, the fabrics I talked about, non-volatile memory, the accelerator solutions, Silicon Photonics, data center software like our Hadoop distribution. The technologies that we're providing into the data center continue to grow beyond CPUs.
As I said, we invest substantially to enable our customers to deploy the best solution, complete solution, running on Intel, whether it's working with the industry on next-generation memory technology or next-generation IO, to the compiler work we do to make sure that their application runs best on Intel, interoperability testing. There's just so much we do for the industry to make sure that they have a very easy and simple deployment of Intel into their solution. These are investments that we uniquely make and will continue to make as the data center IT industry goes through this transformation. To conclude, the rise of the digital service economy is transforming the data center infrastructure, as we've seen. We are uniquely positioned to win given our unmatched ability to deliver leadership solutions across this ever-growing spectrum of workloads. We do continue, as Brian said, to deliver double-digit growth.
We continue to forecast that out through 2017, we certainly look forward to continuing to capitalize on the significant growth that we see in both high-performance computing and the cloud service providers, maintaining our share of that environment while accelerating the market with new solutions. We look forward to driving the transformation of the network onto open solutions from proprietary onto Intel and benefiting from that massive transformation that's happening in the network. With that, I want to say thank you.
Good morning. I'm excited to talk to you about yet another era of PC transformation that we're going through. It's a pretty exciting time, if you've been watching the last year on the PC and the innovation that's going on. There's really four things I want you to take away today. One is, we are seeing that stabilization led by mature markets and business down to kind of a negative small single-digit growth as we look 2013 to 2014. As Brian said, we've become a market-sensing company. In fact, we're doing over 250,000 end-user surveys a year now, and I'll show you a bunch of data that kind of reaffirms that PC is really continuing to be the primary computing device in the home, and that this 2-in-1 trend that we have is really resonating.
The third is that I want to give you some perspective on segmentation. Just like Diane said, between HPC, cloud, enterprise, there's a lot of different trends going on within the PC market, and in many cases, we're achieving record highs in very strong growth segments. I'll show you that and how it's helping our mix, our market segment share versus our traditional competitors as well as our new competitors, and our average selling prices. Lastly, I just want to give you a perspective of the significant amount of innovation that we're doing around everything from these amazing new form factors in desktop and in mobile. The embracing we're going to do of operating system choice in Chrome and Android and Linux and Windows. How we're aggressively tackling now the value segment. Post the netbook era, we let that get away from us.
You're going to see a significant investment back that we think helps us grow share. Really focusing and capitalizing on our business leadership and the vPro momentum we've had. Last but not least, the new experiences that we're bringing into the market. As Brian said, if it consumes electricity, we think it's going to compute. If it computes, we think it's going to connect to the internet, and if it connects to the internet, we think that it'll do its best on IA. There's a lot of different perspectives on what the PC is. At the end of the day, PC means personal computing. Whether it's the gray area between tablets that Hermann will talk about, these new detachables, two-in-ones, or even large-screen tablets going into portable all-in-ones, our strategy is very simple. It's going to do it best on IA.
What's very clear is we're right in the middle of yet another transition in the PC. We've gone through these as we went into Centrino in 2003, and the world moved from desktop to mobile, and we're going through that once again. Today, I'll talk to you about five mega trends and strategies we have. How we're embracing operating system choice based on meeting what the market wants and getting ahead of that. Second is kind of defining for you what we mean by this two-in-one computing trend. Third is what we're doing in business, what we're doing to reinvigorate the desktop, which continues to be a very strong segment for us. Lastly, the new experiences. What do we mean by client operating system choice?
Well, when I was up many years ago talking to you about the data center, it didn't surprise you that we were talking about Windows and Red Hat and Solaris as we brought Sun and Oracle from SPARC to IA or even HP-UX on Itanium. The reality is, for the last decade, we've been essentially 100% Microsoft on the client. What we're seeing is trends where certainly we're excited about the increases in Windows 8.1, but we're starting to see out of emerging markets demands for Android. Certainly, we've had strong strength in Apple that I'll talk about today, and also Chrome and even Ubuntu Linux. The strategy is very simple. We're going to support what the market desires. These are tier 1 operating systems, and we're aggressively investing in all of these.
In Android, as Brian said, essentially a lot of Android apps were written for a phone in portrait mode. A significant percentage don't even exist in a landscape mode. They can't scale to large screens, whether it's a notebook or an all-in-one. We're going to do a number of things here. We're going to scale Android to 64-bit. We're going to allow it to scale from Atom all the way to the high end of the Core processor family. We're going to enable it to scale and deliver a great experience as we go into larger scale screens, allow multi-windowing. While most of us would put an Android device using MobileIron or Good onto our corporate network, we don't yet have the ability to really treat it as a full-on enterprise vPro client. There's a significant investment as we embrace Android and move that way.
On Mac, we're extremely excited about the MacBook Air. As we said, essentially cooperating with Apple doubled the battery life from generation to generation from six to 12 and a half hours on MacBook Air. It's beyond that. If you look at FaceTime and the amazing video conferencing experience you get under the OS X, it's because they're optimizing for our Quick Sync video underneath. On Chrome, I'll go into this in a significant way. We have a significant opportunity to gain share here versus Arm, and we'll talk about the devices there. In Ubuntu Linux, in education, there's millions and millions of units actually running on Linux around the world. We're working on ensuring that our graphic solutions and everything run there. Then on Windows 8 to Windows 8.1, there's a lot of excitement happening.
In the business side, remember, Windows XP end of life is essentially in April. We have some confidence that that business refresh, which typically comes with a hardware upgrade, is heading our way. Enterprises are on track for that transition. Whether they move to Windows 7 or Windows 8.1, that's a big change for business. If you look on the Windows 8.1 side, there are now over 100,000 apps with new apps like Flipboard, Twitter, Facebook. Some of those big misses that we were all wanting to embrace the Windows Store and the Microsoft operating system are there, in addition to being able to boot into a desktop and stuff.
As Brian said, the advantage for us or one of the key advantages, is we can now deliver a common motherboard, and that's either good for our OEMs and our design wins in our OEMs because they can deliver a common motherboard that runs Android and Windows, for example, and they can load that at the last moment on their manufacturing line. We're also seeing people that are actually interested in doing dual operating systems, putting the ability to boot into either one, and giving the user the choice at the boot time through BIOS to boot into Windows or boot into Android. Even switch back and forth, as you've seen with some of the recent ASUS products. As an example, one of the things we're seeing is we've really embraced the China technology ecosystem.
Taiwan is still a mainstay of where computers get built. If you go into Shenzhen now, you'll see hundreds of companies that are building tablets, and as that becomes more and more crazy on the margins for them, they're migrating up to notebooks and even all-in-one computers. This is a product from THD that will sell at $299, built on Bay Trail-M, and it's got a nice 295-degree hinge and is running Windows 8 or 8.1. What you can see here is the exact same product is actually running the latest version of Android.
This is just an example where they actually have a key on the board, and you can switch between the two operating systems as a user, boot into either one, and it's given them tremendous manufacturing capability as they look at what to do in their China market and as they expand worldwide. Great price points and embracing all the operating systems. If you look at the segmentation we're doing, I think the first thing to note is that we have a pretty even split in our revenue between business and consumer. In the premium segment, a lot of people don't realize, but if you just look at the third quarter, we had record mix to Core i5 and i7 across our Celeron through Core i7 product lines. Record absolute units for Core i5 and i7, and record units on our vPro technologies.
The market is absolutely embracing the high end. We'll talk about how we grow back in the value space, some of that which was competition with AMD, our traditional competitors, some of that was elongation of a refresh cycle due to tablets. If you're looking at the high-end space, vPro, more than 20 vPro Ultrabooks we'll talk about heading to the market in the fourth quarter. A new Haswell Y-Series that's kind of right in the seam between tablets and notebooks for business detachables, and then a whole new range of value-oriented Bay Trail products for small business we'll talk about that are really brand new, cost optimized, and go head-to-head with the best, just like the product you saw there. In the consumer side, at the high end, we'll talk about our Iris graphics solutions.
We're putting more and more transistors into graphics, and we're getting paid for that, taking share from discrete. That's a great upsell for us. In addition, we're launching up to 50 two-in-ones by the end of the year in the consumer, and that again spans performance and mainstream. We'll talk about really going aggressively after this entry space where we think we'll gain share versus Arm in things like Chromebooks, embrace Android, and gain share against AMD in the value space as we go into 2014. Let's start at the top. In 2011, we really sat up, Tom was up with Jonney Shih, and Sean was up with Jonney Shih at COMPUTEX introducing this three-year journey we said around Ultrabook. We said that Haswell was really the first product that would be designed from the grounds up for the Ultrabook.
The first generation was really making everything thinner. We stopped counting when basically the entire market went about 20 times the size underneath an inch thick, everything thinner. From the Ultrabooks all the way down to the most valuable products. We think that's a huge success from the marketplace. If you look at touch now in large format retail, the latest NPD data says we're at about 40% touch of Windows 8.1 systems in large format U.S. retail. Ultrabooks on a week-to-week basis are somewhere between 66% and 80% touch as they're going out the door right now. Touch, we're on a journey. Costs roughly about $45 or so today for a 13-inch, dropping to probably $35 and continuing that trend as we go into 2014. Trying to make touch, I think most people would agree touch will be relatively ubiquitous.
It's just a matter of time. About 50% of our fourth generation Core Haswell designs are touch-based from a design win perspective as well. All that's getting into the market now. Really, it's really this 2013 product with fourth generation Core that has been so amazing. I stood up and talked about systems at $699. We now have 19 Ultrabooks on bestbuy.com that are less than a $699 price point and 11 Ultrabooks on bestbuy.com that are under $600. Those price points are starting to come down, and it's really just been amazing battery life. More than nine hours of battery life on a typical Ultrabook, hot swappable batteries that can get you 24 to 27 hours just by changing a single battery. You can see wired.com, basically instant access to your data, record boot time, and amazing battery life.
What we're trying to do with Ultrabook, on the left here is what we said at our Intel Developer Forum in 2011. We said, "Hey, it's going to be a tablet when you want it and a PC when you need it." All day, every day. We wanted it to be instant on, we wanted it to recognize gesture, voice, and we're well on that journey. One thing has changed. I don't think we recognized how amazing the bow wave would be underneath the Ultrabook spec. Ultrabook is still Intel validated. It goes through robust testing of the dual array microphones that are on the system, the touch ecosystem, the performance, the boot times, all that goes through a test to get the Ultrabook logo. We want that to be the halo for the products of what's possible with our technology.
It delivers Intel Core processor performance, all the latest technologies from Flash and security and manageability. What's also happening is a whole range of value platforms are coming in below it. They may compromise on the amount of non-volatile memory they put in the platform, compromise on some of the security, but we're actually selling more than 50% of our ultra-low voltage parts into the value segment. Same price, same product, same Core family, it's just they're not branded Ultrabook. People are just trying to hit lower price points by compromising a bit on the spec. Our strategy is keep Ultrabook as the premium product and ride this wave where as we went from thin, to touch, to now 2-in-1, to perceptual computing, we drive a whole set of designs below that as well. Let me talk about this next big wave that we call 2-in-1.
As I said before, we had about five designs that met this category in the first quarter. We're on track with our commitment to have 50 in the market in this quarter of 2013 and growing into 2014. Very simply, what we've come to the conclusion on through our research, through talking to our customers and the end users is that tablets more and more are becoming consumption devices. We track every six months what people do with the tablets, what workloads they run, and for now a very consecutive set of surveys, the amount of creation that's being done on a tablet has been going down. The number of people that actually do real creation. It's becoming a great consumption device, especially as the momentum goes from 10 inch to seven and eight inch.
At the same time, the notebook, a number of third-party surveys are saying 97% of people still view that the notebook in a multi-device household is their primary computing device. Because at the end of the day, when you're going to do creation, it's thick. What we're doing with 2-in-1 is really eliminating the need to carry both devices. It's a tablet when you want it and a notebook when you need it. What are our high-level criteria here? First, we think it needs to have greater than or equal to a 10-inch screen size because of the productivity is starting to compromise when you get to a smaller screen size. Can't really call it a notebook per se when it's a very cramped screen.
We think it needs a full PC operating system, which means at least Windows 8 on the system because, again, all that x86 compatibility is very fundamental to being the best of a PC and the best of a tablet. We think it needs an integrated keyboard design, and that can be either wired or wireless, but our user experience data says under 10 inches, the keyboard experience goes down significantly, and people have a very bad reaction to it. We're generally recommending 10 inch and higher. If you just slap a Bluetooth keyboard on a tablet, they tend to be much heavier and much clunkier. In order to be the best of a tablet, it has to be instant on, touch, responsive, and have all-day battery life. A lot of people have asked me, well, is Ultrabook different than a 2-in-1?
At the end of the day, Ultrabooks are the best 2-in-1s, and we'll have value-based products all the way down to $299 and below as we go into 2014. The Ultrabook is the best of 2-in-1. We think now is the time to buy. If you just look at fourth generation Core versus the installed base, you can now get work done about twice as fast as a product that's four years old. You can consume your media and video 13 times faster. Not 10%, not 20%, 13 times faster. You get basically 8x faster uptime as you're booting, and these things are 50% thinner, 50% lighter than a four-year-old PC. I'll show you those here in a second. As I mentioned, or Brian mentioned, we're becoming a market-driven company. I thought I'd share with you data.
You guys, after the investor day, say you like data. Here's why basically every signal we have from Intel surveys to third-party surveys saying this thing hunts. First of all, we just did a Intel market research holiday wish list, and the whole concept of this 2-in-1 is now a 30% higher interest for the holiday season than it was just a year ago. We're seeing interest in the category as more and more investment goes into advertising from Best Buy, from Microsoft, from Intel. We're seeing this as a category that users are seriously considering. We also went and we do device studies. We set up mock stores. We put all the devices out for people, and we watch them and what they do.
In over 900 people in the U.S. and PRC, we decided how many people picked the 2-in-1, and if they hadn't picked a 2-in-1 and walked out of this mock store, what would they have bought instead? It turns out that 48% of the people said they walked into the store expecting to buy a 10-inch tablet and walked out with a 2-in-1. That's very good from our perspective that potential tablet buyers where a significant % of tablets from some vendors are still selling above $499 before the keyboard and the cover. We can take that premium tablet space and upsell to 2-in-1 computers. If you just ask the people, do you think this is an interesting device?
Some people have accused us of trying to put two devices together and having something that's a compromise. 81% of the people when they actually hold these pieces up, and this is U.S., China, and Germany, 846 people, 81% of the people say, "Yes, I find value in that." Okay, 19% of the people say, "No, I just like a slate," or, "I just like a clamshell. I don't need this whole conversion." A significant people are resonating when they actually hold these new devices, and it's just going to get better with Broadwell. Lastly, we absolutely know one of the reasons the PC market has been declining is that it's not so much people don't want a PC that have a tablet. In fact, this last data point says that 82% of tablet users that have one or two tablets still want to refresh their PC.
It kind of refutes PC is dead. The question for all of us is, well, why do I refresh this year? Why not just wait another year? The data right now is astounding. There's 650 million PCs out there that are four years or older and longer now. As those things have elongated, it's really becoming tough to use those class of devices. At the same time, when Best Buy did a study of over 800 people that walked out of the store with 2-in-ones, 66% of those people said that they had a PC four years and older, they got interested in the category, and they bought. A large % of people that are buying at least the convertible are coming over and refreshing that older PC. You see the quote from Walt Mossberg. He's been critical of this segment.
He now believes that for at least the detachable category where the keyboard removes, is something that's getting better and better and less expensive, kind of hitting the mainstream price points. What are we doing out there? Right now about 50 devices that are hitting the market in the fourth quarter. I'd like to show you some of these. The top ones are the Ultrabooks that are meeting the Ultrabook specification. At the bottom, these are 2-in-1s and/or convertibles or detachables that are running the Android operating system. I'll show you a few of those if I can. One of the first true convertibles was the Lenovo Yoga, and this is the Lenovo Yoga 2. I'd say that the traditional convertible category is something where the keyboard's attached, and it doesn't detach mechanically.
It's typically built as a PC first and kind of a tablet for free kind of model. You can see this is now 15% lighter, 15% thinner. It has a stunning Quad HD display, under $1,000 for the system, a backlit keyboard. Again, it's just a real simple conversion mechanism to get it into tablet mode. Lenovo's done a really nice job, and it only gets better now moving this into business that I'll show you in a few minutes. If you look at something that's more in the detachable category, Sony's launched, this is literally a Core i7 machine. Is this a tablet? Is this a PC? This keyboard here is one of those that's mechanically designed to work together.
They have a magnetics here, and it simply just snaps on, and you can use this with the keyboard on your lap, and the kickstand kicks out, and this is a full highest performance Core i7 vPro machine. Consumer or business. Again, stunning performance, and a really nice price point at $799 even for a Core i5 or an i7. If you look at HP Split x2, here's an example of, again, a full notebook computer but again a really thin detachable device from HP. They've got a companion to this that if you've seen it for Black Friday will be down at that $499 price point with lower-end Intel processors. Lastly, the Sony VAIO Flip, kind of a really nice conversion mechanism here that Sony's done. You can sit here, and then you can just literally flip it down. Pretty simple.
You can also use it if you're a real estate agent, and you just want to show somebody the other side of your screen. You could be showing them a house. You're on the other side of the Starbucks screen, you just flip it around, and you can show them the other side of the screen. It's more than just a tablet and a clamshell. It's really the ability to do simple screen sharing as well. Uh-oh. What did I do with my clicker? There we go. Okay. Those are the class of devices we're seeing. At the end of the day, what we're excited about is Broadwell's just going to make this better. We'll commit here on stage again. We're on track for second-half production.
We talked about this at the Intel Developer Forum, where we are literally at the same performance level, we are getting 30% less power. What that means is we are seeing a whole range now of fanless form factors that are going to bring just amazing performance, but in the fanless very dense devices. A whole range of new amazing two-in-ones, socket-compatible upgrades from Haswell. The whole range of these two-in-ones will be very quick to market because you can just drop in the processor into the existing motherboards. The fanless designs will be brand new with our Y-Series processors, significantly greater performance, lighter et cetera. We are super excited that Broadwell, although we made the 3-month delay, we are on track to get products and stunning products into the market in the second half of 2014. It is not all about premium.
One of the things I said, the high end we are doing extremely well in with vPro, i5, i7. We will talk about that. What does the holiday and spring lineup really look like? Ultrabooks, two-in-ones at $799 and above, Core processor-based two-in-ones at $599 and above, two-in-ones around the Bay Trail product line at $349. This is an example of a product, again, from the China tech again based on Bay Trail running Android down at $299. Snaps into a keyboard and becomes a detachable. A product, same product from CZC, I will show you over there running Windows. Another company who is running both Android and Windows on the same set of hardware. Bay Trail detachable is down to $299, then non-touch clamshells down to $199. These aren't just 10-inch netbook-looking devices. These are large 14, 15-inch screen sizes in many cases as well.
We think that, I will show you why we are so excited about Bay Trail in a second because it is really a refresh at the high end with Haswell and a refresh at Bay Trail all coming together literally this quarter then going into spring. Okay. Let me talk about the value in the entry space. As we announced, Silvermont is a microarchitecture, right? What is wonderful about these microarchitectures is they are highly scalable. We literally have products that Diane talked about most recently for Avoton, for lightweight servers and storage and networking, like branch office routers, et cetera. We have Bay Trail-T that Hermann will talk about next for tablets, whether they are 7-inch or 8-inch or 10-inch tablets. A range of two-in-one devices that are using the M and T products.
The M does things like add PCI Express and standard hard disk drives and things that are traditional to a more PC segment. Then we are able to actually scale the wattage up to meet the needs of mainstream towers. We just had some of the largest OEMs in the world this week announce new towers on Bay Trail down at $199. Then Bay Trail-I for embedded, where we do things like high temperature and long life, where you will see these in kiosks, point-of-sale terminals, robots that walk around your house to see if your parents have fallen and can't get up, which I saw in Japan. Just about anything you could imagine on the Internet of Things. Diane talked about Avoton, I will talk about the M and D, then Hermann will talk about Bay Trail-T.
What we're seeing is we basically candidly exited the value space after netbooks. We decided to focus on the high end, we really did not innovate at all. In some cases, it may have been surprising to you, but our market segment share was down to 40 or even 50%, depending on how you count it in below $399 devices, traditional clamshell, traditional desktop, a significant amount of the market, we were anywhere between 40% and 63% share, depending on how you count it relative to our other segments where we're well over 80%. You saw two-in-ones now from CZC at $349. You saw the 295-degree hinge at sub $299. More than 50 of these designs are now coming to market.
A few dozen this quarter, but if you look across the Shenzhen, the Taiwan ecosystem, and all our multinationals, as we get into 2014, we're really going to unleash this value space, and we think that means that we gain market segment share as we go into 2014 against our traditional x86 competition. At the same time, we're also setting ourselves up to participate in Android and Chrome. The fastest-growing parts of the market now are below $499. Ultrabook's been great to hold the high end, keep the innovation going, and be able to waterfall that down. These are the products that are enabling us to win against our traditional competition with AMD, but also with Arm, whether it's a laptop, a two-in-one, a traditional desktop, or an all-in-one.
One of the things I've heard as I've done one-on-one interviews with a lot of you in the investment community is, wow, Bay Trail is stunning, right? The performance is stunning. Good job, Intel. Quad core, significantly higher performance and leadership battery life that Hermann will show you versus our Arm competition. You should be really worried about sell down from Core. What I wanted to do is just show you across the range of benchmarks, what I've done is I've taken our best performing quad core Celeron part, not the dual core, but the highest performing, best Celeron we can deliver, and I'm benchmarking it against the lowest performing, cheapest Core i3, okay? Basically showing you that across a range of benchmarks here, you're getting anywhere from a 2x to a 3x benefit.
Candidly, what keeps me up at night is not the sell down from Core to Bay Trail. It's attacking and winning share against our competition in the value space where we literally had exited post netbook and winning that in all screen sizes. Rather than show you this, I thought I'd just do a quick benchmark. We'll have some things out there as well. Obviously, one of the things we do, guys, is we take photos and we take videos, right? We all know that most of the internet traffic is videos. What I'm going to basically show you here is this is a Core i3 Yoga machine consistent with what I have up here, and I'm benchmarking it against the CZC machine, the same one you had Android over here. I've loaded Windows on that same motherboard.
This is showing the best quad core Bay Trail we have, Celeron. What we're going to do is very simple. You're taking a bunch of photos from your vacation. You're going to compile it into a slideshow, you'll see that that's roughly on the photo editing. It's about 2x. What I'm going to do is do the same thing with videos, we have something called Quick Sync, which is the same thing that we said Apple's using within FaceTime and people like Tencent are using. It's just going to show you how stunning the video compilation is when you move from a Bay Trail product to a Core. Remember what Hermann's going to show you that Bay Trail is higher performance than the Arm solution.
You could also substitute this certainly with Arm. Guys, why don't we click it.
Okay
click it here, hopefully the team can see.
Release that
the delta here
Basically it's going through the collection of photos that we selected, analyzing them, and it's going to create a slideshow based on a theme that we've chosen here. You can see the Core-based system in the upper left hand of the screen and the Bay Trail system in the lower right. It basically goes through the process quite quickly of the analyzing, and now we've actually already started the preview on the Core-based system while the Bay Trail system is still finishing up analyzing and getting ready to prepare for the movie.
It hangs a long time at 95 just because it's not a linear progression. It's basically where the last compilation happens. We've already finished and it's just now starting.
There we go.
Basically about a 2X performance delta. Why don't we do that with video now? Tell us when you're going to start here.
Let's go ahead and switch over to TouchXPRT, this gives us a range of benchmarks across the different variety of applications here. We're going to do some transcoding. We'll go ahead and start, Umesh. Ready, set, go. Now basically what it's doing is it's taking a couple video clips and it's transcoding them into a different format. This is something that's commonly done on a lot of different systems. With Quick Sync, you really get that capability of doing transcoding very quickly. Transcoding is something that you might not even know that you're doing, but when you're copying a video to, say, a tablet or a phone device, you're typically transcoding. When you're uploading to YouTube, oftentimes your application is transcoding. It's something that actually you do quite often.
You can see here in this example, with Quick Sync and the Core-based system, you're getting about 3X performance improvement in doing things like that.
This is basically the worst Core i3 we have against the best Bay Trail. When you do go up to Core i5, you're getting turbo mode, which is, again, we've had record shipments, as I said, on turbo. When you go from i5 to i7, you're increasing the threads, the cores, and the overall performance. As we said, people are upselling to Core i5 and i7 just based on those kind of attributes. What we're showing you is the worst core performance, 2X to 3X on something that we all do, photos and video.
Yep.
Thanks, guys.
See you, man.
Okay. This hopefully clears that up. The other thing that's obviously happening in the market is the other operating systems. We're extremely excited about Chrome. You saw Sundar on stage with us at Intel Developer Forum talking about our collaboration on Chrome devices. Worldwide retail share in Q1 prior to our Haswell Celeron announcement was actually 40%. While we had interesting design wins, the Samsung Exynos Arm product was doing very well at $199. Just since we've announced the Haswell Celeron part, our share's gone to 51% in worldwide retail, and now we've got design wins from Acer, from HP, from Toshiba, from ASUS. You can see the devices here, and that little product there in the lower right is actually called the Chromeb ox, which we're already seeing now replacing some thin clients.
If you're always connected to the internet, you actually just have this small little hockey puck like device, and it's running Chrome as a stationary computing device. All of this market segment share gain we're talking about is before we even introduce the Celeron-based Bay Trail-M products that we're committed to launch in the first half of 2014. A good opportunity for us to go gain share. I thought I'd just show you two of the boxes. This is HP, and again, these things aren't netbook-looking like in any regard, right? Lots of different colors available from HP. This is a 14 inch, and it actually runs at Walmart now with a wireless WAN connection and I think 250 gigabytes of cloud storage at $329.
It's outselling, in some cases, the $249 and even $199 machines because people like the wireless WAN connectivity and the cloud storage. From Acer here, we have something more traditional. Again, down at $199 price point, a pretty nice looking machine, again, running Chrome that'll enable us to compete where we hadn't had a price point at the $199 directly against that Samsung Arm competition. You can see the Laptop Magazine said a Haswell in $249, which is now going to $199, offers significantly better performance versus its low-cost competition. Transitioning to business. This has been a segment where we've had well over 90% market segment share for a while. As I talked about, we have record unit shipments of our vPro technology. If you look at what IT needs, they need low total cost of ownership, same as Diane talked about in servers.
In some cases, there's a set of compliance requirements, something as simple as you need to turn the PC camera off when you go into certain parts of the building because we don't allow in this government location cameras to be on. Things like ease of integration. How can I get these things onto the network? Compatibility, just like Diane said, is a huge asset where if you look at the tablets better together, device security and manageability, things that are inherent in Windows, like putting domain join, common security, common manageability. We've worked with Hermann's team. We have actually now Atom SKUs that are ready for business where we've waterfalled a number of the security and manageability features that traditionally were in our vPro products so that there's commonality now between our Atom tablets and our Core processors.
When we walk into IT, we say, "Hey, you can configure these on the network together, secure them, and manage them consistently." What we're seeing now is that's a huge asset if you want to have a tablet and a vPro device working better together than, for example, an Arm tablet. We're just now getting the first third-party research back on the value of a two-in-one in business versus having an Arm tablet and an Intel notebook. In some cases, we're seeing $1,100 lower total cost of ownership, and we'll make this white paper available to everybody from Principled Technologies. Just the ability to have licenses for your remote application support, your Good license or your MobileIron license, managing two devices, the TCO value of going to a single device is pretty significant. What does that mean?
It means that we feel confident in the future as business stabilizes, that we're extremely well-positioned. As we talked about before, vPro has crossed 100 million units, and it just continues to go up. We're expecting high single-digit CAGRs for vPro, both historically this year and going into next year, and it just keeps continuing. Companies that I've had on stage, like BMW, for their dealers looking at a combination of vPro Ultrabooks and two-in-ones, companies like PepsiCo that are, maybe had deployed tablets before with someone else and are now using Lenovo's Helix detachable for their salespeople to go into the stores, just because they're finding it more productive. By the end of this year, we'll have more than 150 4th Generation Core vPro machines. We'll have more than 20 Ultrabooks on vPro, and we'll have more than five two-in-ones that are in the market for IT now.
Beyond that, we're taking vPro from not just, hey, it's security and manageability, which we've been known for, but now we're adding to the experience, not just for the IT manager, but for the end user, all of us that might walk these things around. We're now working on, for IT, Pro SSDs. Instead of having to go walk around your cubicles when someone forgets their drive encryption, you can go down the wire and reset someone's SSD password. You can encrypt now what used to take hours and hours and hours in literally a few minutes for IT with the new Pro SSD. Pro WiDi, where you can eliminate the 10 minutes in the middle of every meeting where everyone's looking for the right dongle, the thing to plug in the thing to get to the projector.
As we go into 2014, you'll be able to basically just swipe off your notebook and your system will automatically project up on the wall. Location-based services. We worked with a company, Stanley AeroScout. We're basically now integrating a $70-$80 RFID tag directly into the wireless of vPro. You can do things like for Intel's most confidential documents, when you walk out of our building, it's locked from getting access to that document. I walk into a new building and I want to find the local network printer, it'll automatically connect me to the printer that's closest to me. It'll find out if there's people in the conference room because this thing is pinging up to Cisco and Aruba's Wi-Fi access points. A significant value here and a huge differentiation opportunity for people to take location-based services.
You can do things like, hey, when you enter this part of the fab, disable the cameras on your tablets or your notebooks. We're also doing things to eliminate passwords. Our vision has been, the average U.S. person has 18 passwords, basically eliminate passwords. We're already doing a pilot with Cisco where you never have to log into your VPN password again with our identity protection technology. Lots of goodness here as we go through and the poll, we think, is continuing to be very strong and the appeal for two-in-ones very strong here as well. If you look at this, I love this quote from Forbes.
It says, "End users may literally hug their IT guy with the two-in-ones because it has the security and management that IT wants, but it's a great device." The best way to think about this is, this is what people are typically carrying around Intel, right? You have to put the extended battery in the back, and you can get a feel for what an average person is kind of walking around with in IT today. You're going to look at this Lenovo ThinkPad product, and it's just completely a different experience for the end user, right? They can get the tablet and a notebook, but it's something that's much, much easier. You no longer have to compromise between what the IT manager wants and what the end user wants. Let's see here. This is the Dell [inaudible] platform.
Again, a really nice fold-over design, things like this for business. Detachables from Fujitsu and products from Dell as well that are going to go into this market. This is the Dell detachable. Imagine a Core i7 vPro machine that is every bit a tablet, but the highest performance i7, again, in just a very simple detachable device. Lots of innovation there. In addition, things that traditionally had been huge tower desktops are now ending up in these very small devices. This is the Lenovo Tiny, just absolutely blowing away Lenovo's expectations because you can literally put a Core i7 vPro, all the legacy DVD drives, everything into something that's this small. Okay, desktop. If you look across desktop, we're segmenting this as well. Small form factor, like the thing I just talked about, double-digit growths within the traditional desktop category.
If you look at all-in-ones, more than 15% growth. You see the figure here from IDC as we look at the forward and the backward CAGR of the all-in-one, where you're basically putting the CPU behind the screen. The reason we like this is it's moving the desktop out from underneath the office at home where it was collecting dust for five years and it's putting it into areas like the kitchen. The picture here is actually like a typical hotel room now. If you go over to the Marriott half a mile away from here, all of their front desk clerks are using touch screen all-in-ones behind the desk. If you actually go out around the corner and look in the side of the lobby, all the all-in-ones are there to have you book your hotel reservation and your tickets for the airport.
The density that we're seeing in desktop now is helping density in both consumer and in business. In value, again, the Bay Trail-D part going into products as low as $199. In the enthusiast space, this high-end, overclocker space, we're doing incredibly well there. In high-end gaming with record core volumes, and I'll talk about what we're doing in graphics there in a second. What's interesting is even the desktop is going mobile. You traditionally had this tower, then you had this all-in-one that was interesting. We started getting those to lay flat, now they're becoming portable, and we're moving into a space where you get the whole family together. You can move these things around in the house. It turns out a large percentage of desktops do move around the house more than you would think.
It's all about getting multi-user, multi-touch around a common screen. On the top here, these are new products that are stationary. At the bottom, these are all the new products that are portable. The best way to look at this is here's the new Dell XPS 18, full-size desktop, keyboard, and mouse has it here, but with literally just a simple tug, you're now at 4.7 pounds and five hours of battery life. You can actually take this out. It has legs on it, so you could literally just sit it in your kitchen like this. Again, it's the full experience.
You can put this down flat and collaborate with multi-user, multi-touch, play Monopoly on it, and we're working with a number of people in Microsoft to get a whole set of your traditional family board games and things onto here so you can get the whole family around this class of device. All that says is the desktop is no longer a tower. The fastest-growing segments are high-end enthusiasts, all-in-one, portable all-in-one, et cetera. Okay, last but not least, experiences. First and foremost, touch. As I said, about 40% of large format retail on Windows 8.1 is touch, and more than 50% of our fourth generation Core design wins are touch. Costs about in the $40s. I know you guys asked that. We think it'll get down to the $30s for next year. Doing extremely well in 10- and 11-inch form factors.
As you get up to 15-inch and above, the costs have been higher and the penetration's a little lower, but we're continuing to see the trend, which is quite strong. On voice, as you've seen the Best Buy flyers lately, you've seen there's a voice icon on there. Best Buy is significantly increasing their presence on their premium collection on the Intel Voice Assistant that we're doing in collaboration with Nuance. We now are delivering voice in seven languages across 25 countries. We have nine multinational shipping versus the two or three we had just six months ago. We have over 100 platforms validated with voice. Now is currently shipping seven multinational companies with voice solutions.
Think about it as doing everything from asking your PC what you traditionally have been able to do on a phone, how many quarts in a gallon, who's the President and CEO of whatever company, are there any good movies playing tonight? And over time, we'll integrate with Fandango, Wikipedia, OpenTable. Every major website you know, you ask it the question, naturally speaking to your PC, it'll come back with their answer through the various websites. On 3D depth camera, essentially the world is in a transition from 2D to 3D, and we plan to lead that. A lot of people don't realize, but we've purchased multiple camera companies in the last couple of years, as Brian said. Traditionally, you've had a big camera in a gaming console that you plugged into the wall because of the power.
Today, you can buy these at under $200 around U.S. retail through our collaboration with Creative, where you're basically plugging this into a USB port. But what we've basically announced at our Developer Forum is we've, for the last year plus, been working on integrating the camera into the bezel of all-in-ones, of 2-in-1s, and as Brian said, this is a great example where PC technology will actually drive down into tablets as well. And we're doing things like augmented reality, where you open a storybook and the characters come to life and build up in front of you, blogging in the Internet, a whole range of usage models that, by the way, we'll have available a number of the demos out there around 3D so you can see it firsthand where we're at.
As you get into 2014, I think this will be one of the game-changing things. You've seen some of our partners announce their own acquisitions of 3D companies as well. I think we feel pretty confident we have the world's best 3D cameras and solutions coming to the market. We have OEM commitments now from HP, Dell, ASUS, and Lenovo to bring these products to market in 2014. No passwords, very simple. 18 passwords for the average person in the U.S. We're going to eliminate passwords with a collaboration between Intel and McAfee. As we get to 2014 and into 2015, it's quite possible that the only thing you'll need is your voice and your face to log into everything in the world that you want to get access to. I don't know about you, but resetting my password has just been a complete nightmare.
No wires. We're basically going to eliminate every wire off of computing. We're a board member of A4WP for wireless charging. We are committed that the way you charge your notebook, your phone, your tablet, your two-in-one is all in a consistent way. We're committed to wireless display so we can eliminate all those cables going in to project in business. We're committed to wireless docking with WiGig technology. We'll be bringing WiGig solutions to the market next year. We're committed to wireless data sharing, where you can just seamlessly walk into a room and share data between devices. Battery life, as we said, has been absolutely amazing. We essentially doubled the battery life on a Mac Air from six to 12 and a half hours in collaboration with Apple. Quad HD and higher-end displays coming across the broad range of product lines.
Last but not least, Iris graphics. Let me just talk about sell-up opportunity on Iris Real quick. We've been investing in processor graphics for some time. Some people haven't realized that Intel is the largest graphics company in the world. If you look at the commitment, we've met all our commitments on performance. We're now basically have increased graphics performance 75 times. You can just see the huge generational bump that we get. Thank you, Bill. I have to thank Bill at least once for Moore's Law. As you shrink those transistors, it's highly valued to the number of execution engines you can put in there. The benefit of integrated processor graphics is lower platform power, more and more aggressive form factors, et cetera. It only gets better from here.
We were extremely pleased that Apple has moved a significant part of their iMac and their MacBook Pro line to Intel Iris Pro graphics. Sometimes a harsh critic on AnandTech says this was the biggest endorsement of integrated graphics ever. We're seeing a $20-$30, if you look at our list pricing, upsell generation on generation to our ASPs. Today, we believe that we can go after about somewhere around more than 75% of the discrete graphics market, regardless of the competitor with the performance. This brand is also very important because in countries like Germany and China, it's not been so much the technology as the brand of the Nvidias and the ATIs that has been pulling through their products. We now have a brand to go after this. Obviously, Apple's pushing that hard and believes in what we've got here.
We're pretty excited, and the conversion to 14 nanometers is just going to make this even better. There were some leaks in the press today talking about 80% better graphics performance in Broadwell. I don't respond to leaks in the press, but you can just see how people are excited about how much graphics potential could be coming. In summary, PC market, we believe small single-digit declines, operating margin will be better, and Stacy will talk about that directly based on the efficiencies we're driving. Studies reaffirming the end-user desire for both PC as their primary computing device as well as the two-in-one category. The client segmentation is helping us attack the bottom, whether it's with Chrome or Windows against our traditional AMD as well as Arm competitors with Bay Trail-M and Bay Trail-D.
While at the high end, we're hitting records still on unit volumes, not just on mix. Lastly, amazing new form factors coming, a commitment to operating system choice, business leadership by expanding the vPro franchise we have, and then a whole set of experiences from touch to voice to perceptual computing and 3D. Okay. Thank you. I'll be back up for Q&A, and it's my pleasure to introduce Hermann Eul, who's Corporate Vice President of our Mobile and Communications Group, the other half of the client.
Thank you so much. Kirk did a little bit of training how to pronounce my name. Since I'm in the U.S. here for two years, I do also training on English idioms, and the devil catches the hindmost. We are late on schedule, don't worry, I will continue, and it will just eat into your lunchtime. I would like to pick up one thing from Brian this morning. At 2:00, Brian came up with a late change to his presentation, I do this all the time. People like this very much when I come with a late change, here's my latest change. There's my latest change. What do I want? I want to have a little bit of time to lead you through a few considerations before we go into the material.
One thing is we came a long way, we made great progress. We have busted all the myths around what Intel can do, what Intel can do in the mobile space, that this architecture is not going to do this, the power consumption, all of this, we have busted it. We brought performance to this market, we brought graphics to this market, we cranked all of this up, we brought also LTE to this market. We came a long way, we made great progress. Just when I walked down here, somebody came, said to me, "You are far behind." I didn't say that, or I already have some space for this. Yes, of course. We would like to do better. We would like to be faster.
This market is going ultra-fast, competition is not standing still while we are catching up or waiting for us to catch up. Of course, we recognize that we don't have an integrated product as we speak today. Of course, we recognize that some of our products came late to the market. With this coming late to the market, they were intended to be a performance device, but they didn't hit the performance segment anymore. With this, of course, that hits me on the ASP, and because they were designed so rich, that hits me also on the bill of material because they go into different market segments. We recognize this. Also it took us a few months longer than we promised on when LTE is shipping. Yes, we recognize this. We have to get faster. It's not a reason for a doomsday.
It's a reason to just keep our sleeves rolled up and continue plugging away. We have all the assets we need. We have all the skills that we need for this market. With this, we continue plugging away, keeping up our sleeves, and just do work. What I will talk about this work, of course, how we accelerate our mobile roadmap, how we accelerate our capabilities, how competitive our platforms are today, and how we are growing towards the leadership. How we focus on winning share in mobile. Brian already gave a lead to this, and how we are moving this ecosystem onto our platforms. I will be speaking about this. Why are we in this market? Because this market is a growth segment. It's important for us, we are here in this market to stay and to make a difference. This market is fast-changing.
Within the next 3 years, the entry segment, the value segment, will take more than 50% of the smartphones. While we speak, it is changing fast. Also the air interface technologies, within 3 years from now, about 50% will be all on LTE. LTE is going to be a mass product, introduction of an air interface technology faster than ever before. Having said this, why is this all? Because that market is driven by innovation. That started on phones, now tablets are accelerating it. People have multiple devices, the replacement rate is as low as one and a half years. New channels come. Operators are working on innovations on data plans in order to embrace tablets more. With the advent and the increase of tablets in this market, also, the distribution channels change. Open retail is coming more and more.
Non-subsidized models are coming more and more. All those so-believed holes in that market, it must be subsidized and blah. This is going away. This market is always changing as we speak. Applications are growing. Number of applications are growing. They are exploding. With more and more and more tablets, what you can do on an app is more and more rich. The complexity is increasing, this asks for compute power. This is all good for us. Amazing uses are coming that we have not thought about. A $50 accessory to a tablet can turn a tablet into an ultrasonic machine. Maybe not of that quality today what that lady needs on this slide here, but this is just the beginning. People already do feature shots and produce on a tablet high-quality movies that can stand up to professional movies.
A good reason to be there. Now let me explain to you on how our strategy is to participate and to push this market. One thing I would like to pick up from Brian mentioned market-oriented pragmatism. We are recognizing that we have a large product portfolio in the feature phone segment, that has consequences. Having said this, I want to remind you, these early touch devices are built on a single chip, which is our product portfolio. You may recognize and remember we have the team that invented the first single-chip device, the single-chip phone. Put together all the integration of RF and baseband, adding the memory to it and adding the power management to it, just bring these devices into the market.
Today, as rich as low-end smartphones, feature phones like this Asha one, they run on such a highly integrated single chip. We bring a lot of innovation into this market, with this, we span from those feature phones over smartphones into tablets. Different level of integration, of course, different level of performance. Not everything is available everywhere, but we have the complete bandwidth, we are shipping more than 360 million platforms a year. We are an incumbent in this market. Moving on this to the Intel architecture, this brings more compute power to this market, this brings an even longer battery life to this market. We have spoken about that before. You heard that from Kirk, you heard that from Diane. Most efficient core in this industry. Our new products, Bay Trail, Merrifield, are all based on Silvermont.
This is a core which can be extremely frugal on power consumption, can scale up to the performance much beyond that, what I'm interested in in a tablet. You heard Diane speaking about the deployment even in servers. With this flexible architecture, we can span through all those markets. Of course, they are built on Intel's industry-leading process technology. Thank you, Bill. You have seen me skipping one on that slide, that was 64-bit. Yes, all of them are 64-bit. Those products that come to the market we are shipping today, start shipping next year, all consecutive products will be 64-bit. We have 64-bit. I will come to this more later. The strategy, as already teased up by Brian, we are boldly increasing the focus on winning in tablets. Winning share with all key OEMs and in all segments.
At the same time, we refine the focus on the phones, we are working with leading volume-targeted customers in that segment as well. How do we do this? We accelerate our performance. We accelerate our integration. We embrace more and more Android. We are growing with the volume share of Android, we go after aggressive cost reduction. I mentioned that at the beginning, we double down on LTE. You will see this later. These are the measures that we deploy into this market with the energy, the maximum energy we can bring behind this. That leads me to the 4X campaign. Brian mentioned that this morning, we are going to drive the 4X growth in this year. That this can be a reality, you see at the right point, a part of the slides. These are the different segments.
We are predominantly placed in the premium segment, but we work hard with our customers to find ways to bring this into mainstream and value and even into the entry. We deploy our complete product portfolio, the 2012 Medfield and Lexington generations, the 2013 Clover Trail generation. The Bay Trail, which is currently ramping. We are currently seeing more than 30 designs in our customers that they bring to the market on Bay Trail. Merrifield will also contribute this in next year's product. This is how we do that. We work hard with our customers. You will see us doing marketing campaigns, and you will see us expanding the markets and putting all efforts behind that to make our customers successful on those platforms. Let me come to one of the working horses. Bay Trail.
You heard that multiple times before. Bay Trail is one of our big working horses for this. It is 64-bit. I mentioned that before. It has performance on demand. It is very well playing on great imaging. It can run Windows and Android. It has more than 10 hours of battery life, and we get good remarks for it. Fastest CPU performance out of any Android tablet on AnTuTu. Or such a prize-winning device like the T100, all based on Bay Trail. It's a competitive product, and will be one of the working horses for this. Having said this, more details. Today, we already have best experience on very popular usages: browsing, media editing, tablet gaming. In the second half of next year, we will augment that with a lot of experiences out of the perceptual computing area. Voice, visuals, 3D will come to it.
If I take those vectors and compile them in a different way, if I go through the operating systems, today, we have best in modern of Windows, and our architecture is the only that can play the legacy and all the commercial programs that are available for Windows. It can play full PC games, and it has a wide set of security functionality. Kirk already mentioned that. I can save some time on this. Moving this to Android, next quarter one next year, you will see Android devices coming into the shelves based on Bay Trail, and we use that to scale into a market that we are currently not tapping with the Bay Trail architecture. These are the steps.
If I put another step to this, having said the last two parts, we have a great experience for Windows on it, and we bring good experience on Android on this as well. That means we have two legs that we can give our customers. We are in a perfect position to help our customers do cross-operating systems optimizations. This is something that nobody else can do in this fashion. We have customers asking for this, and of course, we support what it needs to make those products successful in the market. That is what I mean with expanding markets. This can create new markets that have not been there before. All of this comes with the goodness of the Intel brand. Let me come to experience. This slide you have to read from right to left.
TouchXPRT, Kirk already mentioned that before. Here's a comparison of our devices against competition on how we perform. That is a Windows comparison. You see that already that what we are shipping today, last year's product, the Clover Trail, outperformed the competition here. You may argue, okay, Intel has always been good on Windows, and with Bay Trail, we just multiply this by a factor of 2.5. It's Windows. Let's turn the focus point towards Android. Similar test environment, MobileMark. Of course, you see our last year's product, what we are shipping today, the Clover Trail, has been overtaken by some competitor's product. With Bay Trail, we again set a new benchmark here, outperforming that what is out there from the competition. This is the way on how we want to drive this market. Now go back to the right one.
This is still 32-bit. That brings me to, what does that actually mean if we bring 64-bit into the game? Here's a comparison, and this is a real look into the reality. For the market, it's future, but for us, it is reality. We have 32-bit Windows and we have 64-bit Windows 8.1. That is a direct comparison out of the box on popular usages on what can be done on a 64-bit Windows compared to a 32-bit Windows. For the photo editing, 20% faster. Podcast publishing, 44% faster. That is what can be done. Cameron, can you help me and run your demo?
I will. Thanks, Hermann.
Thank you.
You're talking about the performance of our Bay Trail platform. What I wanted to show everybody today was here we have two Bay Trail tablets. This one is actually running Windows 8.1 64-bit, and this tablet right here is running Windows 8.1 32-bit. Identical hardware. Nothing's different on these devices. The only thing that's different is obviously the 64- and 32-bit. Let me go ahead and I'll start the demo. If I can do it. As you can see, the 64-bit on the top is running through it. The 32-bit is starting to run. We're both running Adobe Photoshop here, and what we're doing here is we're adding a special effect. As you can see, the 64-bit is running quite a bit faster.
This one didn't run at all.
This one did not run. The important thing though is 64-bit went faster.
It's really slow.
Yeah. It's really slow.
Yeah. No.
If you want to see, I will show that later. It's important to know that we have not made any optimizations on Bay Trail to get it to run faster. You're seeing this kind of performance, and what you'll see is up to 30% performance difference from 64-bit to 32-bit on these types of applications.
That is wonderful. Thank you.
It's not only about Windows 64-bit. You've been talking about Android as well. What I'd like to show everybody today is this is the first ever showing of a 64-bit kernel running on a Bay Trail with Android.
That is a Wow.
It's pretty snappy.
Yeah.
It's a great little system.
Wonderful. That is a 64-bit kernel.
Yes
running under an Android as we have it today.
Absolutely.
Right away.
Right away.
Super.
Great. Thank you.
Thank you so much. That is really great. Actually, you could not expect anything else from Intel. We have great experience in 64-bit. We are doing this for generations. The Silvermont core, which is in Bay Trail, is running on Diane's servers. We have that there. We have it next quarter shipping on the Windows part. Needless to say, we will run fast to make this happen on the Android world as well. Now let me turn the page towards LTE. LTE is shipping. It's in this device here. It's a very wonderful device. I use it every day. I thought I have left it back in my hotel room yesterday when I rushed out of the hotel room to catch a customer meeting, but it is still there. Otherwise, I would have bought a new one, so don't worry.
You can subtract one piece out of the statistics on how many we are shipping. I have mine still. We are shipping this. That is a multi-mode data and voice LTE. Of course, to bring clarity into this, the bottom line part of it, VoLTE is not shipping today, so that you don't get confused. We have VoLTE on it. Our customers have the VoLTE software. We are currently working with our customers, with the networks, and with the network providers to get VoLTE running in the system. We have VoLTE as promised, and when our customers find that VoLTE is in the market and that VoLTE is a market and the networks are stable enough, it will ship. When that will be, I assume it will be quarter one, this is not up to us anymore.
That is up to our customers, what kind of plans they make when they want to introduce it into the market. It is the smallest solution, lowest footprint on the PCB, and it is very, very good in power consumption. Here are a few statistics on how good it is in power consumption compared to competitor solutions. These are real products. That is even drawn off this device here and another shipping device from another company. This is real MOS mean opinion score based data, this comparison. It is a global solution. It can do 15 bands out of one SKU. With this, we can build a real global solution. We built a leadership product. Going into the future, what is coming on leadership products? First of all, I want to give you a complete scope towards 2016.
Within the next three years, our plan is to crank up the CPU performance by a factor of five, graphics by a factor of 15, media by a factor of seven, imaging by a factor of seven, and display by a factor of seven. That is what is to be expected over the horizon of the next three years. Now I take you a little bit deeper into real products. For tablets and phones, in the performance and mainstream segment, we have Merrifield. We are introducing Merrifield. It has 1.7 times the performance compared to the previous generation. It has two times the performance on graphics. It has a longer battery lifetime, and it comes with an advanced sensor hub. Product based on Silvermont core, 64-bit, not to forget. Next step will be the introduction of the 72 generation of LTE.
For most people, it's just AKA carrier aggregation. That's what most people think about this. It adds not only carrier aggregation, it adds TD-LTE, and it adds TD-SCDMA. I just come from a customer meeting. Returned yesterday night. Some people asked me this morning, "How are you doing on your jet lag?" I said, "I don't know yet." Let's see. I thought I'd go to this customer and give him the present that he was asking for, pull in the schedule by two months. What did he send me back with? I want another month. I thought I returned with an empty tornister.
I returned with a full tornister, and I have to go back to Aicha and say, "That customer was not even satisfied with the two months he was asking for, is now another month asking for." That is how our customers like that product. TD-SCDMA, TD-LTE, and FDD, of course. You can play with it. It is in the demo room outside. Also, TD-LTE is running on it, go and put your fingers on it. I can only tell you, customers are asking me day by day to pull the schedules in to get this product earlier into their hands. It has faster capabilities on the networks. That is the reason why customers want to have that.
It can go up to Cat6, 300 megabits per second, we push the envelope on the bands a little bit more, 17 FDD bands, and we add five TDD bands. It is again, a real global SKU. Moving into the second half of the year, talking about a quad-core sister product from Merrifield. We call that Moorefield. It has two times the CPU performance, easy to calculate from two to four cores. It has extended battery lifetime. We add more graphics capabilities, and we, again, bring more security to this product as well. The second half. This is not the end of the second half of the year. Towards the tail end of the year, we will introduce our next generation for the high segment of the market. We call it Cherry Trail.
It's next-generation core of Atom, the Airmont core, and it's the first on 14-nanometer. It has Tri-Gate transistors, everything of the newest technology, what you can imagine. Of course, it comes also with the next generation of graphics. Here is something where I have to add a few more words. I got some questions in the break already, I take a little bit of time on where does that come from and what is the history and what is the feature set of this one. Where does that come from? We have products that ship in feature phones. For example, feature phones like this or feature phones like that. Next generation of this will be so highly integrated that we believe it will be the highest integration which is in the market.
One RF chip that has Wi-Fi in it, Bluetooth in it, FM radio in it, of course, 3G RF in it, GPS in it. Absolute high integration. This is a feature phone product. In order to make this a smartphone product, we need a real application processor. Those feature phone products, for those who are familiar with this, we call them usually application enhanced modems. They don't have a particular application processor. They just run all those apps on the headroom that is available in the communication processor on it. For going into a smartphone, we need to have an apps processor. That's what we are doing. We bring the Intel richness of the application processing, the Silvermont, into this generation. That is the secret sauce. It comes with all the rock solid 3G that you know.
The 100 and more operators that are familiar over three, four years of that technology that is shipping in hundreds of millions. It comes with this as a backbone. We add the Intel Architecture to this to make a stunning entry smartphone out of that. Remember what I said before, all our future products will have 64-bit. That is the history of this product. Next year, 2015, Broxton, that is the high-end product. It will mark leadership in the performance for Hero devices. It will have our next generation of graphics. It will be in 14-nanometer, and it will be converging our tablet and our high-end phone roadmap. That brings a lot of efficiency to my R&D because I only have to work off one architecture. Today, I have to maintain the two of them.
It will be a cutting-edge product, we are just about today speaking with customers on what kind of features they want to have in this particular so that we nail it exactly to what the market needs. Market-oriented pragmatism, that is behind this. Then it comes to the low end of the market, integrated LTE. I got a question of what LTE is in it. I can go push the envelope a little bit beyond of what I was supposed to say. Yes, of course, it will draft off the 72 generation. A few of you were asking, is that a 71? And whatso. It will draft off the 72 generation. This doesn't mean that it comes exactly with the same feature set. There is one secret sauce in building entry devices, and that is as simple as fight feature creep.
Every feature creep pushes it just up in cost and pushes it out of the market segment. Don't take the 72 spec and believe that this comes here. We will do, again, what we have always done in that market segment, a very subtle, to-the-point feature selection, that it hits the market segment and the cost, what it needs to have in that market segment. I dragged you through some details here, and I would like to sum it up and say, we believe we have all assets. We believe we have to continue to keep our sleeves rolled up and work hard with those assets. We have 64-bit architecture today, not a future, not an announcement. We have that today. We are pushing the envelope and the gas pedal on the integration. We always look at it from full platform view.
We have leading manufacturing technology, we are pulling in differentiated user experiences. This comes all with the promises that you can assume under an Intel brand. With this, let me level this up again. You have seen, we are accelerating our roadmap in mobile. We are accelerating our capabilities. You have seen how competitive our platforms are already today and how we think our way to what leadership will look like, and how we think we will be focusing on winning share. Winning share, that's the reason why we do this. In 2015, we will have an absolute leading performance part and very competitive integrated solutions for the low end of that market, 3G and LTE.
We want to have a significant market segment share when those products hit the market, that they jump on a running bandwagon of a market segment share and an Intel architecture presence in that market when they come. That is the reason for all this heavy lifting, what we do to plow us into the market this year. Thank you so much.
Thanks. Ed Snyder, Charter Equity Research. Bill, one question for you. Obviously, you're running to a little bit more of a challenge moving down Moore's Law, 14 nanometers you've shown here. Sooner or later, the party ends, however long that takes remains to be seen. How does that manifest itself? Does the tick-tock cycle stretch out to three or four or five years? Does it get to a point where we just stop? What kind of lead do you think, and I'm sure you think about this quite a bit, what kind of lead do you think you'll have over your competitors when and if that should occur, say, in the next 10 years? Thanks.
Well, let me start to say that I'm not about to start predicting the end, since anybody who has tried has been wrong. I'm not going to try that. The other thing I would refer back to is, Craig many years ago said, when asked this kind of a question, that, yeah, there is a wall out there somewhere, potentially, and he was going to run into it as fast as he could. We have no intention of slowing down. If we slow down, it will just be because we cannot keep up. We will see. The goal is to keep pushing that wall out, and that is what we are doing right now. As far as hitting it, we are not going to slow down because we see it on the horizon.
Great. Let us move over here.
Thank you. Vivek Arya from Bank of America Merrill Lynch. Thanks a lot for the presentations yesterday, today, I mean, very insightful.
Do not rush me.
It's been a long day.
My question is, traditionally where Intel has succeeded is where you have had a very dominant share of the market, right? 50, 60, 70% market share. If you look at mobile, do you think you can get to that 50, 60, 70% market share as an IDM, or do you think it's better to look at a foundry, or do you think it's going to take some combination to get there? First, is 50, 60, 70% a reasonable goal to expect over the X number of years? The path to get there, is it determined as an IDM or is it built as a foundry? Thank you.
I think that is something that I have to take. We have not given any of those projections. What I think is new today is you see us coming up with a very market-oriented pragmatism. What counts in the end is what gives you the value that you can expect out of our business. That is our leading mantra behind this. Yes, of course, we believe in our manufacturing. Yes, of course, we believe in our architecture, so we will be deploying that. We will stay very pragmatic in the way on how we do this. You have seen that today, and Brian explained that this morning on why we do this and how we do this.
I think the other thing is, if you think back to where we've been successful, when I joined Intel, we were using SPARC workstations to design our processors. We moved to 386 workstations. We had zero share, now we have 98% share. When I moved on to Compaq in the field, we basically had 0% share of the server market. We had 0% of storage. We're now, as Diane said, going up the networking and the comms. I think we have a track record of driving in a new market, not just being successful where we had the share you mentioned.
All right. Over here on the left.
This on? Stacy Rasgon, Sanford C. Bernstein. Thanks for taking my question. Around PC demands, it sounds like things are stabilizing a bit in mature markets, in enterprise. Sounds like emerging markets are still the open question, we didn't get an answer on where you thought that might be going. I think part of the reason emerging markets have been a little weaker is there are alternative form factors, particularly at the kind of price points that you're really looking to enable. How do you guys think about what you're doing in the entry-level and value part of the tablet market in particular? What impact that actually might have on emerging market PC demand? Does it matter because if the market's going there, that's simply where you need to be anyways, but how do you think about making those trade-offs?
Yes, of course. We go after all of those segments. On that one transparency, I went through all the segments that we have, we even stretch it below $100. We work hard with our customers to help them be competitive, go and scrutinize every single piece on the bill of material, every single piece that can somehow be sourced cheaper or whatsoever, optimized. We are very aggressively helping our customers to make a success in that market, which also can mean that we do marketing, that we go jointly to the market, help them with marketing, and so on and so forth. You can assume that we will be very aggressive to address all those market segments, knowingly that not all those market segments today we can address with the perfect product.
Anything to add there, Kirk?
Yeah, I would just add that I feel very good that if you look at Shenzhen as a growing ecosystem, even in the Brazil ecosystem, we're ahead of the curve on their desire to get into traditional PC form factors, whether it's clamshells or towers. A year ago, if you said we were up here showing CZC, THD, Lengda, Jumper, these are names that maybe many of you haven't heard, but they're also some of the people that have been the fastest-growing tablet customers. We're ahead of the curve, and one of the highest value propositions they see is this ability to run Windows and Android on a common device. Pretty much everything there that's coming out of that Shenzhen ecosystem values that IA compatibility.
I think we're ahead of the curve, and the price points we're hitting, we think, whether it's on Windows, Android, Chrome, or even Ubuntu for emerging markets, will be in the price points where that volume is moving.
Great. Let's go to the back here.
[inaudible] from Citi. Bill, the question is for you. In your presentation, you showed us some specs on 10 nanometer, but you didn't make any mention of material science changes. I guess my question is, one, is it implicit in the numbers you provided us that there are no material science changes? Sort of attached to that is how important do you view material science technology in terms of progressing where Intel can go with Moore's Law?
Well, as we've said for a number of years, simple scaling where you just change geometries and didn't have to change materials ended at 130 nm for all practical purposes. Every generation includes in it many material changes, whether they're small ones or they're massive. Obviously, there will be material changes. As far as predicting, what I think you're asking, I'm putting down numbers is what we think we're going to do, but I'm not implying whether there are or are not significant changes.
All right. Let's go right to here to Mark.
Thanks for the presentations. Perhaps this is a question for Bill. You showed some very impressive charts about how you guys are getting a material die size advantage over your competitors. Does it make sense to use that advantage or transfer that to your fabless competitors? For example, maybe a specific example would be, would you manufacture an Arm-based processor on your leading edge to compete head-to-head with x86 processors that you're trying to put into tablets and smartphones?
Well, I think you have to answer that on a case-by-case basis. Brian certainly gave an indication that we are open to building parts for our customers in the way they see most beneficial. Paul also was pretty consistent in saying that we'd rather get paid twice than once. I think we're certainly going to do everything we can to maintain our internal competitiveness, but we are obviously opening up our foundry, and we'll be open to proposals. If we can make good money at it, we'll probably consider it.
Thank you.
Great. All right, here.
Timothy Arcuri, Cowen. Bill, does the consolidation of your suppliers, the proposed merger, and there's been a bunch of things that have happened recently, does that help you stay on Moore's Law, or does that hurt your ability to stay on Moore's Law? Thanks.
Well, I don't think anybody ever likes to see suppliers consolidate. Certainly, in the limit, it's not good for any of us to have a single supplier. We don't look at that as a positive. The industry is what it is, and we'll just have to watch and see how that works out.
Let's go over here to the left side of the auditorium.
Thank you. David Wong, Wells Fargo. SoFIA, in 2014, what line width will that be on? In 2015, just to clarify, is it internal or externally foundry, the 2015 version, and what line width will that be?
The product in the lower part, which we call SoFIA, those two products that you have seen will be out of the foundry landscape.
All right. Just one more over here. Ruby?
He said 2014.
He said 2014.
2014 versus 2015.
John Pitzer, Credit Suisse. A question for Kirk, just around some of the characterization of PC demand stabilizing down low single digits. Is that for the outlook of how you're operating the business for all of 2014? Is that both a unit and a mix comment? When you think about the enterprise stabilizing near term, can you help us understand how much of that might be driven by the end of life of Windows XP in April, and what happens post-April to the corporate demand?
Yeah. I think relative to high-end demand in general, I think we still see very strong Core i5. In fact, I was just with one of the world's largest retailers, and they are selling out a Core i7 right now of their two-in-one SKUs. They're driving our OEMs hard, even in Core i7. I think absolute unit volume, I feel good about a Core i5 and Core i7. The mix, in general, if you just look as a percentage, as we go aggressively into value, I think you'll see the blended ASP go down because we're growing share with good margins with Bay Trail-M and Bay Trail-D. As both Diane and I said, that doesn't mean they're bad margins.
It just means now we have a cost-optimized chip to go after that space and gain share against Arm, where I'm confident we'll do that in Chrome and against AMD and x86. Relative to business, I think vPro will continue to go up. What was your business question again?
Windows XP, Microsoft.
Oh. Yeah, no, I think everything appears to be on track. I'm not predicting a large spike either this quarter or Q1, unless we're surprised, but I think people are on track to migrate, and the April deadline is fast approaching. I think small businesses are behind large businesses in the transition, but we know when you do an OS flop, that it typically comes with a hardware transition. I think you saw a nice linear line, and that's what we would expect, not any major spike.
Great. Let's go here to the left middle row.
Thanks, guys. There we go. I had a really provocative question I planned on asking, BK answered it in the first three minutes. Unfortunately, good planning, and I'm relegated to something a little more mundane. Diane, you had just one slide on Arm servers. What sort of % of the market do you expect Arm servers to take up? Do you see it as a significant threat or not? Why or why not?
Yeah. Today, as I mentioned, the portion of our volume that is going into the microserver space, which is that low-end, light workload space, is less than 1% of our total volume. There is yet to be a production solution in Arm for the servers. We actually thought it was going to be early this year. It now looks like it'll be early next year. It continues to move out. We take, obviously, all competition very seriously, but we do believe we have a very compelling roadmap with the Atom SoC line and then coming with the Xeon SoC line. We believe we have leadership performance at very attractive costs, as Kirk was just saying, so we're able to really cover that light workload segment.
If you look at what Arm Holdings is saying about their penetration of the data center, as they talk about it, their most recent earnings call, it had 90% of their data center penetration in 2017. 90% was in the networking space, not in servers. When they say data center, and then when they put the numbers behind it's really the networking space. If you think about it, the networking space is where we're 5% share, where alternative architectures are obviously the majority, whether it's MIPS or PowerPC, or Arm. I think there's a continual recognition that the software ecosystem and the consistency of architecture is a fundamental advantage for Intel, making it very difficult for an alternative architecture to come into the data center from a compute perspective.
Not to pin you down, but just to pin you down on a specific number, do you think that Arm is going to be 5%, 10%, or negligible share eventually?
We've always said that that microserver segment will be about 10% of the total server market at some point, where 3% of it is light workload, Atom-class microservers. We haven't made projections about what Arm will do, but we think that segment remains very small, 3%.
Let's go here. Is it on? One over here. Excuse me. What? Is it on? Okay.
Over here on this side?
Go ahead, Ross.
All right. Ross Seymore from Deutsche Bank. Sorry, whoever's over there. Question on the LTE side of things for Hermann. A number of aspiring new entrants have talked about the same timetable as Intel talked about and unfortunately had the same lack of success with LTE thus far. Can you give us a little bit of information as to why you think you're ahead of the Nvidias, the Marvells, et cetera, the world Broadcoms? Do you think there needs to be consolidation in this space for the economics to actually work out in a way that's favorable to Intel?
Let me take the back end of your question first. Semiconductor industry always has shown consolidation, it's a repetitive pattern that there is consolidation. That prediction is very easy. I do not need a crystal ball for this. It's just what always happens. The front part of your question is also very easy to answer, and it is we are shipping. Obviously, we outpaced everybody else being in the race, and we are shipping. Why they struggle, what they struggle with, you may want to ask them.
All right. Let's come back over here. I think we missed somebody the first time. Go ahead, Chris. Excuse me.
Not on yet. Okay, here we go. Hi, Chris Caso from Susquehanna. Question for Hermann, also on just clarification about the positioning of the SoFIA products. I think what I heard is the 2014 product is positioned towards the 3G feature phone, 2014 on LTE. Is that more the entry-level LTE market when you get to 2014? Perhaps you could talk about what % of perhaps the LTE market you think you'd be addressing with that product, then for sort of more mainstream higher end, what would be the roadmap there? Obviously, I would assume that getting it into the Intel fabs is a priority as well.
The positioning is not a feature phone device. The positioning is an entry smartphone segment. We derive it from the feature phone legacy that we have, also the technology for the modem and so on and so forth. That's where we derive this from. The target segment for that market is entry ultra-low-cost up to the value. That is where we position this in the market. We believe in 2015, the volume in that part of the market will also include LTE. For that reason, we are already shooting at that point in time to have an LTE solution as well. That is also targeted for the value entry low-cost part of the smartphones.
Anything you could say just going forward with regard to sort of mainstream higher-end LTE and then bringing that product into the Intel fabs?
Sure. I said we believe in our manufacturing. We believe in our architecture. That is the path we go.
Let's go to the back of the left middle row here. Trey, it looks like we've got someone right here.
Thanks. Blayne Curtis of Barclays. You threw out a big number for tablets, $40 million next year. If you could talk about, you've talked about your processors going down to the $10 price point, and you showed where the share gains were coming from. It seemed at the low end, $75, $125. Is that the kind of processor price point you are targeting? If you talk about what operating systems, are these all Android, or would you see Windows down that level?
I did not completely catch the question. Could you repeat it?
Where does the 40 million unit-
Trey, if we can hand that microphone back
price point or operating system.
Take another stab at that question.
You talked about tablet share gains, 40 million units next year. I think, Kirk, your slide deck had where the share gains were coming from, and it seemed that the $75 and $125 price points at the low end had the biggest incremental gains. Can you talk about what the processor price point is at that very low-end $10 processor point?
Yeah, we address all of those segments. The products that we currently have in order to address those segments, they are of course best targeted for the performance and the mainstream segments. We will also work with our customers to aggressively put them into the other parts of those segments. I have not spelled out on how many it is, but what you could see on that slide is how much we participated this year in those segments and how many designs we are currently seeing for next year in those segments. You see the largest growth in the lower price segments compared to this year.
That was one part of this question, the other part is, I addressed that in the other slide where I said, we also in the tablet space, we embrace Android, and that will be the growth driver for the tablets. This year, we did not participate in the Android market for tablets very largely. Of course, the Samsung device does, but in a larger scale not. This will completely change next year.
That sound great. Let's come back over to the right side of the room. Is there anyone over here that we've missed? All right. I think at the back here, Tiffany.
Hi, this is Josh Harrington with RBO & Co. I had a comment or a question for Diane. First of all, I really appreciate the breakdown of the customer segmentation and your growth outlooks. I was just wondering if you could give us some more detail about what's changed relative to a couple of years ago in that enterprise segment and the reduction in the expectation there.
Yeah. If you go back in time, if you go all the way back, 2005, 2006, we had nice high single-digit enterprise growth, and then 2008, 2009 hit the economic downturn. We saw a contraction. 2010 and 2011, the economy comes back, and everyone virtualizes, so we see growth. There's this very tight, as I showed on the slide, there's a very tight correlation between enterprise IT spending and GDP. Once the CIO has given back their budget, as I said, once Andy Bryant gives me back my budget when I was CIO, then you take that money, and you look for how you could either generate the biggest return to IT from an operational perspective or the biggest return to the line of business from new services and capabilities.
That virtualization growth spurt in 2010 and 2011 was bringing down the cost of running IT, so direct value to IT. Now we've seen another economic downturn, 2012, 2013, we saw the market contract. Now we believe the market is coming back, as you saw the GDP projections. Now the next big wave of opportunity for IT to deliver value is through private cloud and big data solutions. There's this kind of cyclic wave of how IT from a unit perspective grows and contracts based on GDP and based on how compelling the solutions are that they can offer. Private cloud as an opportunity to drive down the cost of running IT is tremendous, and it also at the same time delivers tremendous value to the business from that on-demand service delivery.
That's why we believe now with GDP and budgets going back that there is a compelling reason to deploy new infrastructure and hence our projection on 8%.
All right. Back here, Doug.
Thanks for taking my question. Doug Freedman from RBC Capital Markets. Can you each share with us what key metric you're targeting to hit to meet your goals to corporate next year?
I can start. Cost per transistor.
Stay on that curve. High five. Good job.
Return on invested capital.
Diane, your thoughts?
Oh, it is hitting our growth targets, right? Continuing to see that growth in cloud and then continuing to see the value of our technology and getting enterprise IT back spending through giving them compelling solutions around big data and private cloud.
Hermann?
Yeah, Brian laid out mine. 4X.
4X.
4X and tablets.
Yeah, for PCs, I think we're in the year where we're stabilizing units, and we want to start growing units again through all the innovation we see, and then Stacy will outline the operating profit target, which I think will be, as I foreshadowed earlier, better than our unit, slightly unit decline.
All right. We have one more over here, and then we'll come back over to this side.
Thank you. Alex Gauna at JMP Securities. Bill, I don't recall ever seeing a chart like you put up where you're trying to close the yield gap between 14 nanometer and 22-nanometer. What I recall seeing at this event in prior years is the faster and faster ramp to yield that Intel seemed to always be achieving. Can you confirm, is this the first time you've gone through something like this in a node progression? Maybe some color around what you see causing it. Is it the FinFET? Is it the architecture? Then I noticed that on the yield curve, you're starting from a lower point than you started with 22-nanometer. What gives you the confidence it doesn't asymptote at a lower level?
Let me take the second one first. It only starts at a lower level because we cut the graph off to make it more visible where the problem is. They both start at zero. That's. We did put a little break in the line, so you don't know how far down zero really is. Zero is a long way down from there, but they both start at zero. I wouldn't say this is the first time we've had substantial yield challenges in development, but it's been the first time in quite a number of generations. It's just getting very hard.
The fact that we're not seeing fundamental improvements in the patterning equipment means all of that has to come from extracting more from the existing equipment set. The primary way of doing that is adding more layers of multiple patterning. Those layers have lots of complex interactions that you really don't find until you actually see them in devices. That just takes a lot of, you could call it blocking and tackling, but you just have to go find them and identify what it was and put in fixes. It's nothing more than that, but it is hard. As hard as it is, we think it's going to be hard for everybody else when they get to that point. It is not FinFET related. It is not fundamentally related. It is just scaling down the size of things is getting harder and harder.
Okay. We have one here. Go ahead, Romit.
All right. Hi, Romit Shah from Nomura. A question for Diane. You mentioned in data center a general improvement here in the second half of the year. First, can you just tell us how big enterprise is as a percentage of the data center business? I guess second, how should we reconcile the improvement that you're seeing with some of your customers like IBM and Cisco, which have been, I guess, signaling more caution recently?
Sorry about Cisco. I forget the first part of the question is, you just said.
The size of the market.
The size of the market. I think you saw on BK's chart. The enterprise segment is obviously becoming a smaller and smaller portion of our total business. It'll be about 50% this year. About half of our revenue will come from enterprise. We are seeing some recovery in the second half of the year. We look at MP servers that are four-socket or greater. Those shipments tend to go into enterprise IT, for instance. The cloud service provider world is all two-socket. We're seeing recovery in the high end of the service space, which tends to reflect enterprise IT spend. Cisco's earnings last week, they actually highlighted that enterprise being a strength area in their UCS group, and the enterprise segment actually grew very nicely again. I think it's a bit of a mix, what you'll hear from the various OEMs.
I will also say, if you look at the total server market, it continues to diversify significantly from an OEM perspective. There are more and more participants into that market. We've seen a significant shift over the past five years from very tight consolidation to now many different players, many of them coming from China as well. The local OEMs in China are growing as well. You have to look at a broader distribution to get a complete picture of the enterprise market now.
All right. We'll take one more, and then we'll break for lunch. Why don't we go right over to? Do we have a microphone on this side?
Thanks for taking my question. Gus Richard with Piper Jaffray. There's a lot of companies these days designing their own processors with Arm, Lenovo, Huawei, Amazon, et cetera. I was wondering, how open-minded are you to working with those customers going forward? How practical are you going to be?
To working with-
Gus, your question is in the foundry space.
You meaning to working with them to convert that Arm core to an IA core, or working with them despite the fact they're competing?
Either.
In the foundry.
Oh, from a foundry perspective. Sorry.
Yeah.
We're going to evaluate each one on a business-by-business basis and see what it does for Intel. There are some cases where we have very little play in a given market, and it may make a lot of sense for us to take up a foundry customer. In other cases, it won't. We're going to have to look at that. As Brian said, we're opening up that availability more than we have in the past.
All right. We'll wrap up the first Q&A session of the day there. We will have another one this afternoon with Brian, Stacy, and Renée.
Good afternoon. Oh, there's Stacy. See? At first, I said, "Well, talking after lunch, they're all going to be tired and relaxed." I realized I was the only thing that stood between you and Stacy. I know you're all here and alert and ready to go. I am going to talk about the final piece of the product strategy, which is really filling out the what are we doing with services. Based on our conversations at lunch, hopefully, we'll answer some of your questions about foundry services, because that's one of the things I'll be talking about. If we just recall from this morning, the framework that Brian started with, if it computes, it does best with Intel. What we have been working on for the last several months is really focusing our services efforts to adjacencies around these core areas.
Specifically, we have narrowed the team down to three areas. We had a lot of services efforts going on. You've heard us talk about them for the last several years. I would call that the letting a thousand flowers bloom or the chaos reigning, and now we've reined in chaos. We've narrowed down to three key areas: security, cloud, and foundry. I'm going to talk about each one of them in a little bit of detail, then I'll recap what we've talked about here today before Stacy comes up. Each one of these areas has a unique strategic opportunity for us. Well, usually, it's multiple things.
It's a source of differentiation for our platform, as well as an extension of a product offering in the case of cloud. Diane alluded to some of this in solutions. I'm going to talk about that. It's also an additional source of revenue. Case of security, it's a cross-platform technology. We've been talking about security for the better part of three or four years with you, certainly the last two years since we acquired McAfee. I'm going to talk about the final step on that piece of our strategy. The cloud services platform is maybe a little bit less known to you. I want to introduce some of the things we've been doing the last couple of years. You'll see that it ties very clearly into what Diane talked about for her customers and where we're going.
I'll lay out for you some of the things that we've been doing and where we're going to take that. In foundry, which I know many of you are interested in, you can mark this day. We're going to talk about foundry as a service line or as a product. I'll give you an idea of the two different product lines that we're offering and what we think our unique advantages are there. Starting with security. In the past two years, the need for security has only grown. Nothing has changed. It's only more so. The threat volume is growing exponentially. We know this. You see the threat reports. We put them out, Symantec puts them out, everybody puts them out. Nothing has changed. It just keeps increasing. The thing that's happening that's a little bit disconcerting is the threat complexity is growing.
It's no longer just malware and viruses and simple things. What we see now is a level of professionalism in the attacks that really is forcing the security industry as a whole to really rethink how we're going to solve some of these problems. Finally, the point of insertion or origin is changing. In the point of origin, it used to be, we used to talk about hacker groups or activists. It's a professional business, and I've said this before in a couple of other conversations. It's a significant double-digit billions business. It is a professional, criminal, organized, not just nation-state, just for commercial purpose is what's going on today. The point of insertion has moved to one of the more vulnerable places, which is the cloud.
A lot of what we see now are things hiding in the cloud, I'm going to talk about that. The opportunity to deliver value across all of our platforms here has only grown, and we are in the unique position that we talked about before, which is the vision that we've had all along, which is unchanged, which is to be the leader in delivering safe, connected computing experiences from the clients through the cloud to the data center. In Diane's slide, she showed you end-to-end security from the device all the way through to the data center. I'm going to talk about how we accelerate that in a second. We think that the differential advantage for what we can do with our platforms is from silicon plus software from edge to edge. That's been our vision since we acquired the assets.
We've made additional acquisitions in security since then. I'm going to talk about what we've done with those. Just going back because a lot of you have asked me as I've been out talking to you, "How's McAfee done?" And there's been a lot of misconception about them, and I know you can't always see that in the software and services segment. McAfee, in the last couple of years, has done well. In a flattening PC market, they were able to maintain their clients' business and transition their business on a forward-looking basis towards network and into next-generation firewall and into the data center. You can see from the slide the growth of their business into the new segments.
The most important thing that's happened is that they've been really reorienting their business towards where the high-growth segments of security are and to create this broad footprint end to end. That's the foundation for our platform strategy. The other thing they've been doing since they've been part of Intel is we infused a lot of engineering. I've talked about that, but now these are the results. In the last 18 months, we've re-engineered the entire product line of McAfee. Over the course of this last quarter and on a forward-looking basis, they have 30 new products that will be coming out, including the first re-engineering of the entire client suite in the last eight years. They have all new network products. They have next-generation firewall products. They have mobile products, and they have new enterprise management products.
Across the board, from consumer endpoint all the way through to the data center, they have re-engineered their product lines to be integrated, to have superior performance, and to take advantage of Intel hardware. The results of that, and I'll just share this, some of you know, that they are in the upper right hand as the leader in Gartner's Magic Quadrant. This is across multiple different categories of security, from endpoint protection all the way through intrusion protection, email, web, client. This is a multi-category that I didn't want to put up all the magic quadrants just to say. In general, there's multiple different third-party accreditation who've been looking at their new product lines and really are giving them very high scores, taking care of a lot of the things that we had heard about back as feedback from our customers.
Again, performance, integration between products, the ability to all be cloud-based, and to take advantage of unique Intel platform differentiation. As I said, security is fundamentally changing. The pace is changing, the rate is changing. Everything that Brian talked about, that Kirk talked about, Diane talked about, talked about the pace and the volume of devices and the proliferation into the Internet of Things, all of that volume is all new attack opportunities for security. On the flip side of that means that the TAM for security is growing very fast as well. What we're going to do is we're going to take where we're at today, which is where all vendors are. McAfee is a 27-year-old business. The security industry is older than that.
For 27 years, this industry, at least as far as I can tell, 27 to 30+ years, has been very focused on detection and cleanup of what you find, and then hope that what you found, you can prevent in the future. What our strategy has been, and now we are at the point of deployment, is about protection, proactive protection of known good, preventative security, and a focus on a movement from building off of the foundation of AV into identity, protection of who you are, protection of your data, and protection of assets as they move between different networks and across different devices. Creating ringed fences, which is something that's unique that we can do, and it's a combination of hardware and software, so that you can have known safe environments.
We're very much going to be moving, over the course of this year, towards proactive enabling of these safe interactions and in different environments across the network. To do that, McAfee has had a security strategy called Security by Design, that's the integration of the different elements of security, which is very unique to them as an asset where they had endpoint in cloud and network data center. That's been their plan. They've re-engineered the product line. As I said, they've been adding capabilities using the underlying hardware features that we've been shipping and are shipping in new products. We are taking the final step on the integration of McAfee into Intel. We are going to, and have already announced, effective immediately as of over a week ago, the Intel Security Business Unit. It's a single business unit.
It will still be reported in the software and services segment. I want you to think more broadly about this revenue because we will get paid multiple ways. We will be paid not only in the continuance of the McAfee software products, the new solutions that are going to be created as services between our hardware and software assets, but also in the hardware for the hardware features. Revenue can show up in multiple different places for security. Our hardware and middleware assets that we have in Intel will be integrated into this business unit as well. It'll be one end-to-end hardware, software, services, solutions organization. The McAfee team is excited about this. What this will look like to our customers is the McAfee products will continue, and over time, new products will be introduced under the Intel services name and with McAfee as core technology.
The organization is going to be focused on delivering ubiquitous solutions for client. What I mean by that is taking the foundation technology, the base, the thing that everybody's been working on for so long that everyone should already have deployed, but they don't. As I've told you, over 60% of the clients in the world have no security turned on, and making that ubiquitous. That is the very first thing that we've asked them to do. I want to thank Brian because, as Brian and I were appointed, he has been a very big supporter of continuing on this strategy and actually accelerating to the final vision that we had of this integration.
We'll be working on ubiquitous client solutions, raising the bar, the base for everybody across all platforms, not just Intel, and then building unique solutions for Intel platforms on top of that. We'll be working on integration into the network and cloud and taking more advantage of Diane's platforms, as well as storage network and where we're going in both of those business. We'll be building unique solutions that are new products, if you will, that are unique to Intel platforms only. That's what's going on with security. A little bit of news. It's effective immediately, and of course, as you can imagine, there's a large number of people who are all coming together now, but I think in the next quarter or so, you're going to see a lot of new things from us.
I'm going to move on to cloud and big data, which is the other very exciting area for us to talk about in services. Diane talked about how enhancing our data center customers' cloud is a big part of what Intel can offer that others can't. In addition to that, she talked about enterprises wanting to use private cloud technology. One of the things that they need in order to be able to move forward with that is the assets, the core assets that really ride right above the silicon that help them be able to build out a cloud infrastructure. As you saw on some of Diane's slides, she mentioned the Intel optimized Hadoop. What we have been doing for the last two to three years is acquiring and building assets that build out a cloud services platform. Why do we care about this?
We care because it allows customers, even very large customers that you would know with commercial services that everybody uses, to very quickly take an Intel Xeon server and have an optimized version that's 40% faster of Hadoop, a very fast file system in Lustre, and security optimized for Intel, all going to scale across their data center and build their services right on top of it. They don't have to worry about what they would call, in the old days, we would call it in the software world, middleware kind of plumbing. It's kind of middleware for services, if you will. It's very sensitive to the performance and features of the underlying hardware. In addition to which, it has a very fundamental interaction at this level with the client. You want these components to be client-aware.
You want to know the security on the other end and be able to read it in. You probably want to know through the API what's going on on the other side of the cloud. These assets we selected very specifically because they're unique points in the stack that are very sensitive to being optimized. They all go together to create a platform that extends Intel's position in the data center, and they're all assets that our customers are already using today. Either they were leading technologies that they're using today or that they've already decided to subscribe from us or use from us now.
Our strategy here is to deliver a performance enhanced solution stack for our data center and telco customers, private and public cloud, to optimize those elements, as I said, across our product lines. We will continue to focus on additional key technologies that are close to the hardware. There's other things that weren't on that list without going into a lot of technology discussion about different software, but we will be continuing to fill out this offering, and it will become, as Brian said, we're going to move beyond silicon into solutions and platforms. This is a big piece of enabling into the high-growth segments that Diane talked about and for additional revenue opportunities on top of it. The final piece is the foundry services. We have had lots of discussion this morning already about foundry.
What I'd like to do is present to you what we're doing as a business and how we think about it as a product offering to the market. This is the first time that we've done this, so I'm very happy that the team is ready to do this kind of thing at the analyst conference. Let's start by talking about the most fundamental part of who we are, which is an IDM. Stacy will tell you how much that costs. You all know us as an IDM, and we're one of the last few in the industry. It gives you tremendous advantage. The advantage that Bill talked about this morning comes from the integration between design and manufacturing. That same advantage, we can extend to our customers.
We can extend it to our customers who are already great customers of the Intel architecture, who want us to help them do things uniquely for their businesses. Our foundry offering is actually two pieces, and I want to make sure that we explain that. Unlike our competitors, we have the ability to do a dual approach. We can offer semi-custom Intel architecture products. Diane talked about having, I think it was 15, that you mentioned in your presentation this morning, already out there. It extends even more broadly. We know that lots of people think that they want to build their own silicon. They don't really need to.
They can have access to the wonderful leading-edge technology of William Holt, and they can use Intel architecture, and they can leverage all the design that we've already done, all the tests, all the validation of hundreds of millions of units by adding their unique IP alongside what we're doing and running it in our factory. That is the semi-custom Intel architecture opportunity. It allows us to extend our product line. It allows us to be very customer-oriented for the customers who feel that they need that. Lots of customers think they need to build custom ASICs and other products to do things that we know we can do. This gives us an opportunity to do that for them. We draft off of our already integrated design and manufacturing relationship to offer this as a new, I'll call it a service, approach for our customers.
The second part is the custom foundry piece. I'm going to talk in more detail about how we offer this to customers. Obviously, it's an opportunity for us to use our knowledge of design and manufacturing on the leading edge to offer to other people who we think can take advantage of that. Let me talk a few seconds about that. Our business model is a little bit different. We're a little bit more flexible in that we have a complete array of foundry capabilities on top of our IDM manufacturing. You can select from us everything from just wafers and standard foundry to optimized design, integrated supply chain, assembly, and full shipping products. We'll look at the full range of those things.
Like William said, we look at each deal on a case-by-case basis, and we work with the customer based on what their needs are to assemble the right offering. We allow our customers to select at what level they want to interact with Intel and Foundry. We also embrace all of the industry-standard ecosystem support, so it's easy for them to use not only Intel's tools but also industry-standard tools as they come in to our factory. We have had a lot of questions at lunch about this. It does take a bit of time for people to transition in. That's not unusual. When a customer is announced, it takes about 18 months, right, William, for them to get in. We work with them over the course of the design. As William said this morning, we entertain whatever our customers think they want to build.
What we do is we offer them the Intel architecture, and if they want to build as well. Some are, some aren't. Some applications, it makes no sense, and other applications, they would get better performance. Customers decide on an application-specific basis, and we deal with the deal on a deal-by-deal basis, not on an architecture-by-architecture basis. Hopefully, that answers one of those questions. What we do is we fully leverage the capability that we've already invested in to extend our footprint in the semiconductor industry. In many cases, these are segments we wouldn't otherwise serve. As Brian said, if it computes, it's best with Intel, and we think we can offer this opportunity to be best with Intel. We have a couple of customers who are public, more than the few that Bill put up. These are the only ones that are public.
We do not disclose our foundry customers because they do not want to be disclosed. If they wanted to be, they were on this slide already. We have many customers, most of them come to us. I think Brian mentioned this, we don't necessarily have people going out yet selling. Everyone's heard that we're open for opportunities, so they're coming to see us. What they really get is the leading transistors and the scaling for their business. Just to summarize the approach, first, the semi-custom really enhances our market position. For many customers, especially the largest cloud customers, this is quite important. Many of them have application-specific offloading or unique intellectual property that they want combined with what we are doing in software. Without this offering, I think it could cause us to have some limitations. We use our capability in IDM.
We allow the customers to combine their IPs, and it gives us an enhanced position. Foundry allows us to capitalize off the investment we've already made, I just said that, we are evolving and expanding our business model. I think the thing you should think about is if the customer value is leading edge, and we think that there is a value proposition that works for Intel, we will engage with them, even in areas where there may be some competition with businesses that we're in. With that, I will summarize the adjacent services pieces. We're moving very quickly on security towards this ubiquitous solution. We're moving quickly to get to the value proposition that we originally identified when we acquired McAfee with the Intel Security business unit. The cloud piece of our services offering is getting very focused.
It's very much about enabling the adjacencies for Diane's business and creating a new value opportunity and stack on top of our silicon. In Foundry, we're extending our capabilities, and we're open for business. We get to Stacy. I want to quickly review what we've heard so far today. It's been a long morning, I want to make sure that we leave you with all those messages, then we move into the financial presentation. In data center, obviously, we'll start with the first part, what Brian told us, if it computes, it does best with Intel. In data center, what Diane told us is that there are strong growth segments. We're very well-positioned. She told us especially about networking, which is a new opportunity for us, and it's forecasting the 15% growth CAGR over the horizon.
Kirk let us know in PCs that the market's stabilizing. We have a disappearing, wonderful array of 2-in-1s and convertibles that we're very excited about the new innovation that's coming to that market, and that we're going to have the best experience on every operating system. I know this doesn't surprise you. We've always had a long history in software, supporting all of the popular operating systems. I think the new news is that there's lots of choice now on the client, and more importantly, that we're going to have products at every price point. I think is really great news for us. We heard from Hermann in mobile, very important, our 4X goal in tablets. I got the question at lunch, "Is that tablets and phones?" No, that's tablets. We are in the phone business. You heard from Hermann.
We're very focused on now having products again across all price bands and all segments. We talked about increasing our integration and our focus on comms. What Hermann told you was that our customers are very excited about us being in that business and continuing and are asking us to go faster. That's all great news. On the innovation side, Brian hit on some of this, very important. We are continuing to invest in innovation in areas like perceptual computing that really push the boundaries of compute. We'll take those investments, and we'll waterfall them across the entire product line. That continues, and it has been the driving force of how we move forward in this industry. What are we going to do next with gesture? What are we going to do next with camera technology, 3D? Where does language go, biometrics, et cetera?
Security happens to be one of those driving forces over the horizon as much as perceptual computing. Two big adjacent areas where we're making investments. Of course, Foundry, where our shift in strategy to be not only providing semi-custom IA for our customers, but also foundry services based on our investments in leading-edge technology. In summary for the day, I think we've tried to show you what's different, what's changed, where we're going for the next little while, and hopefully, this was clear for you. We're going to have some Q&A. With that, and no further ado, I would like to introduce our CFO, Stacy Smith.
Good afternoon, everybody. Good afternoon. Good. I am sympathetic to the fact that you've been sitting in your chairs a long time. Some of you even think that these presentations started yesterday. That was before lunch. I'm cognizant of the fact that I'm the only thing between you and a reception, which is always a highlight for you. As you know from prior years, I usually try to start off with a funny anecdote or a story of some kind. A couple of years ago, I was talking Arnold Schwarzenegger and his issues were in the news, and my ending line was, "Don't worry, neither Paul nor I have domestic help," which everybody in the room liked except for Paul.
Last year, I stood up, I started by exaggerating my credentials because of the Yahoo issues, which everybody in the room liked until Paul got the call from Yahoo. Not liking that. As I was testing out a couple of my stories, one of which did involve the mayor of Toronto with Andy, he said, "You know, you got a new CEO, first investor meeting. Maybe you want to kind of go a little light on the story." I thought, "Well, okay, you've been sitting here a long time. How do I start in some way that's interesting and fun?" I don't have any demos, right? A finance demo is second only in excitement to a photonics demo that Diane's going to take you through later. Sorry. The reason that you all come here, it's really not to hear me present.
It's for the guidance that you get. I thought, I'm just going to make this easy on you guys. I'm going to give you the guidance up front. That will make it exciting for you. I am going to start in a way that I've never started before. I'm going to give you the guidance up front in the form of a haiku. Isaac is struggling right now to figure it out. He knows he could get an hour jump on the rest of you if he can figure this out. My haiku is units, autumn renew, warmth of data fireplace, spring mobile motion. All right. This will all make sense to you at the end, by the way. That is to the real haiku format. I'm pretty proud. In consultation with legal, we thought that was an FD-compliant way.
Actually, legal showed a remarkable ability to generate a haiku. We may just do this for guidance from now on, I think. What's up, Oh, Andy wants the mayor's story now. We just had iced tea at our table. I don't know. All right. Let me jump in here. I'll get serious. Today, I am going to try to give you a lot of insight into our business, give you a sense of how we see our market and our financials as we go into 2014. I'm going to build on three themes that you heard today. The first is that we really are repositioning the company to sell into a broad range of devices. We're investing in the PC to reinvent that market.
You're going to see, as I go through, the investments that we're making to shift the company towards low-power SoCs, towards Android, towards phones, towards tablets, towards the Internet of Things. I'm going to spend quite a bit of time on tablets to talk to you the kinds of investments and the kind of actions that we're taking to build share rapidly next year. Second, there's this wonderful spiral that Diane showed you, and you should ask her what Grove thought of her spiral when you get with her. He loves these spirals and loved her spiral. I think it actually is a great visual of what's going on in the data center, and this explosion of services feeding devices, which drives the build-out of the data center. We are uniquely situated to benefit from that. I thought that was a great presentation.
It kind of walked you through our position in the traditional data center, how we're benefiting in networking and storage. You heard some strong hints, and maybe going even beyond the hints, about how our ambitions span beyond just the silicon, but it's using our software capabilities. I thought Diane's example with Hadoop was a really good one of how we create this platform that give us more value, to contribute to our customers and give them an ability to build on top of us to add value to their customers. Third is our technology leadership. I will not be the one to thank Bill today. Everybody else up here thanked Bill. I'm going to tell you, he's the one that makes my life hell with you guys when I'm defending an $11 billion CapEx number.
I'm going to spend a lot of time talking through how that is a fundamental advantage for us. It's not just a science advantage. It's not just an ethereal advantage. It allows us to lower our unit costs. It's increasingly something where we can provide direct value to our customers through our products and direct value by making our transistors available to others, which at the heart is what we're looking to do in the foundry space. I'll tie back to these themes over the next hour or so. Okay, just a simple graphic. Sometimes in the simple graphics, you really see the power of an idea. This shows how we think about our markets. If you go back a few years, our market was really confined to the PC market.
Let there be no mistake, the PC market is a great place to be confined to. Even today, after a couple of years of unit declines, it's a $33 billion business generating $12 billion in operating profit. That makes it one of the largest and most profitable businesses on the planet. What's equally as exciting is the opportunity that we have in this much broader market. We're well on our way to reinventing the PC market. You saw a lot of that today from Kirk. As you heard today, we're focused on winning share in tablets and then driving our transistors and our SOC capability into this much broader range of devices ranging from phones to the Internet of Things. Some of these markets are markets where we don't have a lot of share.
In other cases, you look at the Internet of Things, it's a market that we've been in for a long time and actually have commanding share and a commanding presence. Also say on a personal note, what you're going to hear through my presentation, and I think what you heard from the other executives as they spoke to you today, is this really leaning forward to these market opportunities. We're not just embracing these market opportunities. We're on our front foot, and we're going after them aggressively. You've heard various executives, starting with Brian today, talk about how we're repositioning the company to focus on this broad range of devices.
As I was preparing for this discussion, I actually thought one of the more powerful ways to show that to you was to let you look a little bit under the cover of how we're making our investment decisions as we go into 2014. As a finance guy, I always look at where is the company putting its money. That tends to be a good indicator of where are they serious and where are they not. What this shows is, over a couple of year horizon, how have we shifted the investment profile of the company. I've put in some key areas there, where we try to look comprehensively. What you can see here is a significant increase in our investment in tablets. That shouldn't surprise you based on what you've heard today.
A significant increase in our investment in low-power SoCs, security, Android, LTE, the Internet of Things. We're also making significant increases in our investments in technology development, and we're investing more in the data center to make sure that we can go after the opportunities that we talked about there. The first element of guidance that probably wasn't obvious in my haiku is our plan is that we want to keep spending roughly flat next year. That means that as we focus on these new priorities, we're also looking at places where we can be more efficient or where we can leverage the assets of the company in a different way. One of those is in one of our largest organizations. We're going to invest less next year in the PCCG segment.
Again, under the covers there, they are shifting rapidly towards investments in things like perceptual computing, two-in-ones. They're biasing their investment towards the kind of use model-oriented things that you heard about today. Generally, we're going to invest less in that space. Phones is another one. You heard from Brian and you heard from Hermann, the focus that we have, we think we were successful with our strategy of proving that we could fit into low-power devices through our FFRDs. You're seeing a focus now that's focused on a handful of large market-making companies, where we're doing less at the in-system level than we were a year or two ago. You also heard a lot from Hermann about how we now have this converged SoC roadmap that gives us common cores that we can take into phones and tablets.
When he went through his timeline, those were converged cores that go after both segments of the market, we get some efficiency there. Just to give you a map of where I'm heading, my presentation today is going to be organized into three sections. First, I'll take you through the reportable segments of our business. I'm going to take you through our core financial advantage, which is our ability to manufacture the world's highest performance, most energy, and most cost-effective transistors, generation to generation to generation. I'll close out by sharing some information I know you all love on cash generation and our priorities of how we use that cash between investing in our business and returning cash to our owners.
I'm going to start by talking about the two client organizations, the Personal Client Computing Group, which I may shorthand to PCCG at times, so bear with me if I do that, and the other Intel Architecture Group, other IA. I'll talk about the data center, and I'll close out the section with some financial information about Software and Services Group and NAND. Starting with the PC Client Group. In 2013, we expect that this group will generate about $33 billion of revenue. It's down about 5% on a revenue basis year-on-year. The TAM is down about 10% in PCs. The total available market on a unit basis is down about that much.
As a reminder, as you try to reconcile the different segments, netbooks for us, the netbook volume doesn't show up in PCCG, it shows up in other IA because that's an Atom-based core. As you're doing your reconciliations, you just need to be cognizant of that. If you exclude netbooks, the TAM for PCCG is down about 7%. Our billings units are down a little bit less than 7% in this segment because we believe that we gained some share over the course of the year. This segment will generate about $12 billion of operating profit, which is about 35% of revenue. As Brian showed you, we are seeing this market start to stabilize. You also heard that from Kirk today.
He showed some of that quarter-on-quarter trending, which would suggest that as we get into the back half of 2013, we're starting to see the unit declines abating on a year-on-year basis with some chance that Q4 for us on a billings basis is actually pretty flat with Q4 of a year ago. As we look to 2014, what we're seeing from the third parties is that they're forecasting PC declines overall that are in the low single digits. That's pretty consistent with our view. As we look at 2014, we think that PC units decline in the low single digits. When you start breaking that out, the expectation is that we're seeing some stabilization in mature markets, some stabilization in enterprise, but that the tablet growth continues to impact us in emerging markets.
Additionally, in emerging markets, we think there's some economic issues that will have to be worked through. That out of the way, I just want to tie back to something as I was struck as Kirk did his presentation by how much innovation is happening in this segment of the market. It really is unlike anything that I've seen in my 25 years. When you look at that slate of products that range from things that are priced very low to this amazing technology at the high end, the technology that's going into the two-on-ones, the two-in-ones, the touch, the features coming to the market, perceptual computing features and things like that, it really is pretty astounding. We are steely-eyed focused on reinventing this market.
We're starting to see it stabilize, we believe that we're bringing some great technology to the marketplace that will bring value ultimately to the people that are replacing their PCs and the new buyers in emerging markets. I want to click down a level for PCCG and show you a couple of long-term trends, pricing, and cost. Average selling prices for PCCG have been flat to up since 2009, which is a change in the curve that we saw to that point. If you go back to 20 years before 2009, what we saw is kind of a 4%-5% decline in pricing. This is really driven for us by mix. We have seen high-end demand stay strong at the high end of our stack.
I think it's a combination of where we've lost share at the lower end, also the compelling features and performance that our Core-based products deliver has kept demand in this segment pretty strong, Kirk shared some information with you on that. Again, here, the mix is not by accident. I'm tying back to that innovation theme. It's because of the innovation, it's because of the features, it's because of the performance that we're bringing to this segment of the market, and it's also because of our branding program. We've done a very good job, I think, of being able to articulate the differences between what you get when you buy a Core-based system and what you get when you buy a value-based system. This graph shows our cost curve over the same period of time.
Again, this is for PCCG, we've isolated for that segment of the market. You can see that even as our mix has gotten richer, our costs continue to come down. A nice hint for the next section, this is Moore's Law at work. I'll show you cost curves for different segments of the business, I'll actually show you cost curves for different segments within this business in a second, this long-term trend is Moore's Law at work. If you added up the quarters, you'd see here that the costs come down some in 2013. We expect them to come down a bit more in 2014. I'll give you a bit more on that in a minute.
I am going to come back as I get into the next section and make the direct tie between how we're able to keep bringing our costs down year-over-year, generation over generation, in a world where capital intensity is going up at a faster rate, because I know that's something that is difficult to intuitively understand. Putting these trends together shows how our PC group isn't just a large group, it's a large group that delivers a very healthy margin. It's a big business. We do think it's stabilizing. If I were to articulate for Kirk the priorities for next year, which I think he did a good job of articulating on his own, it's first and foremost, drive innovation, reinvent the PC with 2-in-1s. It's extend our presence so that we're port of choice across other operating systems like Android and Chrome.
It's invest in differentiation and features, that's across all of these different operating systems so that we have a nice selection of value-based systems that hit price points, nice sell-up features as we move up the stack. Very importantly, it's drive that volume growth by enabling lower system price points, I think you saw a lot of great examples of that today. A key part of that is segmentation, product segmentation. For us, critically, the product segmentation is enabled by cost segmentation. This is a journey that we've been on now for several years. What this chart shows you is the unit costs across the various PC segments.
For each of the segments, performance, mainstream, and value, it shows the actual cost through 2012, it shows where we're on track to be 10-11 months through 2013, it shows you our forecast for 2014. Starting with 2013, what you see is that costs came down in both performance and mainstream, relatively significantly in the performance segment of the market. Went up a little bit in value. We made a choice to dramatically beef up the graphics capability of the value segment of our product line, that consumed more die size, conscious choice, but a little bit of an uptick there. What you see for 2014 is that costs come down across every segment of our business with the most significant decline, this is an average cost for these different segments, in the value space.
That decline there is driven by Bay Trail coming in, becoming a large percentage of our overall total in the value space. Let's now just click down a level and look very specifically at Bay Trail, because I think Bay Trail is a great example of how we enable unit growth through this cost segmentation. What you see here is the Core-based Celeron product at the end of 2013, that really dominated that mix in the prior chart. You see here where we expect to be with Bay Trail next year, then Broxton, which is the next generation of Bay Trail. In this case, it'll be a Broxton Celeron branded product, where it is in 2015. What you see here is that the product that we'll have in the market next year is a 30% lower cost than Celeron is today.
If you go out one more generation to Broxton, you're down another 37%. Over the course of between now and 2015, we'll bring down our product cost in this segment of the market by on the order of 70%. This gives us the cost structure to grow that market at the low end while we continue to drive features, use model, and segmentation at the high end. This becomes a really important capability for us as we think about how the market develops going forward, because we will go where the market is. To do that profitably, we have to have the cost structure to go after it. To wrap up for the PC Client Group, here I'll give you some very specific guidance for what we expect in that group next year.
For 2014, we expect that revenue is down in kind of the mid-single digits. That is, as Kirk said in the Q&A, that is down a bit more than what we expect the unit growth to be, and that's because we believe that there's going to be a bit of a mix down just based on Bay Trail becoming a larger % of the overall mix. Because of the cost structure, and this is important, we believe that even with a revenue decline that's in the mid-single digits, we're targeting flat operating profit. That's a result of lower costs. Part of it is lower startup costs, and reductions in investment level in this particular segment of the business. Okay? I'll put all this together at the end, but this is what we expect for the PC Client Group.
I'm now going to shift gears to the Other Intel Architecture Group. This is our other client organization. This group contains our Atom-based tablets. It contains our phone business, multi-coms, that was the piece that Hermann was talking to you about today. It contains our Intelligent Systems Group, which you haven't heard about yet today. I'll share you some information on that. Then we also have the New Devices Group that is part of this operating segment. This is a $4 billion business for us. It's down about 8% year-on-year, based on how we expect 2013 to come in. The biggest decline in this business is a result of netbooks coming to the end of their life.
There was still a pretty sizable amount of netbook volume in 2012, pretty close to zero in 2013, and we expect all of that is shifted over to tablets and low-end PCs by the time we get to 2014. Our phone and multi-com business, particularly the multi-com business, is also down as customers transition from 3G to LTE, and I'll show you the implications of that transition on our revenue curve in a couple of slides. We had a wider operating loss in 2013. It's multi-coms, it's tablets, it's phones, they were all down with a wider operating loss year-on-year. Those losses, though, were partially offset by pretty sizable revenue and profit gains in the Intelligent Systems Group, which is why I'll put a little emphasis on what's going on in that group because it's a pretty good story and I think a nice benchmark for where we're heading.
Consistent with what I showed you earlier in the investment, this is a place where we're making significant increased investments. This is a place where we're beefing up our product portfolio. We've been beefing up our comms capability. We've been investing a lot in low-power SoCs, and making incremental investments where we see opportunity in terms of the Internet of Things. There's a pretty big shift in investment over to this segment. I'm now going to focus on a few key topics within this segment, and I'm going to start it with tablets. I'll put together some of the things that you heard over the course of the day. Let's start with a conversation about Bay Trail. Bay Trail is going to be probably the big runner in the tablet market as we go into 2014.
It's a Silvermont core. You saw it on the chart that Kirk showed you. Bay Trail is a great product. It was designed for the high end of the tablet market. We are seeing a lot of designs coming to the market this quarter, including some Android-based designs. As you saw from Hermann, a lot of the growth in this segment is happening at the lower end of the tablet market. We've made the decision, implicit in that 4x ramp, that we are going to take Bay Trail into a broad swath of the tablet market. While it has great performance and features, it does not have the level of integration that you typically see at the lower end of the market, and it's driving a systems bill of materials.
Beyond the integration, but the systems bill of materials that people have to put around the tablet that is more appropriate for the performance segment of the market. Our goal, as you've heard, is to quadruple the tablet volume to get to north of 40 million units. This is the year for us because Bay Trail is a very good product. We want to build momentum in the marketplace. Then you see products coming on the back of Bay Trail, which you saw from Hermann, Broxton, SoFIA, that we think will be lights out great products for us. We want to start building momentum in a very significant way in 2014. In order to do that, we are going to be aggressive in the market, working with our customers.
We're working with them from the standpoint of providing non-recurring engineering, where we're helping them port over designs, and we're being aggressive with them in terms of providing contra revenue to them in order to help them through the bill of materials that our product line drives until we can get to a bill of materials reduced platform, which happens over the course of 2014. The impact of taking Bay Trail broadly into the broad tablet market, and including the value portion of the tablet market, will be a significant increase in the operating loss in this segment. At a corporate gross margin level, it's going to drive about a point and a half of gross margin negative impact in 2014. As we get to 2015, I expect that that gets a whole lot better.
It gets better because of the products that you saw from Hermann and what you heard from Brian in terms of this segmented slate of products and the level of integration that they have and the bill of materials that they drive that we bring to the marketplace. Let me talk about the products that you see on this graph. By the way, what this graph shows is it shows Bay Trail, which is on the far left. That's the product that we're taking broadly in 2014 across the tablet market. Then it breaks out our tablet products in 2015 between the high end of the market and the low end of the tablet market. These are to scale, so you're seeing actual relative die size between the products.
As you go from left to right, what you see is higher and higher levels of integration of the product line. Broxton is the high-end product that we're going to be bringing to the marketplace. It's designed for the performance segment. You heard a lot about it earlier today. It's a 14-nanometer product. It's going to have a stunning level of performance, particularly in graphics. It will also have a higher level of integration than does Bay Trail. If you go to the far right, you see two versions of SoFIA, which you heard about both of them today. The one furthest to the right is integrated 3G. The one in from that has integrated 3G and LTE. Let me come back to that bill of materials issue. I want to put it in perspective.
The bill of materials to our customers as we get to the Broxton generation, some of this is directly driven by Broxton, some of this is other engineering enablement that we're doing with our customer base. The bill of materials, in terms of building the system based on Broxton to our customers, will be $20 less than what we're seeing with Bay Trail. We are putting a whole lot of energy in driving that bill of materials down so that we have super competitive products across every segment of the market. If you go to the far right, when you get-- That was Broxton. If you go to the far right, and you're now looking at SoFIA, there you're getting into products that have highly integrated comms as well as other integration around them.
You'll see an even greater reduction with SoFIA for the value segment of the market, that's on top of the significantly lower product cost that we're going to have as we start driving that die size down. The strategy here for us is win share in 14, get to 15, where we have this segmented product line, great performance, great features, and a much better cost and integration profile than what we have in the market in 2014. One last thought on this chart to just come back to one of the questions that was asked and give you some insight into where we're heading. This is SoFIA before we land it on our 14-nanometer process technology.
When we land it on 14-nanometer process technology, you will see a die size that is a fraction of the die size of this, you're going to see a big boost in terms of the energy efficiency and the performance. We're pragmatically trying to get this product to the market fast. You kind of see the economic reason of why, beyond just building share. We'll land it on our process technology, where we get an even further cost benefit. We'll get a big benefit in terms of the performance and the energy efficiency. All right, I'm moving on to our multi-comms business. You heard a lot about this today. I think you all know the key trend here is that shift from 3G to LTE. That shift clearly negatively impacted our revenue year-over-year as we went from 2012 to 2013.
We're pleased to say, though, that we started shipping our data LTE modems in the first half of this year. You may have seen that with the key design win on the Samsung Galaxy Tab 3. On top of that, we're now shipping voiceover LTE products. This is a place where we've been investing significantly. I think we're a clean second to the market. I loved Hermann's answer of why do you have confidence you're going to be second to the market, and his answer is because we're shipping. Other than Qualcomm, no one else is.
I think you can see. We're a little later than we wanted to be, by the way, but we're a clean second to the market, and our goal here and our investment level here is commensurate with staying time to market with the market leader, but we had some catch up to do to get there. This shows the revenue ramp, you can see, I think the market has been hungry to have a second source. Our expectations are that as we start shipping, or as these products, because we're shipping already, as these products come to the marketplace, the end products come to the marketplace, that we see a rapid increase in revenue as we get into the back half of next year. All right, I'm shifting gears now to the Intelligent Systems Group business.
You haven't heard anything about this business. Just for those of you that have been around as long as I have, this is the artist formerly known as the embedded business. I started my career in this business back in 1988. Intel has a wonderfully long and successful history in this business. We are the market leader. It's been growing fast over a long period of time. To give you a scale for this, because sometimes we don't always tell you the scale, the revenue for this business was right at about $2 billion in 2012, will be significantly above that as we close out 2013. This business has been growing at a 16% CAGR since 2009, just to triangulate that growth rate, we're expecting a roughly similar growth rate as we go into 2014.
The gold line on this is a pretty important line, this shows on the same scale our design win pipeline. We have a methodology, as most companies do, of when we've won a design, we look at the size of the design, we put some probability around how much this will actually come to the market, that gold line shows our design win pipeline. For this business, this is by far the best predictor of future revenue. When that line's going up, you have a high probability, barring some economic event, that your revenue's going to follow. When it's flattening out, you know that you have a problem that usually hits you about 18 months later. As you can see for us, that the design win pipeline is quite robust.
I also want to point out two changes in this business that illustrate our focus on this market and the Internet of Things. First, we started a group this year that's focused on emerging technologies and devices under Mike Bell. He's the one that's wearing the formal Tommy Bahama shirt there. You all may know him. That, I think, really illustrates one of the points that Brian made, which is we're putting our eyes to the horizon, and we're trying to get to these markets first. We won't always get it right, but we want to make sure that we're engaged, we're looking at where the usage models are starting to develop, we're putting real energy behind that. Additionally, we combined two organizations.
We combined Wind River with the Intelligent Systems Group under Doug Davis, who many of you know if you've been around for a while. That's building on what Renée walked you through. We find that being able to provide a solution at the platform level in this business becomes really important. It increases our ability to bring a solution to the market, to bring value to our customers. It improves the stickiness of that solution, and frankly, we get paid more for it. We get paid twice. We get a software sale. We get a hardware sale. In some cases, we're actually selling a system in addition to that, some sort of a platform that the customer's building on top of. I think a great example of this is in the automotive space.
You can look at some of the design wins we have, where we now have this IVI platform with Wind River software that sits on top of our hardware. It improves our customer's time to market. It improves their cost structure. It's giving us a lot of design win momentum. We're selling at a higher level of integration, so we win and they win. It has also resulted in a place where I can say with absolute certainty, we have 100% market segment share. This always makes legal nervous when I start talking about 100% market segment share, but this software platform is 100% of the autonomously driven car market on Mars. This software platform powers the Mars Rover. I think there's two of them, and we power both of them.
I also read there was a problem with it, and Doug assures me that's a hardware problem, not a software problem in this particular case. All right. Tying all this together for the other IAs group, we expect significant unit growth and share gains in tablets. We expect revenue growth, I was pretty explicit with that, in the Intelligent Systems Group. Overall, we expect lower revenue year-on-year in multi-coms, but with that very fast ramp as we're ramping 3G into the back half of the year.
Very importantly, we expect that the non-recurring engineering and the contra revenue support that we're targeting at gaining share fast in tablets is going to offset some of the other revenue growth in this segment and will result in a wider operating loss for this segment and will have an impact that you're going to see at the overall company's gross margin level, and I'll quantify that for you in a little bit. All right. I'm shifting gears now from the client to the data center. I ended up taking out most of my slides on the data center because I thought Diane just did a tour de force of a run-through of what's going on there. I'm not going to take you through any of the market trends. I'll just focus a little bit on the financials.
This is a business I think most companies would be envious to have, both from the standpoint of the growth rate, the product line, which is spectacular, the breadth of our offering, and the profit. It's on track to be about $11.5 billion of revenue. In 2013, we'll grow, wait for it, within spitting distance, I know I introduced that term on the call a quarter ago, of 10% versus 2012. Spitting distance was within one point of that, by the way, if you want to define financial terms here. Since 2010, we've seen a revenue increase in this business that's up by about a third. We are seeing both unit growth and a richer mix inside this segment. It's generating over $5 billion of operating income and has an operating margin that, depending on the year, is right at or just a little bit under 50%.
Diane, again, did a great job of articulating the unit growth here as a result of very rapid build-out of the cloud storage growth that we're starting to see in networking, high-performance computing. There are four trends to keep your eye on in this business, just to summarize her presentation. First, it's that explosion of devices, services, and data and how that's driving a build-out of the cloud. I loved the data that she shared that shows that every 400 smartphones, every 100 medical wearable devices, every 40 connected factory tools, as you deploy these incremental devices, they drive a server. The second trend that's driving this business is frankly the almost insatiable demand for high-performance computing. Every time we look, somebody's implementing a high-performance computing cluster that would not have been even imaginable two years before.
Third, we are now participating in a very broad way across storage, and we're seeing that same trend play out in networking. We're starting to see an expansion of our footprint across the data center. Last, we continue to offer IT shops a cost advantage. At the end, I was a CIO before Diane. She was a lot better at it than I was. At the end, IT shops, I think, are brilliant at making the right economic decisions. It's because they're all under budget pressure. To the extent that we can come in and show them how they can lower their costs by going to a private cloud, just as we did to show them how they can lower their costs by going to virtualization, they'll start to allocate more of their constrained budget to that priority. I think it's as simple as that.
I want to just take one second on microservers. We talked about the fact that our strategy is to participate broadly across the data center. You saw the kind of breadth of our product line. With Xeon, we have clear performance per watt leadership, but there's also this segment of the market, and it is pretty small today, but it's a segment of the market where the solution does not require the horsepower of a Xeon. For that segment, we've now introduced into the marketplace our second-generation Atom-based SoC. It's a product called Avoton, which you heard about earlier. It's a true 64-bit SoC. It's got the RAS features that you want in the server market. It's got the software compatibility that we think brings real value to the people trying to implement solutions on this.
Because it's Atom-based, we do sell it for a fraction of the cost that we sell a Xeon for. Because it's Atom-based, it also has a fraction of the cost of Xeon. One of the ways that I look at this is we look at revenue and margin per wafer start. What's interesting is when you look at the revenue and margin per wafer start of starting a wafer of Avotons versus the revenue and margin per wafer start of starting a wafer of Xeons, you can't tell the difference. It's the same revenue within a few %, and it's the same margin. If you guessed which one had the slightly higher margin, my guess is you'd guess wrong.
This is less expensive, and we can provide a lot of value to our customers, but don't think of it in the terms of that it's the segment where we don't make money because we will make money here. Looking forward for the data center, this business will approach $13 billion in revenue this year. The trends driving our data center are these secular trends that I've been talking about, these devices connecting to the cloud, the services, high-performance computing, participation across a wider range of the data center, the needs of the enterprise. As Diane showed you, we're projecting a strong CAGR based on these trends into the future. You can ask your question when you're trying to reconcile the presentations why Brian only showed it through 2016 and Diane showed it through 2017. That'll be fun to watch them figure that out.
I think we're going to 2017 now, yeah, for sure. Our revenue next year, our revenue growth is expected to be well into the teens, and we think that we'll actually see operating margin expand. Our goal here is to have operating margin grow a bit faster than revenue. I'm going to shift gears now to talk about our software and services group and then our NAND business. They each get one slide, and then I'll jump into the Moore's Law section of this. Software and services, you heard a lot about that from Renée. It's a $2.5 billion business. McAfee is the vast majority of the revenue. As you heard from Renée, McAfee's executing well in a down PC market. Under the hood, we're seeing this strong shift that she talked about towards enterprise security solutions.
To put that shift in perspective, a few years ago, we saw that mobile and cloud would've been relatively small. Today, mobile and cloud for new sales, kind of half of the new sales in this segment. The other significant business in this segment is Wind River that I talked about. As we combined Wind River in with our Intelligent Systems Group, and we have this focus on deploying solutions as far afield as Mars, we believe that we can provide value for our customers here, and we're expecting a pretty significant increase in revenue for Wind River, which will drive a noticeable increase in revenue for the software and services group next year. Last but not least, a business that's very close to my heart is our NAND business. Here, we are expecting to see significant revenue and profit growth in 2013.
Looking forward, we expect that NSG, at a top-line level, will grow pretty significantly. It'll be well over a $2 billion business. To put that profit chart in perspective, it's several hundred million dollars of operating profit that it's generating this year. The market driver for this business is what we've been talking to you about the last several years. It's the demand for compute storage in the cloud and the enterprise. What we've seen in this business is that over the last several years, we've been on this journey to successfully take advantage of our technology leadership in the NAND space, just as we have technology leadership in our own factories, such that almost all of our shipments now for this business are in these higher value segments of the market. There's been, and you can really see that in terms of the financial results.
If you go back to 2008, significant loss in 2009. As we've executed on this shift, you've seen not just the top line grow, but we've seen significant profit growth. All right, in the next section, I'm going to talk about our cost advantage being driven by Moore's Law. I'm going to start this by re-grounding you on the operating principles that we've been talking about the last couple of years. Actually, I bet if I go back, some of these were introduced by Andy more than eight years ago, which is when I became CFO. If I were to summarize these operating guidelines that we've talked about, it's first off that we get lower costs by advancing Moore's Law at a rapid pace. I think you all get that. That the faster that we can advance it The more benefit we get.
We're not limited by our desires here, we're not limited by the economic benefit of moving from generation to generation. We're just limited by how fast the rest of the industry can keep up with us and enable us to make these transitions. Second, I shared several principles with you over the year that really get at a couple things. They show our responsiveness in how we deploy capital consistent with expected unit growth. I've shown you how fast that we can respond to changes, and that's a real shift, I think, in our velocity over the last five years and if you need to, you can go back to last year's presentation, I did a case study. We can actually move quite fast now if we see a shift in demand to realign our capacity to that demand level.
We've shown you over the years that we do put a slight bias in our planning systems to make sure that we can respond to upsides. For our business, we get a great cost structure when we're loaded in that kind of 80%-90% range, but it also gives us the ability that when there's a segment of the market taking off, or we have an initiative, we have the ability to respond to upsides, and there's a very high NPD associated with having that capability. Last, I shared with you how complexity and capital intensity is going up for everyone in the industry. Yet how our leadership in advancing Moore's Law looked to be extending our lead relative to our competitors.
Not in spite of the increase in complexity, I think it's actually because of the increase in complexity in advancing Moore's Law, our leadership is extending. These are all true today. In fact, as I think about the world today, I think they're probably more true than they were even a year ago. Today, I want to build on this with one additional observation. That additional observation is that I'm going to show you why I believe that the increase in capital intensity will not result in an increase in our unit costs, nor will it lower our company's gross margins. Now, here I have to put a disclaimer. There are other things that can impact gross margins, right? Like what we're doing in tablets. That can have an impact on gross margins.
A rapid shift in mix has an impact on gross margins, although we mitigate that by having a segmented product line-up. Coming back to this point, I firmly believe, and I'm going to show you some real data here, that the increase in capital intensity will not cause an increase in unit costs, nor will it cause a decrease in gross margin. At least as far as the line of sight that Bill showed you continues to exist of us being able to bring that cost per transistor down generation to generation. He showed you his expectations through 10 nanometer. He gave you a very strong hint to seven nanometer, which I'm a little surprised he did, but it helps my story a lot.
If you remember last year, we talked about how our capital intensity, these are slides from last year, as measured by capital cost per wafer, is increasing at a rate faster than it historically has. That our scaling, which is the density of our processes, also improving at a rate faster than it historically has, thus enabling us to stay on the historical cost per transistor line. We said we expected to stay on this through 14 nanometer with strong line of sight to 10 nanometer. Bill now, I didn't realize his convention, he's now made that from a dotted line to a solid line. I guess that means our confidence has actually gone up. By the way, I'm going to show you his slide again here in a minute that shows that we're actually achieving all this.
I did realize as he was presenting that I had a material misstatement last year. That plus sign should've been a times. This is what happens when a finance guy does technology charts. That's clearly been the thing that's been impacting our valuation for the last year, and I'm glad I can come back and clarify that. I'm surprised that Bill, who had the line of the day of, well, we always start yield at zero, didn't point this out to me at some point, because he doesn't usually shy away from telling me when I don't quite get it. I obviously borrowed this slide from him, and it shows exactly what we've been talking about, which is that through 10 nanometer, we can continue to bring down our cost per transistor at the historical rate.
That means that for us, Moore's Law is alive and well, and it's still the economic engine that powers our business, and I'm going to click in on a couple of things here to really show you this at work. Now I'm going to show you real data, because Bill showed you a normalized chart, I'm going to show you real data that quantifies the increase that we're seeing in capital cost per wafer. What this chart shows is our equipment capital spending. This is equipment. This takes out any impact from space. Per peak wafer start per week for different process nodes. You can see here the significant increase that we see as we went from 32 nanometer to 22 nanometer, and then again as we go from 22 nanometer to 14 nanometer.
The top line driver of this, as you've heard from us in the past, is moving to multi-patterning and lithography. That's the piece that's going up the fastest. Just to calibrate this slide, because I know you'll all go with your rulers later and try to figure out exactly how big is that, I'm just going to tell you this time, the increase from 22 to 14, that's about 27% increase in capital intensity on this slide. That's maybe a little different from our theoretical. This is as we're looking at putting in our peak wafer, what is the peak wafer capacity, how much equipment capacity does that require? I do know that you believe Bill, even if you don't completely understand him. I feel the same way about Bill, by the way.
I also know from my conversations with you that you struggle to connect the dots here around how that cost per transistor curve equates back into our product cost and our gross margin. You say, "Well, I get that your cost per transistor comes down, but you have this significant increase in capital cost. We see it in your CapEx. We see it in your depreciation. We hear it from others in the industry. Help me again with how that doesn't destroy your gross margins over time." The simple answer here is that in a world where density is improving from generation to generation, it gives us the ability to shrink our die sizes, even as we increase the performance and the energy efficiency of our product.
The offset to this is that density, and that density should translate into smaller die sizes while still taking up performance at the historical curve. This is what this chart shows. It shows a 7-year horizon that the weighted average die size for the company comes down. Two things to note with this chart. The first is that the CAGR of that reduction, if you go from 2008 to 2014, it's on the order of about 7% reduction in die size over time. I know that the cynical among you, I'm looking at you, are looking at it and go, "Well, wait, that's an average." That's impacted by the mix of your products, right? As a proof point, I'll say yeah.
If you look at what happened in 2009 and 2010, that was when the first really high volume Atom products came in as netbooks. You can see the die size came down at a faster pace. As that segment of the market kind of flattened out, and as servers grew faster than the rest, you see that curve kind of bubbling up and moving back down. This absolutely can be impacted by the mix within the businesses. We've shown you this data before. Now I'm going to show you data that we haven't shown you before, which is isolating to one segment of the business so that we can take out that mix effect. It's frankly our largest segment of the business, so it should give you a sense of what's going on.
This chart is showing you the Core notebook die size in 2013 versus what we think it's going to be in 2015. Think of this in 2013 as a product line that's dominated by Haswell, all those machines that you saw from Kirk, which is a 22-nanometer product in 2013, versus a product line in 2015 that will be dominated by Broadwell, which is a 14-nanometer product. What this chart shows you is that independent of the performance and power benefit that we're going to get, we were able to drop the average die size for the segment by 29%. If you think back to that capital intensity slide that I showed you, the die size here is coming down by more than the capital intensity is going up. This is Moore's Law at work. Capital intensity for us is up, die size cost is down.
This is the economic beauty of Moore's Law. It allows us to deliver those improvements in energy efficiency and performance while still bringing the unit cost down. By the way, the importance of this is not just that we can continue to bring our costs down. Bill shared this slide with you. I just want to come back to it for a second. As we look at others that are skipping a node transition, it also means that as we target our designs at segments of the market where our competitors use the foundries, we believe that not only do we have a decline in cost, but out in time, we have a performance of power and a cost advantage relative to the people that we compete with in the marketplace, and that's a fundamental competitive advantage for us. There's one other topic I want to cover.
I know it's another conversation that I have with you all, which is, okay, we get all of that at a philosophical level, clearly at $11 billion of CapEx, you got to be putting in a whole bunch of unit capacity. By the way, Stacy, you've been a little wrong on the unit growth in the past. Just tell me, when is that going to destroy your gross margin? That's the other thing that we worry about. All right. Let me take you there. This is data we've not shared in the past, although I talked a little bit about this on an earnings call a couple of quarters ago. What this chart does is it breaks out our capital spending between capacity, development, non-manufacturing, and 450 millimeters. Let me give you some observations on this.
First, you can see that about $7 billion of our capital spending is for capacity. It's about two-thirds of the total. The next biggest chunk is that non-manufacturing piece. Think about this as all of the capital we spend to run our IT infrastructure. That's the biggest chunk of it. It's the office buildings that we build around the world to house Intel employees. It's the labs that we put in place to do development. There's a pretty big chunk of this that's the interest capitalization of interest, which is how the accounting has us capitalize some of the interest that we have on debt. The next biggest chunk behind this is the spending that we're doing on the 450 millimeter facility, or 450 millimeter, that's primarily facility spending.
Last is that green chunk, which is the slice of our capital spending that's going for research and development. This is, in any given year, a combination of equipment and space. It just so happens in 2013, it's being dominated by space spending because we're building a development facility in Oregon. That's a breakout of our capital spending. Take a second here and make sure you're paying attention. I'll give you some thoughts about 2014. I think generally for 2014, we're going to see pretty flat CapEx in total, and I think that that $7 billion that's going to capacity related also will be pretty flat year-on-year.
I'm going to come back to that $7 billion in a second and show you some math of why I think that's the right amount of CapEx to sustain the business more or less at the current size that it is. We don't need a lot of growth to sustain that $7 billion of CapEx. Before I move on, I also want to just directly make the point of how this CapEx hits the P&L via depreciation, because it may not be entirely obvious. Most of it does, most of the depreciation here does go to cost of goods sold. A sizable portion of it will hit below-the-line spending. For example, most of the non-manufacturing piece hits below-the-line spending, and almost all of the development piece, actually, I think all of the development piece hits below-the-line spending. It hits our R&D line.
To just calibrate that, about 20% of our total depreciation goes to spending, the rest of it goes to COGS. All right. Now let's talk directly about depreciation for a second. You can see here that as our capital intensity goes up, so does our depreciation as a percent of cost of sales. That shouldn't be a surprise to you. What is a little bit of a surprise, though, and what's pretty interesting here, is that over this time horizon, if you go from 2011 to 2013, and I'll show you some stuff on our forecast for 2014 in a minute here, what you're going to see is that overall cost of sales has been pretty flat. Kind of plus or minus $1 billion over this time period. It makes sense if you think about it.
The capital intensity curve has been driven by multi-patterning and lithography, as I said. That doesn't equate to an increase in the other costs of running a factory. It's the same number of wafers that you start. If you think about gases and chemicals, they may be up a slight amount, but not nearly up like capital intensity is. Labor cost doesn't scale as a result of capital intensity. In fact, all of those elements, we apply efficiency initiatives to, so they kind of all come down generation by generation. It's not surprising that depreciation as a percent of cost of sales goes up, but that does not necessarily equate to an increase in overall cost of sales. In fact, we've seen as our business has flattened out, so has cost of sales. The depreciation increases have not impacted cost of sales.
Now I want to tie this back to the prior page and just do some manager math for you. If we're spending $7 billion of CapEx on capacity, over time, that $7 billion of CapEx will equate to $7 billion of depreciation. A little emphasis on over time, because when you're putting in place a facility, it's a long-lived asset, so it can take a while before that happens. Over time, if we just spend $7 billion forever, depreciation will also be $7 billion. As depreciation as a percent of cost of sales kind of pushes above that 30%, you mathematically get to a cost structure that equates to a gross margin that's cleanly in that 55%-65% range that I've been talking about, without having to grow the revenue levels of the company.
That gives me, as I just do the manager math, it gives me the comfort that $7 billion is about the right amount of CapEx to sustain the business as we have it today. Okay? That doesn't mean that we won't have a big battle when you show me your CapEx forecast for next year, there you go. All right. Now I want to just talk a little bit about how we view capacity additions and how that equates to our factory utilization. This chart shows you the peak wafer start per week for different process nodes. As you can see here, our wafer start capacity has been pretty flat since 32 nanometer. I want to put together a couple of themes here. Wait a minute. Yeah, good. The green line there shows it's been pretty flat since 32 nanometer.
I want to put together a couple of themes here. Our capital intensity is going up. Our die sizes are coming down, and we're planning peak wafer start capacity that's pretty flat as you go from 32 to 22 to 14 nanometer. This gives us the ability to increase volumes while not increasing peak wafer start per week generation to generation. Said another way, that $7 billion that we're spending on capacity-related CapEx does not increase our wafer start per week capacity, but it does allow for some modest unit growth. There's one other piece of information that I think is important for you to kind of get the full picture of what's going on in the factory network, and that's our utilization chart. We've shared this chart with you in the past.
This chart, as much as anything else, gets to our ability to respond rapidly to changes in the demand environment, right? We can move pretty fast to keep utilizations high. You see actually that play out some in 2013. We had some underloading at the beginning of the year, as you recall. At the end of the year, we had some 14 nanometer capacity that we weren't yet loading because we were pushing out production. Even within that, we were kind of right at the 80% level of utilization. There's a lot of responsiveness to keep that going. You can see that utilization ran right at 80% in 2013, and we expect it to tick up just a little bit as we move into 2014. This gives us both a good cost structure and the ability to respond to upsides.
It also, frankly, gives us the ability to not build inventory, which is something that I personally try to not do. All right. Lots of charts to make three key points. First, our costs will continue to come down. The second is we're investing in capital consistent with our unit growth, which has been pretty modest over the last few years. Third is that $7 billion of capacity-related CapEx supports modest unit growth and a good gross margin at kind of current revenue levels. That feels like a pretty comfortable amount. On the next page, I'll actually show you gross margin. Here we are. You know our gross margin guidance for 2013 is right at 60%. That's where we started the year. That's where we still are.
It's down a couple of points from the year before, just because 2013 is the year where we see elevated startup costs. As we go into 2014, again, we have a little bit of a range around it, but I expect to be right in the middle of that range. The two most significant impacts as I think about 2014, and we'll give you the formal guidance in January, but the two most significant impacts, that point and a half of tablet NRE and contra revenue that I talked about earlier, offset by the fact that we'll see some good news in startup costs. Actually, I think the good news in startup costs will be a little bit more than that, but generally, that keeps us in the middle of that range. One last thought before I leave this section.
In October, I communicated I'd give you some more insight into the implication of the shift in production on 14-nanometer from what you'd expect to be the shift in production on 14-nanometer. What that does is it moves the period, in which we expect to see a steep reduction in startup costs from Q4, Q1, like we saw in 2012, early 2012, to Q2, Q3, like I expect that we're going to see in 2014. Additionally, in the first quarter, so I'm giving you a sense of the shape of the first half of the year. Additionally, in the first quarter, I'm expecting that we'll see some pre-production reserves on Broadwell. We're ramping Broadwell prior to its qualification for sale.
As I think about that shape of gross margin in the first half, I think that in the first quarter, we'll be impacted by the fact that startup costs don't yet start to see the rapid decline. We'll have some pre-production reserves on Broadwell. That probably takes us down a few points from the average. In Q2, as we don't repeat that reserve impact, and as we start to see the rapid decline in startup costs, I expect that the gross margin would snap back up to the average for the year. You'll see a little bit of lumpiness, I think, as we go through the first half, just based on the shift in production on 14-nanometer. All right, now moving into the last section, and then I will wrap up. Capital allocation philosophy.
My favorite part, always the shortest part of the presentation because it's been pretty consistent now for the last several years. Recapping our capital allocation philosophy, which I think you've all seen and hopefully you all love. First, we invest in our business. Over the long term, we think that the investments that we make in CapEx, the investments that we make in R&D are the highest net present value investments that we can make. Second, we use the dividend to create return for our shareholders. I think we have an industry-leading dividend program. Last, we use the buyback to return cash to our shareholders and to modulate our cash balances. We've been pretty specific. We've said that we target about 40% of free cash flow to the dividend.
Just looking at those, looking at the cash from operations, capital spending, and dividend on one chart, you can see that our business is generating approximately $20 billion in cash from operations this year. Out of that, we're investing about $11 billion in capital and paying out about $4 billion in dividend. In addition, year-to-date, we have done about a billion and a half dollars to repurchase shares. You can really see those capital allocation priorities playing out exactly as communicated in 2013. Just focusing on the dividend for a second. It's currently $0.90 per share. It's approximately a 4% dividend yield. For 2013, the dividend is running approximately 50% to free cash flow, it's a tad above that 40% longer-term model that I've articulated, it's still in a very comfortable zone for me. That doesn't feel like a bad amount.
It's just a little above the long-term number that we've put out there. As I said, I think we clearly have one of the best dividend programs in tech and likely one of the best dividend programs around. Here you see the dividend and the buyback together. Just focusing on the buyback for a second. Over the life of the program, we bought back 4.5 billion shares. Since 2010, when we reentered the market after the downturn, we bought back almost 1 billion shares. It's interesting when you look at this over a long time horizon, you can see how this plays out. Over the last decade, we've returned $85 billion in cash to our owners. I think that's probably about 60% of our free cash flow. Very sizable amount.
I'm sorry, 60% of our cash from operations returned to shareholders via the dividend and the buyback. Putting it all together. For revenue, we expect roughly flat revenue for the company. You'll see a lot of flats here, but there's a lot of movement under the covers, or I'll give you some insight under the hood. Under that top-line number, we see several important trends playing out. We expect the PC market to stabilize some, but it will be down year-over-year, as I showed you earlier, offset by strong growth in the data center. We plan that the investments that we're making in non-recurring engineering and contra revenue to rapidly grow share in tablets will lead to significant unit share gains in tablets.
We expect that gross margins in the middle of that range, 55%-65%, good news in startup costs, offset by the impact of the tablet enabling that we're doing, and we expect flat revenue, but within that, I'm sorry, flat spending, but within that, there's about $1 billion of shift from places where we're investing to places where we're investing less. If you go back to that chart I showed you that showed the pluses and the minuses, think of that as about $1 billion of shift in spending, which really gives you a sense of how we're repositioning the company. I'll just wrap it up here with tying back to some key themes, that we are repositioning the company to focus on this broad range of devices.
It frankly is a lot of fun to be leaning forward into these markets, going after them aggressively to win. If it computes, it does it best with Intel. We expect that over time, we do see unit and revenue growth in that combination of PCs and tablets. When you think about the client across those two kinds of devices. We expect robust top-line growth and profit growth from the data center, and our ambitions in the data center include more than just silicon, it's the software building blocks on top of it. Our manufacturing lead is extending over the industry, and that lead translates directly to our business results. It's our cost structure, and it's our ability to grow a foundry business out in time.
Over time, we will bring down spending as a % of revenue, but our investments today are critical to getting us into these new markets, you can see that shift in the investment priorities as we go into 2014. Finally, we're committed to continue to make investments that bring return for our shareholders, and those are the investments that we make in CapEx, the investments that we make in R&D. With that, I'm done. Thank you very much.
Let's go to Ross over here on the left-hand side.
To start with the "Once there was a man from Nantucket" line.
Please don't.
Anyway, onto the real question. The OpEx side of things being flat as a percentage of revenues, I still think you're a good five, six points above your target range there. Whether any of you can actually answer this, how do you expect to get back to that average range of about 30% of revenues in OpEx, and what sort of metrics can we watch where you either succeed and grow into it with revenues or you're actually failing, and we can expect some cuts on the OpEx side?
You want to start or you can start.
Sure.
Okay.
If you think about the transition in spending this year, a lot of it is going towards areas where we would expect revenue growth next year, in 2015. It's not that far off. The particular increase that's impacting both margin and a pretty big increase on the OpEx side is the investment that we're making in tablets. We have a goal to grow 40 million+ units in 2014, and then as we move into 2015, and we're having to do much less enablement around that because of the bill of materials and the integration that we're able to provide to our customers, we would expect to see revenue growth associated with that, and that'll start to bring down spending as a percent of revenue.
Yeah. I would've said it probably simpler. It has to be revenue growth. In fact, if we didn't have a projection that we'll grow revenue, you would have made other decisions now. Those are investments towards revenue growth. If you remember in the discussion, one of the things I said we're trying to do differently was really drive innovation to the marketplace. We've got to do that quickly, and we've got to do that in a way that generates revenue. I think if you ask any one of the GMs, they'll tell you that the pressure of generating revenue from this innovation cycle is definitely there.
Great.
Tiffany, right behind you.
Thanks. Ed Snyder at Charter Equity. Brian, you mentioned in your talk that you were focusing on the biggest OEMs and handsets now as a kind of a shift in strategy, which is puzzling because Infineon used to supply Apple and was also a big supplier into Samsung. I'm trying to get a grasp of what does that actually mean that you're focusing on now that you hadn't before? You must have been talking. Does it mean pulling foundry services and bringing a wider suite of products to try and compel them to start using Intel parts? How are things changed now that you're focusing on them versus two years ago?
Well, Infineon used to sell the modems to the Samsung and Apple and those, and they still do in some cases, Nokia, for example, one of their biggest customers. What's happening, especially at the entry level, is apps processor and the modem are combining, and into one single piece of silicon. Hermann talked about that. At the upper end, we're talking about the apps processor. The modems are typically separated in those. It's very simple. All we said was we've gone from a model that says, "We need to build our own phones. We need to show ourselves and the world that we can build these devices, and we're going to try and influence through the carriers," to a program that says, "You know what? No. That market is fairly stable.
There are half a dozen big players in there that own the majority of that market, we should just really focus on that." What that means is really going and working with them, understanding what the product requirements are, what they need from us to bring that combined apps processor and modem to them. It's exactly. You heard Hermann, he just landed back. We're having these discussions. Now, in some cases, they're actually getting excited about some of these products like SoFIA and asking merely pull it in another month. It's really that. Instead of going out and trying to. Renée talked about the 1,000 points of life in services, the same thing. Instead of doing that with phones, it's really focusing on these half a dozen suppliers. This side over here. Vivek.
Thank you. The question I have is, Brian, as you look out the next two to five years, what do you think is the sales growth rate for Intel? Where I'm going with that is we heard a lot of very impressive presentations today. Leadership in manufacturing process. The leadership you have in the data center, et cetera. Despite all that, you're guiding to revenue staying roughly flat next year. As you think about the company over a longer-term period, what is the right growth rate to target for the company, and how does that sort of dictate where you're putting in investments? Thank you.
Sure. We're not going to project actual sales growth rate out past next year. Stacy told you it's relatively flat, right? You can start to build a model that says, okay, let's just take tablets and the client market in general, and assume that our thesis of PCs are stabilizing and low single-digit declines or near flat over that time. Tablets growing somewhere in the 20%-30% and we're still nascent in that market relatively. Even at $40 million, we'll be 15% or something, somewhere around there. There's plenty of room to grow. Part of the revenue growth, as we said, was getting the right products. Right now, we have Bay Trail. We're reacting.
We're bringing Bay Trail down into these tablets, as we get to 2015 and we have Broxton and SoFIA, we're able to produce products that can go directly in there with the right price points and the right performance and the right integration. That now lets you play in a much broader market and generate much more profitability in that market. I look at it as you go into 2015 and into 2016, you can get a lot of growth there. We talked about Diane's segment is growing at 15%, Internet of Things at 16%, it's really around those areas that we should be able to grow. We're not going to go project what 2015 revenue growth is yet. Do you have any?
Yeah. I think you hit it. I would say for me, the biggest variable in all this is going to be what happens to the traditional PC market. We're working hard to create all this innovation. If that's successful and you have something that's flat or slight unit growth versus something that's kind of on a slow unit decline, you end up with very different answers.
Okay.
I think the difference in the company is that we're leaning forward. We're going to have great products. Wherever the market goes, we'll be there. You'll end up with a different overall top-line growth rate for the company if you have a PC market that's flattening out versus one that continues on the decline.
Great. Trey?
Thanks. Mike McConnell, Pacific Crest. I wanted to ask on the foundry strategy, seems like there's a pretty material tone change looking forward. I understand now the flexibility with respect to new architectures and different products, and even if the products compete against your own, you're willing to fab them out. From a customer standpoint, how do you convince these customers, these prospective customers to share their IP with you, even if they're currently competing with you? It would seem like that would be a major hurdle given that they're currently competing with you.
You mean the customers that are foundry customers?
Yeah, your potential customers.
The non-custom IA customers.
Correct.
Right. Well, like all foundries and lots of other projects, we have clear separation between our teams and our development for our own products. I don't know if you would add anything. We firewall and we protect our customers' IP. We do that in co-developments today. We've done it for years and years and years in our co-developments, both in TMG and in the software group where we have other people's source and IP. It's pretty standard for us to do that kind of thing. Is there something you want to add to that, Brian?
Yeah. I've been involved with Bill, especially on some of these engagements over the last few years. I'm yet to find one where the discussion has been around protection of IP as the most critical or concerning thing. Most of them, we show them exactly what Renée just said. We have firewalls we've put in.
Dedicated teams
yeah, dedicated teams, dedicated lockdown areas for where their data goes. How we do genericizing of masks so people can't see it, what's going on, even the demand forecasting, we are able to isolate so the rest of Intel doesn't necessarily see their demand. The bigger issue has up until now been, are you really in this business to stay? We've dabbled our toe in the water here before. I think several of you asked or have made comments, "Well, it felt like you guys were 20% in in the past. Are you going all in this time?" I think that's been the bigger question. People want some kind of a guarantee or a discussion about are you really going to do this. Because they trust us. We're a high-integrity company. We show them the detail and the technical level we've gone to isolate their IP.
What they really want to do is make sure we're going to stay in. I think that's the bigger question we've had to overcome, and I think over the last two years, we've done a very good job, and part of what Renée and I are trying to do with going more vocal, going out today, for example, and saying, "Hey, we're in this, and in fact anybody who wants to come talk to us, come talk to us. We're open," is to really continue to establish that no, we're not going anywhere, and in fact, we want this to grow. That's been the bigger concern, I'd say from people.
Stacy, how do you allocate fab overhead costs for the DCG group?
How do I allocate fab overhead costs for the DCG group? Typically, we don't have specific factories for groups, right? Which is, I think, a core advantage for us. The same factory will start a lot of Xeons, will start a lot of Centertons, will start a lot of Core, will start a lot of Atom. The overhead of the factory just gets allocated to the wafer, and then based on how many wafers each group starts, they pick up their proportion of those costs.
In fact, on some of the products, there's the same types of-
Same die
same wafer can produce all of those.
Right.
It just, at the end of the day, you get the number of units, and you simulate that as wafers and charge them.
Yeah.
Thanks very much.
Kirk would say Diane never gets charged enough, though.
Diane would say.
Diane would say, "You're right.
While we're waiting for a mic to get over to this side of the room, we'll come back over here. Joe.
Going back to the foundry question, Brian, you kind of opened the door to working with anybody. How literally should we take that? There are situations where you have direct competitors where it seems like you would hurt yourself more economically, versus the foundry revenue you could get, versus the damage you could do to your own business by making their products better. How do you think about those trade-offs?
I think at the end of the day, you think in the shareholders' best interest, and that's the way we should make all of these decisions. I think as a shareholder, you'd rather have me open that door and at least have those discussions and see what it leads to than to automatically close the door. That's really what we're saying is, "Hey, come take a look at this technology.
Take a look at what we're doing here. We've got to make sure we keep the shareholder value up, though, we're not Andy was talking to somebody earlier and saying, "We're not going to go into the low-cost foundry business here where we erode margins." For most of the people who are looking at the leading edge, if you take a look at Bill's data, even if we charge them the same as what they're getting charged today, they're going to get better performance, probably get lower power, better yield, and quality. I can either take that delta of his cost per transistor or share that with them, pending on the business agreement we want to make. There's room here for both people to do better as a result.
I just want to have those discussions before we just say, "I don't want to talk to you.
Thank you.
Great. Let's come over here.
Thanks. I guess longer-term manufacturing question, probably for Brian or maybe for Bill. EUV. Does Intel need EUV? Does Moore's Law need EUV to extend it? If not, why not? If so, what generation do you think you would need it? Can you just give us an update on the progress? Do you think it's going to work? Do you think it's not going to work? Are you guys looking at other alternatives, such as double patterning, quadruple patterning? Is there any capital intensity issues there?
I'll let Bill-
We'll go to Bill Moore's Law Holt in the front row here.
Yeah.
We're looking forward to using EUV when it makes economic sense. The sooner that is, the happier we'll be about it, we don't see it as a showstopper for the next couple generations. If we have to continue with the double patterning approach, we'll continue doing that. It still looks like the most promising of the technologies. It's the one that we're looking at most seriously. It has a ways to go, it's making good progress.
Thank you.
Do you think that'd be fair?
No, I 100% agree. All of that is purely an economics discussion.
All right. We'll come back over to Mark on this side of the auditorium.
Thank you. Simple question. Stacy, you showed the same slide that Bill had up. It showed 35% cost advantage on 14 versus 16 FinFET, and then 45% cost or scaling advantage on 10 versus 10, you guys versus TSMC. Does that area scaling necessarily translate to a direct cost saving? Is it, for example, if we're doing a competitive analysis of your chips versus Arm chips at TSMC on 14 versus 16, can we just simply say, "Well, that's a 35% cost advantage because they have a 35% die size advantage"?
Yes. No.
Not-
Bill says no.
Not necessarily.
My answer's better.
Spoke like a true finance person.
Would you like Bill to correct it?
Yeah, we'll let Bill correct me if he wants.
Come on, Bill, please.
All right.
Then I'll clarify it back to yes.
In the limit, let's just take this. If you compared a previous generation to the next generation, it would have an even bigger area reduction, right?
That area reduction is not all cost savings because that wafer, the other graph, got more expensive. Since we don't know exactly what their wafer cost is, we can't tell you that that area is all going to translate into cost. It's possible that some of that is, in fact, lower wafer costs. The predominant answer should be yes, most of it should translate into a definite cost advantage. If it would've been two generations of our own technology, the scaling would not all be cost savings because the wafer does get more expensive.
You also lined up the time period, and we tend to be a year to two years ahead. I'm still pretty confident the answer to that-
That's adding another variable in that we weren't-
Right. I think the answer directionally is going to be yes. Bill's more technically right, this should translate into a pretty significant cost advantage.
The real question you should ask both us, Bill showed it, so did Stacy, and anybody else you're trying to compare it to, is cost of the transistor cost. At the end of the day, none of us sell wafers. Bill made this point, I believe. Nobody sells a wafer, virtually nobody. We all sell some bundle of transistors, and those transistors don't really care, from a cost standpoint, where they and how they were built. They have a cost. What you really want to do to be able to compare apples to apples or understand the deltas is those costs per transistors.
What's the answer to that question? Yes?
I think Bill answered it pretty well, where it says the majority of it does turn into a yes. It's not a linear one-to-one.
Thank you.
Great. Let's come over to Stacy on this side of the room.
Stacy Rasgon, Sanford Bernstein. Thanks for the discussion on how unit cost can continue to go down even as capital intensity can go up, you've got more units per wafer driving it down. Doesn't that actually imply an increase in your unit capacity? If your notebook unit die size is going down by 29%, wouldn't that imply that your notebook unit capacity is going up by almost 30% over the same time period? As you start to sell more of the Bay Trail and more of the lower-end products, those have even lower cost, you need even fewer wafers to generate the same number of units or correspondingly even more units to generate the same number of wafers.
Would this not be the case that your requirements for unit growth going forward are going to be higher than they have been in the past, even though your core business seems to be at least at best stable and at worst maybe declining a little bit? How do you think about, I guess, managing your investments, and your cost reductions in that kind of environment? Is this the reason, I guess, for opening up the foundry business more in order to look for more volume to capture these wafers? Just any comments on that would be very helpful.
Do you want me to-
You can start and I'll add it. Go ahead.
Okay. There was a lot in that question.
Yeah.
Let me just start it by taking it back to a couple of charts I showed. We're constantly balancing the amount of capacity we put in place for unit growth. To your last question, we don't need the foundry business to fill factories. We can fill factories just by modulating the amount of capital that we put in place in any given generation. I think you've seen us being very responsive to that in pretty short periods of time. I also showed you data that said we're running about 80% utilized, which should give you comfort that we're not seeing a big utilization impact. I was pretty explicit that says we've got a situation where our die size is coming down, wafer starts per week is staying pretty constant, and that does give us the ability to grow units modestly.
Now that 29% was over a two-year horizon, right? You've got some, if you look at it on an annual basis, there's some modest ability to grow units, and we'd modulate our capital spending if we needed to bring that up, or if we wanted to bring it down. We would just buy a little more or a little less capital.
Is it more appropriate to say it's not necessarily an ability to grow units if you want, but a requirement that you have to grow units?
No.
No.
I was going to answer, because you took a serendipitous route to. The simple answer I would've said is the foundry strategy is completely separate from fill your factory or something like that. This is purely a discussion about the foundry side of, hey, we really think that we can go make a business out of utilizing our leading-edge technology and get paid well for it. We're looking for shareholder return against the R&D investment that we've already made. The discussion around how much capital do you have to have, and naturally, you can go through the math, and you're right. Given mix stayed the same and/or we added more Bay Trail than Xeons and all that kind of stuff, if you had a decreasing wafer size or a die size, you would use fewer wafers, and you would see a declining demand.
We'll manage through that on a different set of issues. At least in the foreseeable future, there is plenty of demand. Even if you assume a modest decrease in the PC business, there is enough wafers between server growth, IoT or the ISG growth, the fact that we're going to go into the tablets, all of those other things, there's enough wafers to sustain the factories as we have them.
Got it. That also implies that you have to add capacity to do foundry.
Thank you.
We can talk later if you want.
David Wong, Wells Fargo. Brian, you are offering your fab with no boundary, which puts some of us in a quandary and leads us to ask, is your next two-year task to get a fruit in your foundry?
Was that a haiku?
It could be.
It could be.
It's the basis of one.
Actually.
I feel one coming on.
No.
My serious question.
I understood your, yeah.
OEMs. You've said that you prefer OEMs. In the past, you've said you're really keen on having OEM customers in your foundry. Is there a preference over OEMs than chip companies when you're considering your list of options of people who approach you?
No. You said one thing that I want to make sure is clarified and isn't. You said you're opening up your foundries with no boundaries. I still said it has to make sense for the shareholder, right? This is still about, we said we're not going to chase the low end. We're not going to get into a price war with TSMC or something. We believe we have something that genuinely differentiates us in this silicon and this performance. That's what we're really opening up. We will open that up to anyone who can take advantage of that and wants to pay for that. We do not have a preference for whether it's an OEM necessarily or a chip company. In fact, the minute I start separating that, I've immediately started telling people not to knock on the door.
I'd rather have a discussion with everybody, because there may be a business deal for almost anyone that's a win-win.
Great. We have one more over here and then we'll come back to the middle.
I'll try to ask Mark's cost question a little bit differently. For the last several years, we've come to Analyst Day and you guys have had fairly impressive roadmaps of where you will be 12 months from now with higher integration, lower cost. We saw the same today with Bay Trail going to Broxton and then SoFIA. Every time you seem to hit those targets, the market's moved or the competition's moved. I guess my question is, as we look out again to the end of 2014 going into 2015, why are you more confident that these new products are going to be competitive relative to the Arm camp or what the market wants than you were over the last couple of years? It kind of does go back to Mark's question.
Where do you think your cost structure is going to be against the Arm camp, especially to go and exploit the low end of the market?
Sure. I'll answer that. Fundamentally, take a look at our products, especially on the client side. We actually didn't do much of that integration. Right? We don't have Wi-Fi integrated. We don't have NFC integrated. We don't have GPS integrated. We don't have most of the sensors integrated. We don't have comms integrated. Yet we've had those individual Wi-Fi I was talking to you at lunch. We've had Wi-Fi for 12 years. Five times it's been on the roadmap to integrate. Five times it got taken off the roadmap for other right decisions at the time. We can't let that happen. You guys need to hold me accountable, and Renée. Don't forget Renée.
No one's taken anything off the roadmap in this game, though.
You need to hold me accountable because those got taken off the roadmap each and every time. That's what's different. We said at the very beginning, there was a portion of this around, Andy kind of said, around bringing the discipline and results orientation back, and we really have to stick to it this time. That's what's different, and we tried to show you that we're trying to be very pragmatic about it even and go way outside the box with something like SoFIA. Ripping out an Arm core, putting in an IA core, and doing it at a foundry is not something you would've seen on the roadmap two years ago. That's something literally by the next Investor Day, my guess is, we should be close enough to say, "Yeah, the launch date is on this day." If not, we have some explaining to do.
That's a pragmatic focus on time to market, because your question was really, it wasn't that the products-
Renée, your mic is-
It wasn't that the products weren't the correct definition at the time, it was that by the time they got there. I think this pragmatic focus on time to market and speed.
Why don't you hang onto that? We'll come up here to the middle.
Okay. Just two simple questions, one for Renée. Why isn't the software business profitable? For Stacy, you mentioned the dividend at 50% of free cash flow is not alarming. At what level does it start to become a concern for you? Thanks.
You want to go first?
No.
Inside the software segment, as Stacy said, the largest portion is McAfee, which has been largely flat. Then on top of that, you saw a whole bunch of acquisitions and other things that we've invested in on a forward-looking basis for cloud. It's basically a break-even segment when you put all of that stuff together, and it's all combined into that segment.
I just say, I'm not going to give you an exact number, I won't maybe answer the question to your complete satisfaction, but I think you can see it play out in 2013. We're generating on the order of $20 billion of cash from operations, $11 billion of CapEx, rough numbers, that leaves you $9 billion. The dividend's about $4 billion. It gives us plenty of cash to grow our cash levels, to participate in buybacks. We're miles from anything that gives me pause. We're a business that generates a lot of cash. We are capital intensive, but we generate a lot of cash, and we generate cash in excess of what we need to invest in the business. Let's hit Chris in the middle of the room here.
Thanks. Just a question with regard to perhaps, say, contingency planning. It seems that a number of the assumptions that you guys are making over the next year, the strategies are based on the assumption that the PC market has stabilized to some extent, declining somewhat, but rather stable. What happens if that assumption is incorrect? I guess into one part of it, what sort of levers can you pull, in terms of cost reduction, maintaining the margin structure? Then strategically, what sort of things would you perhaps need to do if that were the case?
I mean, how long? I mean, that answer is pretty wide open. It's how long are we? Clearly, we have options, right? We could cut spending additionally. We could reduce CapEx. I think I would be careful about some of these investments around the emerging areas like new devices or Internet things or the server area, the data center, we're growing at 15%. I want us to really be careful. This integration, right? I mean, that's what's happened in the past is the crisis of the year comes along, we suddenly cut funding in order to get the integration done, we look around two years later and say, "What happened? Why didn't this get done?" There are certain things I would not do, and I would not change, and I would hold a very firm no to making those changes.
Certainly, if we needed to and if demand, we've shown in the past, Stacy has shown how we can cut CapEx, how we can cut factory spending, how we can cut operational spending, all of those things very quickly within a quarter. Remember, our throughput times in most of our factories now are very quick, we can adjust that plus inventory in less than a quarter, and respond.
Even in a dramatic and very tight-end time event like the financial crisis in 2008, 2009, we were able to realign the capacity in the factories, roll forward equipment, bring inventory levels down within about six months. Even in a real event-driven thing like that, we can respond fast, and that's a significant change for our business. If you go back five, 10 years, that six months would've been more like six quarters to get things aligned, and that's real work that we do in the factories around responsiveness and having forward reuse of equipment and all of those things that we bore you with year after year in the investor meeting. Those are the things that give us that kind of responsiveness in the factory network.
I do want to reiterate, right? What you really want us to do is understand those investments and not let those go away to get into those other growing markets.
Right.
If we do that, then that's the death spiral, and that's what we're going to avoid.
All right. With that, we've actually run out of time. I realize we left a few questions unanswered, and what we'll do is we'll work real hard to get you connected, with the executives for whom you had questions.