Good morning. Let me add my welcome to Mark's. We really appreciate you coming out here, particularly those on East Coast time, and getting in late last night and getting up early this morning for us. I'm going to kick it off and just talk about what I'm going to tell you, some key messages over the course of my presentation. First one is that our investments and the actions that we've taken over the last five years are, in fact, delivering sustained growth and are positioned to give us even more growth going forward. I'll tell you more about that. The second is about this notion that in terms of our potential for our markets, is really about the ubiquity of computing. This growth of computing in all kinds of manifestations is really providing unprecedented opportunities for us as a company.
The third point I want to make is that if you look at what we're building, in terms of on top of what we've had as our core business over the last number of decades, we're building a capability to develop and deliver integrated solutions to our customers to allow them to go to market faster and capture market opportunities that are in front of them. The last thing is that I believe we are developing a set of unique assets that will give us a very competitive solution, increasingly differentiated competitive solution, over the next decade, and I'll show you how I think that evolves. In terms of the first point, though, I wanted to take you back, and I don't expect you to read these slides.
I want to take you back over the last six presentations I've made to you in these meetings, going back to 2006. I pulled one representative slide out of each of them to tell you the story of the journey that we've been on for the last six, seven years. In 2006, you remember, it was April of 2006 in New York City. I got up and said, "We're going to go through a fairly large restructuring of the company." This is well before the financial crisis. Intel was still hyper-profitable. Yeah, there were some concerns about our competitiveness in terms of Athlon coming on and Opteron coming on. We believed then, and it's turned out, I think, to be quite prescient, that we had to restructure the company for where computing was going. We did that. We did a lot of that work.
We've talked to you about the SET project over a number of years, and basically made the company very competitive, put in place new benchmarks for spending and development and systems across the board, and really have enhanced our sustained capability for profitable growth. In 2007, we talked about, for the first time publicly, the tick-tock model, the basic development model that allowed us to take advantage of Moore's Law in a very consistent, predictable fashion and move our products forward year after year after year. In 2008, it was the first time we talked about investing in new areas for growth and where we were going there, the growth opportunities were beyond the core business. In 2009, we showed you the roadmap for that.
It was all about extending Intel architecture from our key position in computing, in PC and server computing outwards on both sides, down into more mobile devices and up into the higher ends of mainframe, and now, as we would call it, cloud computing. In 2010, we told you about how we're going to glue all these things together, that the vision wasn't just to deliver chips into each of these segments, but rather to weave them together to deliver consistent experiences through software integration, through feature set integration across all the spectrum of computing. Last year, we talked about how we are significantly adding capabilities inside the company to go faster in some of these areas, specifically in the wireless area with the acquisition of the Infineon wireless business, and in security with the acquisition of McAfee. That's the 6-year story.
I'll spend the rest of the day talking about where we're going to go, but I thought it was important to put it in perspective in terms of where we've been the last 6 years. All of that has delivered, in fact, sustained growth. Last year was our best year ever for the company, again. $54 billion. We've grown $20 billion in top-line revenue in the last 2 years. Last year's revenue versus 2010 was up 24%, and it wasn't a one-trick pony. 17% growth in the Data Center, which I think you all expected, but 17% growth also in the PC business last year. $4 billion of the growth last year was from the acquisitions from Infineon and McAfee. Very committed process for sustained growth, both through some acquisitions, but principally through organic growth in our core business. That's growing profits.
$17.5 billion of operating profit last year, 12% growth there, 19% EPS growth, and I'll talk a little bit more about that in the next slide. We have really been focused on making sure that it's not just the top line that grows, but also both aspects of the bottom line, that we get profitable growth in operating income and also paying attention to the structure of the company in terms of overall EPS growth. Stacy will talk more about that this afternoon. In terms of cash, we remain committed to delivering returning cash to our shareholders. Dividend yield is now above 3%. This week, we had an announcement of a 7% increase. We're up to $0.90 a share dividend now. That's the third increase in 18 months or so, and one of the top dividend stocks now in the NASDAQ-100.
Beyond that, though, we are also committed to returning excess cash to the shareholders through share repurchase, and you can see the curve here. The net of this is that in the last 10 years, we've given back $80 billion to our shareholders. Very strong focus, not just on profitability, but on the capital structure and returning cash to the people who own the company. That's where we've been. Let me talk about where we're going. To do that, I want to give you a perspective, or my perspective, on computing. In the 1980s and the 1990s, as the PC emerged, it was really about productivity. The PC essentially took the place of prior machines. It replaced typewriters, it replaced calculators, it replaced slide machines. That was about personal productivity. The TAM in the mid-1990s for computing was about 100 million units a year.
If you fast-forward to the middle of the 2000s, the TAM was about a billion units a year, an order of magnitude growth. It was about portability. 2005 was two years into Centrino. Intel had popularized Wi-Fi around the world, standardized it, stabilized it, and made possible the world of high bandwidth wireless communications. That allowed us to do those productive things we did with computers in a highly portable fashion. By the time you get 10 years later, 2015, computing has changed again. It's about ubiquity. It's about cloud-centric computing. It's about devices that are of all kinds of manifestations that are connected to the internet, and the TAM has gone up by yet another order of magnitude. This number is probably low. I've seen other numbers that show 20 billion and 30 billion units. It doesn't matter. It's a big number.
This is the TAM that we are aiming at in terms of the kinds of products we build that touch every aspect of that, of where computing is going. If I blow that up and I look at where computing is going this decade, I think there's four key areas for where Intel is focused. The first is about the cloud and data center, the second is about personal computing, the third is about mobile devices of all types, and the fourth is about intelligent systems. I'll cover each of those in detail here. Let me start with the cloud. The cloud is all about data, and it's about big data, and it's about data growth. This plots out just digital information created. The scale is exabytes. You can see the curve here.
We are sort of halfway through the curve or right at the knee of the inflection of that curve. Today, as we speak, we're getting 7 exabytes of data created every day. Inside of that, there's 17,000 HD movies being created every day, in terms of the capability, not the movies themselves, but the kind of capacity that's being built in terms of data that's out there. On the right side, you see what happens every minute. 60 hours of video, 6,000 songs, 170,000 photos, unlimited amounts of email. All of what everyone on Earth is beginning to do is creating data of many types. Much of that data going forward is going to be unstructured. Historically, computing has lived in the world of structured data, and this is what big data is all about.
You can see structured data continues to grow, but it's dwarfed by where individuals principally are creating data and posting it onto various sites. This explosion of data creates an explosive demand for transistors over time. In 2005 to 2015, storage will grow by 25x, and storage is a big part of our business. One of the reasons we renamed our enterprise group several years ago from the server group to the Data Center Group was that we are focused on not just servers of all kinds, but also storage and networking equipment. All of this needs Intel processors at the end of the day. Cloud computing, as Diane will tell you later on, is growing by 100x over that time period.
The number of transistors required to do all of this stuff is growing by 200x per year by the time you get from 2005 to 2015 in this timeframe. This is a lot of data, a lot of transistors, and a tremendous opportunity for Intel going forward. How do we serve that? We've created a very broad and very deep product line in the enterprise data center computing area that scales from, well, let's say teraflops to milliwatts. At the very high end of it, we're developing a product that is codenamed Knights Corner. We've demoed variants of that to date. This product is a multi-core machine built on 22 nanometers. It is a teraflop on a chip.
That addresses the need that we see going forward for high-performance computing, Diane will drill down a bit into that in terms of the way we see that market growing. The second one is about our volumes server business, which is Xeon-based, and we have a new version of Xeon that's out there today. It's twice as fast, much more energy efficient, very high ROI than the Nehalem generation that it replaced. That's providing very good growth for us in the core data center business. We're not blind to new trends. We believe that we are one of the leaders in microservers. In fact, if you look back at the early instantiations of microserver white papers and those kinds of things, they came out of Intel. We believe that there are certain workloads that are highly optimized for a microserver kind of environment.
We want to lead in that area. Microservers do scale down to milliwatts when you use the Atom processors, but many of our customers are also building them with Xeon processors in them. It's a very exciting growth segment. It's not going to be the entirety of the server market. In fact, we think it's single-digit % of the server market, but it's very good business for us, and we want to be leaders here. That's the hardware side. What you've seen from us over the last several years, though, is an increasing focus on the things that glue the data center together. I wanted to point out just a couple of recent acquisitions. You saw acquisitions in Fabric. You saw us acquire the assets of the Cray Fabric team to help work on this. You've seen us acquire Fulcrum for networking.
We're moving into storage with our solid-state drives, flash-based solid-state drives, and increasingly providing the data center software for doing things like node management and console capability inside the data centers in a very open fashion. It's all about delivering not just the best chips, but the best platforms, the best solutions to our customers globally to be able to allow them to move into these new areas and take advantage of the markets. Diane will talk to you a little bit about the change in the customer base over time. Servers, which we had seen a consolidation in terms of the number of players for many, many years, are now expanding again. Because a lot of the standards that are making servers so affordable are now enabling new companies, particularly Asian companies, Chinese and Taiwanese companies, to be able to move into this market very aggressively.
They are serving those markets. Diane will show you data on the explosive growth of servers in China later on. The next area is personal computing. You've seen us talk about the fact that much of the growth in personal computing is going to come from emerging markets. I thought I would just show you the top 5 PC markets in rank order from 2010. You can see the U.S. was number 1, China, Germany, Japan, and Brazil. We've been predicting for some time that China would pass the U.S. In fact, it has. That was last year, 2011. China became the number 1 market. Brazil became the number 3 market. Yeah.
You fast-forward to 2016. Our line of sight is that, yeah, the U.S. is still on there, it's number 2. The other 4 countries are the BRIC countries: China, Brazil, Russia, and India. It's the 1st time that India's shown up on the slide as a top 5 country, because this has been a market that has been sort of a constant five years away from growth for a long time. Now we see real traction as they're starting to deploy broadband communication infrastructure in India, which is a precondition to high-performance ubiquitous computing. I will also point out for the skeptics in the audience that we have essentially no dependence on France. Our growth is going to be very good independent of what they happen to do there. Last year, we talked about the need to redefine the PC.
I talked about this in the context of the fact that our industry had spent much of the last decade in PCs making things cheaper, maybe a little bit smaller. Mostly making things cheaper. That helped drive volume, which is very good. It also took away innovation, it took away creativity. We thought it was time. The number of changes, the capability of what we could put into a single chip or a chipset was going to alter the fabric of computing, allow us to change not just the form factor of it, but the way we interact with machines for the 1st time since the Xerox PARC days of the mouse and the keyboard. That was about inventing the Ultrabook.
The notion of this product line is not just about form factor, it's about a number of things. You'll hear roadmap comments from me and from Kirk in a few minutes. It's about building an ecosystem to produce these products in very high volume, drive innovation, and keep the price points very low. That just begins with an enabling of silicon. In April of last year, Dadi Perlmutter, who runs our Intel Architecture Group, redid our entire roadmap. Essentially moved the sweet spot for power and integration from what had been the mainstream inch-and-a-half thick notebooks that we all carried around to increasingly lightweight, long battery life, high-performance machines. By moving that center point down, we altered the landscape of the products that are going to be built around Intel architecture.
That allowed us to have a kickoff event at Computex in Taiwan last year in the April-May-June timeframe. Sean Maloney's picture is here with Jonney Shih from ASUS. Thank you, Johnny. Sorry. We created an ecosystem development environment where people came in and talked about how they could enable lower-cost batteries to be able to handle these long life cycles that we need in terms of the products, smaller drive capacity, lower-cost drives. Just all the milling and the machines to be able to build these things had to be able to be moved from a niche product line to a high-volume product line. That was that enabling of the ecosystem that happened mid last year. Intel Capital committed $300 million for an Ultrabook fund. That is not money that's going to our customers.
That's money that's going to ecosystem investments around us, people that are enabling lower-cost versions for things like touchscreens to be able to bring those to the marketplace. By holiday, we had over 20 SKUs shipping last year. They were all Sandy Bridge-based for the most part. We're now tracking over 110 designs. The bulk of those are all based on our next-generation silicon, which is Ivy Bridge. I'll talk more about that on the next slide. Inside of a year, we've gone from idea to very high volume concept, and we still are on track, as I got asked in the last conference call, we are on track to meet our goal of 40% of the consumer notebooks this holiday season being Ultrabook. I wanted to talk about the Ivy Bridge, which is our third-generation Core microprocessor.
We announced this product publicly a few weeks ago. It's a great product. It's a tick in the Intel tick-tock model, 22-nanometer product. It's shipping in very high volume. We'll exit this quarter, the June month, shipping at a rate of a little over 2 million units a week. That puts us on a track to have the majority of our microprocessors, in other words, Ivy Bridge crossing over Sandy Bridge, by the fall of this year. As I said earlier, it's not just about form factor, it's about innovation. We have, coming into the market this year and then a roadmap of feature sets over the next several years, really exciting innovations that change the way that human beings interact with computing.
Obviously, things like security are going to get more and more important, increasingly, particularly with Windows 8 coming on, touch starts to be a major enabler in mainstream computing. As you'll hear from Kirk, we think voice is the next big thing. Voice and gestures, to be able to use the other senses that human beings have to be able to communicate with computers and make them do things for us. I thought I would just show you a couple of examples of this. First one is an Ultrabook from Lenovo. It looks like an Ultrabook, standard form factor here. It's called Yoga. The neat thing about this Ultrabook is it's also a tablet. Oops, I hit the wrong button, sorry. It's also a tablet. Still very thin, but all the capabilities of what you'd want in a PC. Another one is from Compal.
It looks like a normal Ultrabook, except that it's an Ivy Bridge tablet when you pull it out. Compal, as you know, is an ODM, and they have derivatives of this design that are going to go into a number of OEMs' branded systems for this year. We've got customers really innovating on form factor. This, to me, is one of the more exciting things about the Ultrabook story is that for, as I said earlier, for about a decade, PCs were about cheaper, cheaper. Now with the concept of the Ultrabook and with the capabilities that we're building into them, we've, I think, seen a new energy level coming out of our customer base that I haven't seen in quite some time.
Tens of thousands of engineers in the United States, in Taiwan, and China are working on their own versions of these things, and it's created a competitive dynamic inside our customer base that is really quite refreshing. People are competing to see how thin can they make the machines, how light can they make them, how large of a screen can you cram into a small frame, what new features can you put in. Then also addressing how do you hit volume price points. This is not about niche computing. I think this is very exciting. You'll see over the next year and two years, a lot of innovation here that is really, I think, unprecedented. That's why I've said a couple of times that I don't think we've seen the end state of computing. I don't think the Ultrabook is it.
I don't think tablets are it. I think something in the middle, these convertible designs, are likely to represent the next big thing and the next stage of the evolution of computing because they're no compromise. They give you exactly what you need. Increasingly, they're going to give you the ability to choose what you need in very, very different kinds of form factors. The third area is mobile devices. We are a leader in mobile device shipments today in the communications arena by virtue of our acquisition of the Infineon wireless business a year or so ago. We have leadership in low-power RF, very broad customer base as you can see from the list of logos up here. Last year, we shipped over 400 million units into this marketplace, and it added $2 billion of revenue to the company last year.
This is principally, though, in the comms modules or comms processor side of the phone business. We also understand quite clearly that the future is not just about the comms, but also about apps processing, and increasingly about the integration of the two. We've been focusing on that. This year, as you're well aware, we're now in production shipment of our first smartphone chips based upon the 32-nanometer technology. The product was codenamed Medfield, you may know it as that. The product is a Z2460, catchy name. We're getting awfully good reviews from our first phones. We're very proud of that, but we're not stopping there. We intend to be a leader in this business.
We intend to bring the best of Intel architecture to it, the best of Intel architecture not just on the comms side and the computer architecture side, but also on the silicon side. You've seen a number of announcements in the last six, eight months around this point. Beginning on the right with the Google announcement at our Developer Forum last fall, where Andy Rubin got up and said that essentially going forward, there'll be two branches of Google, of Android, one on various Arm processors and one on x86 or Intel architecture. Intel does a lot of the work around that.
Renée's group in the software area is a major contributor, not only to just the Linux community, but also to now the Android community in terms of developing code That reflects the feature set of Intel architecture, the capabilities of Intel architecture, and embedding that into the x86 or IA version of Android. We also announced something with Visa around mobile payments. We're building capabilities in the machine. The companies we've announced with so far are Motorola, ZTE, Lenovo, Orange, and Lava, which is a carrier in India and is now shipping the first Intel cell phone. You can see that during the break later on today. We're just getting started here. We have ambitions. You'll see more announcements over time, and very cool capabilities being built into the phones.
Even though I believe that the end state of computing is not necessarily the tablet, it's a big market today, and we are not oblivious to that. We are shipping Android-based, Atom-based tablets today, and we will be ready for Windows 8 tablets when that operating system is ready later this year. I'm particularly excited about Intel architecture on Windows 8 tablets because we think it's a differentiator. I know there was a lot of debate for some time that this is going to be a real entree for the Arm camp into Windows for the first time. While at face value, that's true, I think they have a big uphill fight against what we're doing here. We have the advantage of the incumbency, the advantage of the legacy support, not just in terms of applications, but devices.
We have all the capabilities of Intel architecture and our ability to scale and power that I think is going to be increasingly difficult for Arm-based competitors. The net result is we have over 20 designs from 10 OEMs for tablets that will launch with Windows 8 when it comes out. If you've played with one of these things, it's pretty cool. You get all of the Metro interface capabilities, very snappy, great form factor in the tablet. With one button, you can get to legacy mode. That's critically important for CIOs who want to preserve all their investments in software that they built for the desktop over many, many years. It's also critically important for end users who want to take the thing home and plug their XYZ camera into the tablet.
All those drivers that are written for the PC are just going to work on an Intel-based Windows 8 tablet. I think there's going to be some compatibility challenges for other architectures there. We're very excited about this. Stay tuned for later this year. I said earlier we're going to apply not just our architecture capabilities but our silicon capabilities to these markets. We're shipping now in 32 nanometers. Next year, you'll see phone and tablet chips at 22 nanometers, and the year after that, you'll see 14 nanometers. Essentially, moving our phone chips to catch up from the trailing edge of our Intel processes to the leading edge of our Intel processes over a three-year period. The net result is that our customers are going to see us moving at twice the rate of Moore's Law in this space.
It is this silicon roadmap, combined with the architectural and compatibility capabilities we bring, the ability to optimize things like Android, as I talked about earlier, that is winning designs out in the marketplace. People see this. They have confidence in us. We'll talk a little bit more about the silicon later on, but these things are not too subtle, very large blow-ups of our Tri-Gate transistor that Brian's going to walk you through later on to give you, in case you needed Transistor 101 knowledge. This is what these are. We make them a lot smaller than this, by the way. But as this stuff scales and advances, you'll see increasing differentiation on Intel-based mobile products. Remember, the leading edge gives you three advantages. You have the lowest cost, you have the lowest power, and you have the highest transistor performance.
That's a pretty good set of assets to bring into any market, but in particular into the mobile computing market. The fourth area I wanted to talk about was intelligent systems. This is an area that for decades, the industry and we called embedded systems, and it was called embedded systems for a reason. You built microprocessors or most principally microcontrollers to put them into these devices. They took 18 to 24 months to design in. It was a terminal or industrial controller or a kiosk or an ATM machine, and those devices did the same thing over their life cycle. They never changed. The software was essentially set. The use model was set. They weren't really terribly connected, and when they became obsolete, they were replaced. The world moved, and the world's moved to what we call intelligent systems.
In fact, we've renamed our business unit from the Embedded Systems Group to the Intelligent Systems Group to reflect this change. What's different about this? What's different about it is it's reprogrammable, it's connected, it's highly intelligent, it's analytical. It takes advantage of everything that we've built in mainstream computing and brings it down into new areas. I'll talk about three of these in particular, about three branches in automotive, in retail, and in communications. In general, this is a $2 billion business for us this year, and probably going to grow at 25% in the next year as these Atom designs kick in. Very strong business. We're very happy with our positioning here. As I said, I've got three areas I wanted to talk about where we are selling not just a point solution in all cases.
In the automotive area, we've announced publicly before our engagement with Daimler. Intel Atom products will be the in-vehicle infotainment systems in the new S and C-class cars from Mercedes and from BMW as well. We've also recently announced Kia, Nissan, and Toyota. What's different here? The world of automotive electronics, particularly the, let me say, the non-mission-critical areas. We're avoiding drivetrain and braking systems, and those kinds of things for a bunch of reasons. We're focused on the connected part of the car, the part of the car that's going to be driver information systems, navigation systems, entertainment systems, communication systems with respect to connectivity, we think is really a ripe area for our product line. BMW has something like 10 million cars already connected through their network, so they can actually talk to their cars.
I've been told that in Germany, when you drive by a dealership, your car actually uploads information to the dealer. They know the status of the car, can send you alerts, and so forth. The world's going to become more and more like that as these cars become smarter and more connected. That's an opportunity for us, not just at the chip level in terms of selling an Atom chip into these machines, but also at the module level, the software level. Wind River is a major player here in terms of the operating system. We've developed standards around this. The automotive industry is now moving to a standards-based computing methodology for the first time, not just delivering their own independent machines, but leveraging an industry standard that's called GENIVI.
The second area is retail. This is about digital signage, about analytics, about customer loyalty, and about generating sell up. Those of you who were at our CES keynote this year saw a demonstration from Adidas. In the U.S., we call it Adidas. They call it Adidas. We did digital signage, and they changed their whole store methodology to have this very interactive capability for the end user to walk in. The signs recognize who you are if you're one of their customers and allows you to basically browse and try things before you actually try on the shoe. Then there's a lovely sell-up capability built into that.
That demonstration captured a lot of attention. We've seen design wins recently from Macy's, Lego, Kraft, and Coca-Cola to do similar things, all of whom want to use the ability to connect either a vending machine or a digital sign to the internet, to the stores, to inventory management, to who you are as an end user, to be able to leverage that to have a better buying experience and a better sell-up experience for themselves. This is an area where we'll sell analytics, we'll sell servers. We'll be able to sell the chips that go into the signs or the vending machines themselves. The third one is an area where I'm very excited, where we're seeing, let me say, the last instantiation of or holdout of the RISC architectures really start to be displaced. This is the communications infrastructure, particularly in the wireless world.
As you know, for decades, this was the home of Sun machines for quite some time. As every network operator in the world strives to have lower costs and higher bandwidth, and more flexible infrastructure, they're moving off of those proprietary systems, and they're moving increasingly towards our Xeon-based systems. These give them the advantage, obviously, of performance and power and low cost, but this flexibility, the same flexibility you've seen us bring to the data center in terms of a consistent architecture for applications processing, for storage, and for network processing, is now available to them. You're seeing major re-architecting of networks around the world. The three customers that have announced so far are Korea Telecom, Huawei, and Verizon. I can tell you there's a lot more behind that, including major telcos in China also working on these kinds of transitions. It's very good for us.
It gives us a strong position in their wireless infrastructure. As the world moves to client-aware computing, where the networks increasingly understand what the device they're talking to is, what the capabilities of it are, they can tailor their applications, their services to the device. Having Intel end to end in that construct is a very powerful tool for us. The last thing I wanted to talk about is gluing all these together, what we call the compute continuum. This is about, from our perspective, not just single-point devices and all those elements, but a scalable architecture, teraflops to milliwatts, to use an expression. We want these experiences to be engaging, we want them to be consistent, we want them to be secure, and we want them to be aware.
That very often, as you think about a cross-device environment that does those things, is a job for not just hardware, but for software. This is what the SSG group is really all about, the Software and Solutions Group, Software and Services Group is all about, is providing the infrastructure to glue these things together. We think we have a very powerful advantage here by having a single programming model for all these devices. At the end of the day, the millions of developers in our developer program and the tens of millions of applications that are written around the x86 architecture are highly leveraged in this environment.
Of course, we think one of the major opportunities we have is that we are, I think, still the only architecture that scales all major operating systems, and they're listed down here below, from Apple to Microsoft, to Google's various ones, to Linux, and then to Tizen, which we're developing for mobile computing and for in-vehicle infotainment. I wanted to give you one example of how this works. This is a product that was recently released from McAfee called All Access. You buy this as sort of a family pack. You, as the purchaser, get to put it on multiple devices, and you have parental control of what those devices can do. The devices, PCs, they span PCs, tablets, phones. At this point, it remains heterogeneous. That is, it's not just Intel devices. It works on iPhones and somebody else's tablet and somebody else's PC.
Our plan is to continue that because we think having that broad base of products that we can attach to is important. Our plan is also to make sure that when it's talking to an Intel device, it's a better, more secure, more convenient experience. This is the product efforts we're doing deeply with McAfee in terms of our co-development. The last topic I wanted to cover is transistors. We've talked about you can't have a presentation from Intel without a picture of Gordon and without Moore's Law. This, as you all know, is about higher integration, more efficiency, lower cost, and better performance. I thought I would just draw a contrast between the first microprocessor and our most recent one. The 4004, 1971, and the third generation Core, Ivy Bridge, 2012. Why is it important to continue to scale Moore's Law?
When you look product to product, Intel generation to Intel generation, there are advantages, but they're relatively small in scale. When you take a step back and look at it over a 30-year+ view, the product's 4,000 times faster, 5,000 times less energy per transistor, and 50,000 times cheaper per transistor. That's why Moore's Law is important. That's why we keep driving this. That got me to thinking about what's happening in our industry. If you'll indulge me, I'd like to take you on a brief history of the semiconductor industry from my perspective of someone who's lived through much of it. In the '60s and '70s, when I started, it was only about integrated device manufacturing. The Silicon Valley was still. There was a lot of silicon here. Not anymore. I don't think there's any fabs operating in the Valley anymore.
It was about inventing silicon process technology on the fly, very often building the equipment you needed to build the wafers to manufacture the products, and deep co-collaboration with the design teams to be able to invent the things you need to make the product work. Every company in the industry was an integrated device manufacturer. You fast-forward to the '80s and things got a little easier. We had evolved to the planar process. We'd evolved to CMOS as an industry. There were common equipment manufacturers out there. I don't want to say the recipes were cookie cutter, but they were closer to cookie cutter then than they were in the '60s and '70s. You saw a lot of explosion in the business.
You saw the emergence of fabless for the first time, enabled by foundries, which were companies that would grow up, replicate these kinds of processes, and allow the multiple customers to be able to use them. That was principally for the reasons of convenience. It was also for the reasons of cost, because fabs were starting to cost a lot more than they did in the '60s and '70s. You had to fill them up. Not many companies could fill them up, so you had this division of resources. In the 2000s, though, it started to shift again, and you saw what I call the first cracks in that curve, where a number of the integrated device manufacturers, AMD, Freescale, TI, started moving towards a fabless model for a variety of reasons, but very often to take advantage of the cost reductions associated with this.
I think that was a necessary move. I don't think anyone did it by choice, because you give up a lot. Let me talk about the kinds of things you give up before I give you where I think where the industry's going. Let's plot out several generations of Intel silicon here. We started strained silicon in 2003 with 90 nanometers, and we've been shipping multiple generations of that ever since. We had to do that because simple scaling wasn't enough. Simple scaling was possible through lithography, but we had to be able to manage performance on the transistor level and power at the transistor level.
Strained silicon gave us a means to do so. We had a three-year lead on that is now other companies have started to use strained silicon, but that early lead gave us a significant advantage. 2007, with 45 nanometer, we developed and launched High-K/Metal Gate. We still believe we have about a three-and-a-half-year lead on that. There are some people shipping now, first High-K/Metal Gate, but we're on our third generation of this, and we've shipped over 1 billion units now with this kind of technology. This year, we introduced these puppies, the Tri-Gate transistor. We believe we'll have about a four-year lead on this kind of product. It took us something like 10 years of research to develop this. This is not easy anymore.
Our technology development group which Stacy will show you, spends about $2 billion a year, has 1,900 PhDs in it. A lot of other engineers, but 1,900 PhDs. Increasingly advancing Moore's Law isn't just about money, it's about invention. It's about material science. It's about creativity. It's about altering standard equipment to be used in ways that no one else knows how to do it. That's the secret behind the kinds of things we do, and I think as a result, you're seeing this become increasingly rarefied. On the right side of the slide, you see the metrics just between 90 nanometers and 22. What changes? A simple measure of the complexity. Twice as many process steps, the database is six times the size, and the transistors on the lead product are 10 times more.
You get advantages, but the complexity, the difficulty, the cost, and the required invention goes up generation after generation. When you couple that with having your design resources, your product design resources, deeply integrated into your manufacturing and vice versa, you get a reinforcing spiral of advantage towards an integrated device manufacturer. We share common goals, Brian will talk more about this later on today, but there's a number of areas where we can decide where we want to optimize things. For example, we try to optimize our products to have most of our yield loss, everybody has yield loss, we want most of it, if we have any, to be at sort and not at the final product, because it's cheaper. You don't have to take it through that whole assembly test process to do that.
If you're a foundry and a fabless company, there's this huge economic tension between where you have that yield loss. We can decide that for the best of Intel. It's tougher for other people. We have shared development goals. We can decide early on that we want to go after the mobile market, and we can change our silicon process technology to provide a wider dynamic range of transistor performance to be able to address not just the high-performance parts of the market that are traditional to us, but the low power needs that we have for new markets. We have this process of continuous improvement, where every day, we get feedback on our product health that feeds back into our factories, that feeds back into transistor changes, that feeds back into continuous improvement, and better costs, and better yields in our product.
You cannot do this unless you're an integrated device manufacturer. As a result, what you've seen is a consistent cost per transistor reduction out of Intel. Stacy may delve more into this, and I think Brian will as well, but we're now projecting that this continues. Certainly true at 22 nanometers, and we're now showing you a first point for 14 nanometers here. You may have seen commentary in the industry from some of the foundries that show that this curve is very tough for them. In fact, there are significant price increases now being talked about as they move from one node to another. To some extent, price follows cost for the foundry model. I think what you're seeing is the bending of this curve in the wrong direction.
As invention gets harder, you throw more resources at it, you throw more equipment at it, and therefore, the cost per transistor becomes more difficult to continue on this curve. My contention is that we will increase our capabilities and our advantage going forward. You add on top of that, the scale that's required to operate in this business. In the 200-millimeter timeframe, factories cost about $1 billion each. I can remember sitting through a strategic long-range plan conversation that was led by Gordon Moore, sort of circa 1984, where Gordon predicted that there would be a shakeout in this industry when fabs hit $1 billion. Right? Well, there wasn't. As usual, Gordon was right, he was just a little off in time on this one. You're starting to see the shakeout now. Fabs cost over $5 billion today for a 300-millimeter factory.
As you look forward towards the end of this decade, the first 450-millimeter factories will cost over $10 billion each. Stacy will show you the economics of that, of running one of those factories later on. They're quite expensive, you need lots of volume to fill them, and you certainly don't want to be in an unfilled factory situation for this kind of investment. Where's the industry going? Well, I think everything we've seen up to date has taken us on this curve. Integrated device manufacturers pushed up the curve of capability, of margin, of return on assets, return on investment. The 1980s and through thousands, the era of the fabless and foundry companies, was more of a straightening out of that curve. The learning curve slowed down. Today, I think we're at an inflection point. The industry's going to go in probably two different directions.
I think much of the industry will move towards increased consolidation. You've seen some of this happening with the various foundries that are out there or people exiting the business. The ability to operate on the leading edge will be harder and harder, so people will be on N minus X technologies, and that means that their costs will be higher, their capabilities will be higher, their customers' products will be less competitive. You move towards an era or an environment that's much more commodity-driven than it has been in the past.
On the other hand, for those companies that can take advantage of this opportunity, that have the economic scale, that have the volume scale, and have the inherent ability to invent and deploy, I think there's a golden age ahead of us, where the IDMs, in particular, are going to have a growing advantage that we haven't seen since the '60s and '70s. The best transistors and the highest level of integration. Obviously, I think that Intel is going to be one of the ones on the top. I think that there may be only one or two others up there. It's going to be a rarefied air, just given the economics of this industry. We intend to be there, and we'll drive it through innovation and investments. The R&D pipeline is quite deep. I talked about 10 years to develop these guys.
Well, not these guys, but their silicon brothers. The invention continues. Those 1,900 PhDs are working on future technologies as we speak. To enable us to bring up 14 nanometers on time and on target. We're putting a lot of money into the ground in terms of investing in the future. 22-nanometer factories are up and running now, or in the case of Israel, coming up. They'll be on three sites in Oregon, in Arizona, and in Israel, and we're in the process of construction and initial deployment of equipment into 14 nanometers and beyond at Oregon, Arizona, and in Ireland. By sum all this up, we will continue to deliver value to our shareholders and our customers through silicon technology, through architectural innovation, and through the brand of the company.
My takeaways I hope you pick up from this are that Intel is incredibly well-positioned for the next decade of computing. We're reinventing computing with the Ultrabook. We are capitalizing on the explosive opportunities in the data center excuse me, and the internet growth. We're increasingly bringing the best of Intel technologies to mobile devices, phones, and tablets. Secondly, our technology lead, which is already large, is going to continue and it's going to give us an increasing scale advantage as we go forward, and we are relentlessly focused at returning significant value to our shareholders. With that, I'm done. I'd like to introduce Kirk Skaugen, who's the general manager of our PC Group. Thank you very much.
Good morning. It's great to see a lot of familiar faces out there from my Data Center days, and just personally, I'd like to say I look forward to continuing that in the next couple of years on the PC side. I got to agree with Paul that the excitement around the PC industry is, what I think, unprecedented as I've looked around and traveled around the world in the first 90 days on the job. I hope I can share some of that enthusiasm with you over the next 30 minutes. Four things I want you to take away with today. Number 1 is, with the Ultrabook, I truly believe we are reinventing the PC.
The momentum we've seen today with 21 designs in the marketplace, growing literally just a few weeks ago from 75 and then to 100 that we announced in the earnings call, and now over 110 today, I think I want to share with you that that momentum and the excitement in the industry is very, very high. Combined with that, we're just now kicking off a few weeks ago, the largest marketing campaign Intel's had since the Centrino launch, and I'll share that with you. Touch and the combination of touch with Windows 8 is going to come across the PC.
We've had it in the All-in-One desktop category, but we're rapidly seeing it being adopted in Ultrabook computing and in a new set of convertible Ultrabooks or Ultrabook convertibles, where we're seeing some of the things, and I'll show you some of the new designs that we see coming up that Paul showed you a few of. Lastly, Haswell. I'll give you a glimpse into Haswell. We're on track for Haswell in 2013. This is the first chip that we designed grounds up with the Ultrabook in mind, and I'll share with you some of the new things that Paul was talking about around voice, gesture, and others that we think that that will help enable in the industry. First and foremost, where do we see growth across the PC industry?
Ultrabook category, I think we'll spend a bunch of time on today, really reinvigorating the PC and delivering a no-compromise experience. If you look in the desktop space, we've also seen a new All-in-One that for the last couple of years, I don't think people realize has been growing over 35% a year, and we think that within the desktop space this year, we'll be continuing that now and into the future. In vPro, this year, we'll be crossing our 100 millionth vPro deployment into business. We're delivering the ultimate in security and manageability so that we can deliver those kind of capabilities to the enterprise space.
At the Intel Developer Forum in Beijing, we announced a new product line called Intel Small Business Advantage, which is taking that and really simplifying it for the small business, putting a set of applications together that enable them to remotely manage PCs. If there's software vendors like Renren in China that have an e-commerce application, they can download patches overnight to their small businesses across China. Really making PCs more manageable and more secure for small business. This is enabling sell up. Intel Small Business Advantage is going to be available on Core processors, so it's enabling us to sell up from Celeron and Pentium to Core, and vPro is Core i5 and above. Part of this rich mix we've had to Core i5 and above processors has been because of some of this level of innovation.
If you just look at why we're excited about Ultrabook and the refresh that we think it drives, just in consumer, just in mobile, if we were able to pull in that refresh of someone who keeps their notebook around for about four years by one year, just get people a little bit more excited that touch is coming or these kinds of things, $5 billion that gets driven into the industry. This is just the bottom line of why I think our OEMs are excited and why you're seeing the innovation that we are in the industry. As Paul said, this is just really from Computex a year ago, and now on the second-generation Core, over 20 Ultrabooks. As we look forward, Time was calling 2012 the year of the Ultrabook, and we're really just getting started.
We were retrofitting products, I would say, on second-generation Core. I'll show you about the performance leap ahead we're going to get as third-generation Core hits the Ultrabook. As we look forward to this third-generation Core announcement a few months away, literally 110 or more new designs or five times the number of designs that we have today are coming, not just in consumer, but in business, not just in clamshell, but in a host of convertible form factors that I'll share with you in a few minutes. What is the promise behind Ultrabook? When we announce the third-generation Core, we're going to tell you that we're going to get another 20% performance at 20% lower power. All those transistors that Paul was talking about, we're applying to new things like graphics and media as well.
We've exceeded our expectations on third-generation Core, we're delivering twice as much graphics as we did just a year ago. We had originally said 70%. I came out a week or two ago, said we're exceeding that with a tick plus because we're really re-architecting the graphics engine. At the same time, we're changing the process. We've had this tick-tock model for a long time, but as you look forward this year, we're delivering a tick plus on graphics, delivering 100% better graphics. On media, I've said on the data center side that media across the Internet, as you look at that data that Paul talked about, is about 50% of what goes across the Internet. We talked about more Internet traffic in 2010 than the entire history of the Internet combined. That's going to go to about 90% media.
Most of us are comfortable taking pictures and kind of moving them around on a screen. What we're seeing is more and more people are doing that with high-definition video. When we launch third-generation Ultrabook, we're going to have 23 times better media than a notebook that's just three years old, 23 times. You can basically start moving high-definition video around on things like these GoPro cameras that you see on skateboarders and skiers and things, just like you move around pictures today. Literally, a PC about three years ago really struggles just to even load the systems into the machine. Performance is a key for both the CPU, for graphics, where we're doubling from last year, and on media, where we're going to be 23 times better. We're literally delivering a no-compromise experience, even though we're making the form factor much, much thinner.
On responsiveness, I'll show you some demos here in a second, we've been waiting to boot up PCs for a long time, right? We've been opening the lid, and it's taken a long time to get out of standby. We haven't had fresh data on the PC like you do on a phone when you open it. With responsiveness, we're literally delivering instant data the second you open up your PC, and I'll show you that in a few minutes. Form factors, we're really just at the beginning. I'll share with you how we're making things thinner, going from inches to millimeters. As we look forward to 2013 and beyond, things get much, much thinner, and we're seeing, I think, unprecedented mechanical innovation in the industry as well in a few of the convertibles I'll share with you.
We've talked about this left brain and right brain. People absolutely want to consume, and that's been great on a tablet. They also want to create, whether you're doing homework, or your report, or you're a reporter typing on the keyboard. We think a mechanical keyboard has been very important. People also want peace of mind. They got the left brain and the right brain, we're putting capabilities into Ultrabooks for anti-theft, and I'll give you some examples of that, to do the best we can to protect malware from entering the system. Also protecting your identity in new ways with hardware-based identity protection technologies that are underlying in the hardware. Whether it's theft deterrence, better secure e-commerce, or protecting you from malware, the most secure experience will be on the Ultrabook.
We're moving towards all-day battery life, as we get to Haswell, basically being able to remove the brick and truly not have to have your power strips, and maybe that'll be our objective for next year, no power strips in the aisles. Obviously, driving down cost in the system so we can get this into mainstream price points, and I'll give you flavor for where we think things will get by the end of the year. What I'd like to do is just kind of drill into the responsiveness category for a second and show a demo on some of the things we're doing to make sure that not just you're going to get the great performance out of the PC, but you'll get great responsiveness. Let me invite Mark up from our demo team. How are you doing, Mark?
Pretty good, thanks.
Good. Why don't we show the folks what we have here?
Okay. What we've got is a second-generation, which is a currently available Dell XPS 13 Ultrabook, we're going to show the RapidStart technology that's been enabled on this platform.
Okay. Why don't you power it down? We're going to put this system into hibernate, not sleep, but in the lowest hibernate state.
After a couple of seconds, it's basically flushed the memory to disk, and now it's basically in a power-off state. You've got standby time here that can last for weeks.
Okay. Why don't you hit the power button now? Typically, I would ask you to go to coffee because it took about 30 seconds or more to actually power on a PC. As you can see from the Dell system
Did I hit the button?
Hit the power button. Better. The other one. Okay, there we go.
Okay, sorry about that.
Usually, you'd have about 30 seconds, and literally on the Dell system, you're going to get up in under seven seconds.
You can-
We'll practice hitting power buttons later. We're talking about four to five times faster, and really this is driven because there's now non-volatile memory in the platform for the first time. It's driving that flash technology into the platform so that you don't have to go wait that kind of time. On the server, the buttons are bigger. On the server, the buttons are bigger. That's right.
I'll have to get smaller fingers for this going forward.
That's Rapid Start technology, and it's in the Ultrabook. How about if we're just coming out of a sleep state mode?
Here, we're using a technology called Smart Connect, and basically, this laptop's been sitting up here throughout the presentation. What it's been doing is it's been waking up periodically and downloading emails, and it's able to update Twitter feeds and Facebook feeds and things like that. That way, it's got the information that you need when you're ready to go. If we go ahead and open this up, it's basically in sleep mode. As it wakes back up, you can see our email application is open there, and we've got email messages in here. We've actually been sending photos to this machine while it's been sitting up here. All of your documents, your attachments and things like that just get downloaded.
If your laptop's sitting in your briefcase or on your desk somewhere, you can just grab it and know that you've got the latest information you need.
Fantastic, Mark. Thank you very much. Rapid Start getting basically 4 to 5 times faster from a deep sleep. From sleep mode, you saw that wake in about 1 to 2 seconds and getting your Twitter, your Facebook, your email now instantaneously into your device. As we get forward to Windows 8, this will only get better with things like Microsoft's Connected Standby. What are we seeing in terms of innovation in the industry and what's driving these mainstream price points? When we first started the Ultrabook category, we were at inches, and I'll get into more detail in a second, but we've driven this thing to 18 millimeters and below in terms of thinness. We had thin machine aluminum, we had full SSDs to do some of these technologies.
What we're seeing now is more and more innovation in actually the plastics technology, so you can get the same rigidity in these very, very thin systems, but much, much cheaper costs. Taking 50%-80% of the cost out of some of these systems. You can see the general trend here as you go from machined aluminum to plastic composites, going from full SSDs down to caches with hard disk drives and solving some of these issues. What that means is you're going to see dramatically lower costs, and I'll go into that in a minute. Let me just drill down to some of the innovation that we're seeing here from the industry that's building and knowing this wave of Ultrabooks is coming.
We went from way over an inch to now under 18 millimeters, and I can tell you much, much thinner as we look towards the end of the year and into next year. The way we did that is starting working with the screen vendors. We had a Taiwan symposium with hundreds of Taiwan companies and Chinese companies looking to thin out the screens from 5 millimeters to 3 millimeters. We're going from cylindrical batteries to prismatic batteries and taking nearly 12 millimeters out of the battery technology. We're going from hard disk drives to SSDs. We're going from pin grid array socketed parts to ball grid array and basically eliminating the socket, which brings down the Z-height.
We're doing technologies in our heat sink development, moving from optical disk drives to either no optical disks or working with companies like Panasonic to get several millimeters out of the optical disk drives. All this innovation is kind of rolling, knowing that Ultrabooks are coming, and this only gets tougher as we go on, but I think we see the innovation to get to 18 millimeters, certainly much, much smaller as we look at some of the hybrids in the future. Inches to millimeters is kind of the goal here. What that's driving is more and more volume and more and more innovation. There's a baseline set of features we're putting into Ultrabook.
These are around the performance attributes of having Core processors, the responsiveness of things like the Rapid Start and Smart Connect technologies we showed you, things like security features around anti-theft and identity protection, then obviously the thickness of the device. We think by the end of the year, we'll be seeing mainstream price points where you originally saw $999, $1,099 class systems will drive down to mainstream price points at about $699. There's going to be a tremendous amount of innovation and differentiation depending on which OEM you're working with. We think we'll move up front and add standalone SSDs or have cache and non-volatile memory variants in the platform. We'll have very high-resolution screens and high def. We'll have touch being added across a variety of platforms, and I'll show you some demonstrations of what that means in a second.
A wireless WAN integrated like LTE into the platform, new convertible form factors, as I'll talk about in a second, adding vPro capability for business, so you don't have to compromise between a thick notebook for business and a nice thin notebook for your consumer. Now, in order to meet the Ultrabook spec and get the sticker, and I'll show you what we're going to do in retail in a second, Intel has kind of a certification test we've had, just like we've had on vPro. We're kind of ensuring a baseline of quality for Ultrabooks overall, then we'll enable the OEMs to innovate above and beyond that. In a few days, we'll be announcing vPro technology for third-generation Core, and we do think that vPro will be coming to Ultrabooks.
I think we just had our enterprise board of advisors with our top CIOs around the world. They're incredibly excited about this. Obviously, when we launched Ultrabook, we were very consumer-focused, but we're basically now going to have a no-compromise experience for business. You're going to see notebooks and Ultrabooks, HP announcing one yesterday, their intent to do one yesterday, where you have everything you love about a consumer Ultrabook today, but add the security, the manageability, and the stable platform where we're committing to corporate that will keep these systems with stable images around significantly longer than our consumer products. Put that all together and you'll get a no-compromise experience for business. I think this is going to be incredibly exciting for the business market and as we look at consumerization IT is a key thing that we're hearing from the CIOs.
On top of that, we've got Windows 8 coming. Simply stated, we're working incredibly close, as we always have with Microsoft, with Ultrabooks and with third-generation Core. We believe third-generation Core will deliver the best Windows 8 experience. I'll share with you that in a few minutes in terms of how we're seeing that. We think we'll have optimized applications that'll be clearly differentiated and scale with the Core processor architecture on Windows 8. Most importantly, we'll be delivering touch onto a large number of platforms. Today, we have over 30 touch designs in the marketplace. Out of those, 110 is growing quite rapidly as we start working with the touch vendor community.
We went around to about 80 families in four countries, in China, Italy, Brazil, and the U.S., really just started watching people and getting feedback on whether touch on a clamshell would be interesting on a traditional notebook. It turns out the feedback has been overwhelmingly positive. We ran a few more studies and it got even more positive. We're very excited about what this is going to deliver to the user experience. If you think about it, we've been interfacing with a mouse for a long time, and we think that touch is going to be a big deal coming later this year. Rather than talk about it, let me bring up Mark again and we'll show you what we have here. Mark, what do we have?
Yeah. What we have, like you talked about, Kirk, is this is basically a reference design and it's a software development sort of platform. It's using the third-generation Core, and we've delivered thousands of these to developers to help us develop great experiences on an Ultrabook. What more so than just a regular Ultrabook because, of course, we've got touch on this one, and so it allows us to develop experiences that work well in touch as well as in keyboard and mouse mode. In fact, I'm using Windows 8 here right now, and if I wanted to, I could type in my password, but Windows 8 actually has the ability to do a more gesture-based login. I can log into my desktop, and there you can see the Metro UI, and I can touch and play with it if I want to.
I can pull up different apps. It's nice because I've got the keyboard too, so I can actually bounce back and forth between different applications and modes. Whichever way I want to actually interact with a computer, it's nice. I've got touch, but then I can do other things as well. One of the other things we can do is obviously play applications. We actually have, for example, a game. I'll tap to get through a few of the banner screens that always precede games and get into it. Again, really easy interaction in terms of being able to interact with it from a touch perspective, and this game's kind of fun.
In order for the audience to actually see the interaction, normally I would probably two-hand it because you can play it with your thumbs, but I'll ask you, Kirk, to play the right flipper, and I'll do the left one. Again, you can just kind of see the type of interaction where we can just touch.
I'll do better.
Touch the screen. Don't let me down.
We'd probably play for 30 minutes and never lose.
There you see it. Touch provides a really nice sort of set of interactions, and I can quickly, because of the performance of the third-generation Core platform, I can quickly move back and forth around to where I need to. Especially what's nice is I still have access to my keyboard and my mouse. If I want to use precision applications like a spreadsheet or a word processor, or in this case video editing, I can go right to it, and then I've got the precision that I need to actually interact in that way, too. It's kind of the best of both worlds.
That's the compatibility mode that Paul talked about.
Exactly
in terms of the desktop mode as well.
Yep.
Okay, great. Thank you very much.
You bet.
The other thing we're doing with the SDV platform or the software development platform, this is also sensor-based. We're starting is to put in all of the sensors that you'd see in a traditional tablet, not just into the convertible form factors that I'll show you in a second, but also in the clamshell. You may ask, "Well, what good is a sensor in a notebook?" You can think of a security kind of application where if you're sitting in Starbucks and you go up to get your coffee, you could have an alarm set that if the PC actually moved, your alarm would go off in the cafe. If you left your office, you could actually hardware lock the device because it knows that you're now outside the proximity of your corporate office, for example.
We're continuing to work on usage models, but the sensors are also going into this. Pretty obvious for the compatibility convertible mode that we talked about, but certainly for the clamshell mode as well, we think there's new usages coming. Ultrabook convertibles. Paul showed you the Compal device, which was kind of what we would call a detachable, where you're detaching the screen from the physical keyboard, and the Yoga product from Lenovo, which is kind of a flip-over mode. I'd like to just show you a couple others that are coming down the path here. This is actually from Wistron and obviously has the Windows interface, full keyboard experience, but the unique piece here is you can actually open it up in tablet mode as well and get a tablet experience.
This is an actual concept PC that we have internally, and we're sending this off to the ODMs as well. What you can see here is you have a clamshell, very thin, and these will get thinner over time, with the Windows 8 experience. If you have somebody in front of you in economy class in row 50, you could actually move it forward and get it into consumption mode. Then obviously it'll move down into a full tablet mode as well. You can get the Windows experience, and if you wanted to go do consumption, you can have it in a reader style, just like a tablet. We're going to see these things get thinner and thinner over time with multiple different usage models. The other fun thing we have is the ability to do concept PCs.
This is something we call Nikiski internally, and you can see it's a relatively nice notebook. It'll get thinner as an Ultrabook over time. You can see here it has a touch pad on the bottom that's see-through. It moves the cursor around. If I move my palm on here, it actually has palm sensitivity, so it won't detect the fact that my palm was there when I'm typing on the keyboard. What's most unique about this one is when you close the system, this is where Mooly said everyone's supposed to say, wow. You close the system. We could actually open up mail, this avoids the ability to have to kind of open up your clamshell to check your email. You could go into CNN Top Stories. Let's try to find one. Not 15 dismembered bodies found. Let's do this one.
Well, anyway, when you open it up, you'll get a full experience. You'll get the full experience here as well as on the screen. This is the kind of innovation we think is coming to the Ultrabook. Okay. Now, there's a lot of discussion about just good old-fashioned thin systems. A lot of this work that we're doing on Ultrabook is definitely going to benefit the thin systems as well. We actually believe that as systems get more and more thin, that we get more and more competitive relative to our competition. Ultrabooks are not a race to the bottom. We believe that $699 is sort of the right price point.
The bill of materials of the current systems will reduce over time, but we'll be adding incremental capabilities, whether it's touch or sensors or voice or gestures as we look out into the future. That drives this kind of $699 and up price point, and we will have thinner systems that are going to be thicker. They're not going to have the same responsiveness, not the same kind of security levels. Intel's working aggressively there, and we're excited about the design wins we have in Celeron and Pentium in that space as well. What's coming from a marketing perspective? As I mentioned, we've had a huge number of successful campaigns in the past. Most of us realize or remember going into a train station or an airport and seeing the huge Centrino banners everywhere.
The campaign that we're rolling out that started a few weeks ago is going to cover social media, print, TV, et cetera, and it's the largest campaign we've had since 2003 with Centrino, and the Ultrabook: A New Era in Computing from Intel. As we get the Ultrabook logos going in the market, we're going to be having a unique retail experience as well. We call this store within a store, where when you walk into the largest retailers around the world, you'll have a separate section where if you've met the criteria for Ultrabooks, you'll be able to go in and experience this. The number of SKUs we see coming in for holiday this year are going to be significantly more than what you've seen in the last few months. This is for both large format retail and for small format retail as well.
Last but not least, Haswell. As Paul mentioned, when we created the Haswell microprocessor, which is on track for 2013, we had the Ultrabook in mind. We're significantly reducing the power and the thermal design point here, kind of completely retargeting the silicon and delivering significantly more efficient and thinner designs as we look out into the future. This is going to deliver 20 times lower Connected Standby power. When I say Connected Standby, we mean always refreshing your data with the lid closed, delivering more than 10 days of Connected Standby, always on, kind of always connected to your system. For the first time, I think, really removing the brick or the need to carry a brick around with all-day battery life. We're super excited about this. On top of that, in CES, we talked about a relationship with Nuance around voice.
We think as voice comes to the platform and gesture, it's going to take full advantage, not just of the thinness, the responsiveness, and the security of the platform, but with new usage models, with voice, gesture, and other things coming to the PC. In summary, I think it's a fantastic time for me to be in the PC space. It was an honor, after 11 years of the Tri-Gate development, to stand up and represent all the employees on the third generation Core a few weeks ago. We've got the vPro versions and the Ultrabook versions of third generation Core right around the corner. The momentum is already five times the number of designs or more that we had on the second generation Core. We've got the largest campaign ramping up. As Paul said, we have confidence in our objectives for the year around Ultrabook.
We've got a whole set of mechanical innovations coming around convertibles that we think gives you the best of the tablet with the best of the notebook, Haswell is just going to make it that much better with a chip that was defined from the grounds up for Ultrabook, making things even thinner, even more responsive, and even more secure. Thank you very much.
It truly is amazing that we create a building this large to produce something on the atomic scale. We're in the middle of the largest construction phase ever at Intel. We've taken Copy EXACTLY! to the next level, wherein Fab 42 is an exact duplicate of D1X, which is currently under construction in Oregon. This construction innovation has significantly shortened the project time schedule. Building these massive structures is big business. To build and equip a fab like this costs more than $5 billion. Why do we build our fabs so large when what we create is so small? It's to accommodate the hundreds of tools that are used to produce the massive volume that comes out of all of these factories. The largest land-based crane in the world was brought in to build Fab 42.
It was needed to lift the huge trusses that support the fab roof. It's capable of lifting 4,000 tons. That's the equivalent of a single payload of 26 blue whales. The boom has a reach of 750 feet, and the back counterbalance has 3,600 metric tons of sand. It took 256 truckloads to bring it here, and nine weeks to build. Fab 42 in Arizona and D1X in Oregon are vital additions to Intel's fab network. These fabs will deliver industry-leading microprocessors with astounding performance gains and a dramatic drop in power consumption. It all starts right here.
Please welcome Vice President, General Manager, Data Center and Connected Systems Group, Diane Bryant.
Good morning. As was mentioned earlier by Mark, I've been leading the Data Center and Connected Systems Group since February. Prior to that, I was Intel's CIO for four years, a great job. Before that, Kirk and I actually were co-general managers of what was then called the Server Platform Group before it became the Data Center Group. All in told, I've been in Intel's Enterprise Products Group for 10 years. Prior to that, I was in the Mobile Group for 10 years, and like many people at Intel, I joined as a college graduate as an engineer a very long time ago. Not that long ago. Someone was being generous. Okay. It is a super exciting time to be in the Data Center Group and for some very clear reasons. It is a period of tremendous growth.
That growth is driven by a couple real big macro events, such as the billions of devices that are connected and will continue to build out to 10 billion or 15 billion by 2015. Pick your favorite number. All those devices generating lots and lots of data, and all those devices connecting back into the data center. For each of those devices, there has to be something in the data center feeding and fueling those connections. Our core assets at Intel are unparalleled. You heard Paul talk about it, and those assets directly apply to the value proposition of the data center. We have a very clear match between what Intel does best and what is valued in the data center market. Things such as energy efficient performance, such as security, such as our software ecosystem of 14 million developers and being able to deliver that software compatibility.
We have tremendous assets directly valued in the data center business. It is a highly segmented market, as we'll talk about, and we are investing to win across all of those segments, across all of the workloads. We're investing to meet customer requirements in each segment, everything from the highest performing supercomputers to the highest density microservers, the entire space. You'll hear me talk a lot about the fact that I know many of us associate Xeon with servers, but really, the Xeon is the solution for the entire data center. It addresses the compute demands across servers, storage, and network. Consistent with what we told you last year, our business is on track to double. We will double both in units and in revenue.
As you can see here in the colorful chart, there are several segments within our business, and all of those segments are growing. Some, though, are growing even faster than others, and a couple of them I want to call out in particular is the cloud growing at over 25% CAGR across the horizon of 2011 to 2016, and high performance computing growing at also over 20%. These are very big segments. The public cloud build-out, public cloud capacity doubled over the past two years, and if you just look at China, the China public cloud build-out doubled in just the last year. Tremendous build-out in the public cloud service provider space. High performance computing, there's just an insatiable demand for compute capacity, and hence you see that greater than 20% growth.
In some segments, they are growing well, but our growth in those segments is even greater than market. One of those segments is the networking space, networking both in data center, enterprise, IT networking, so Ethernet, but as well in the telco space, in comms infrastructure. Paul talked about that a little bit earlier. The network market represents a market that's undergoing a significant transformation, a transformation off of proprietary onto open standard building blocks running on Intel Xeon processors. We'll talk about these segments in a little more detail. First, I want to talk about some of the big trends that are fueling this growth, and the first one is the purchase criteria. What drives the purchase of server storage and network? Historically, the focus from IT and in building out your IT capacity has been all about efficiency.
Driving bottom line growth, bottom line efficiencies in your business by automating your systems, business operations, moving to automation, and improving the efficiency of running the business. That is obviously still true today. It is still true that IT delivers tremendous business efficiency, but it's also about IT delivering new capabilities. It's about using IT to create new business models and to drive top-line growth. Also, another big trend in the market is if you look at the markets themselves. From 2006 to 2011, the total market has grown by 60%. If you look within that and you look at the emerging markets, you can really see a trend. China has doubled their share of the market in that same period.
There's clearly an interest in the emerging markets to make an investment in IT with the clear sense that an investment in technology is a means to fuel economic development, and you see that here as the share shifts by geo over time. The other big trend is, of course, data explosion. Data is exploding, both data in transit across the network and data being stored. As you can see, the growth in data is rising much rapidly than the decline in pricing. This is forcing a fundamental shift in the architecture of network and storage. It's just unaffordable to continue to lay down greater and greater capacity to meet this kind of growth level. We've seen these dynamics in our own product launches.
If you go back just five years ago, when we launched the Xeon 5400, there were 31 designs from our OEMs ready to go at launch, ready to be released, and they were all server designs. Fast-forward five years, we just launched our new Xeon E5 processor family, and at launch, we had over 430 designs ready to go at launch, and almost 25% of them were outside of the server space. They were in storage and in network, both telco and data center network. A tremendous shift in how our products are being leveraged as well in the space of the broad data center build-out. Another proof point for the market growth and the explosion is look at the OEM landscape. It is significantly changing. If you go back to 2009, over 80% of all servers sold were sold by the top three OEMs.
Today, if you take our channel partners, you look at the top 1,000 channel partners, they collectively represent our number three largest OEM. You can also look at the emerging markets. China, significant change in the OEM landscape in China. You have several OEMs that have emerged, both serving the local market, but also expanding into global markets. Folks like Lenovo and Huawei. You can look at the ODMs as well. There are several ODMs that are now addressing end-user needs, so moving up and providing full solutions to the customers and providing solutions in new market spaces such as targeted at cloud, targeted at storage, or high-performance computing. The other bucket of OEMs have grown at 2X the rate over that time period relative to the top three.
A big shift in who we serve in our customer base, certainly a much more diverse audience. To win in this very diverse and segmented data center market requires a broad range of products, and our objective is to cover all workloads, all segments, deliver optimized solutions that provide leadership and value to the customer. Today, that means we provide over 100 SKUs addressing all of those segments that you see, 100 different SKUs across five different product lines, as shown on the right. We're not standing still. As you can see, the market continues to segment. New workloads continue to emerge. New optimized solutions continue to be needed. This year, we'll launch another five products addressing new segments, things such as addressing the network workloads that I spoke of, high-performance computing accelerator workloads, and high-density microservers as well.
The way we do this, the way we provide such expansive products to serve all of these segments and workloads is heavily through reuse. We leverage the investments that are being made across all of Intel into the data center space, and it allows us to respond very quickly to changes in the market. For instance, in 2009, when the microserver space became clear, when the new workloads around Web 2.0, web hosting applications demanded heavy scale-out solutions, we were able to respond very quickly with our products into that segment, leveraging the Atom processors that were being developed for the mobile space. Taking those existing Atom processors, wrapping it with server capabilities, validating it, and getting it into the market very quickly. We also borrow the client core products for use in our storage and networking space. Reuse is a big focus.
We have lots and lots of assets inside of Intel. By reusing and leveraging those assets, we're able to serve the data center market comprehensively from end to end. As I said, our assets within Intel are unmatched and very much aligned to what's valued in the data center space. On the manufacturing side, we have manufacturing capability that allows us to deliver very high levels of integration, which are valued inside the data center. We're able to deliver high levels of integration, such as very high core counts. We're able to do it in high-volume manufacturing with outstanding yields and quality. We have system-level capabilities that allow us to optimize the platforms for the specific application. It's not just about delivering silicon or even the hardware system, but making sure that the complete solution stack is optimized and tuned.
We provide software compatibility, leveraging those 14 million developers on IA. We deliver a wide range of tools from our softwares groups that allows the end user to optimize their code on our platforms. We have process technology leadership that is fantastic. Thanks to our multi-generational lead on process technology, we have the best energy-efficient transistors on the planet. You'll hear more from BK, but that process technology and low-power transistors are incredibly important in the data center, where virtually every segment that we talk about is power constrained, either power availability or power affordability. Our energy-efficient performance that comes from our process technology lead is incredibly valuable. We have the ability now to deliver hardware solutions that we couldn't before, so security solutions at the hardware level. We can deliver hardware solutions that are below the surface of attack through our partnership with McAfee.
We certainly excel as well at driving standards. We have a broad global footprint. We're able to drive standards in new computing spaces. With standards, you get both a faster adoption of technology and you get lower cost and availability. Our assets are unmatched and certainly are leveraged in our continuous win in the data center. To talk about a couple of the different segments, first I'll start with enterprise IT. This has been our traditional segment, still is a big portion of where we sell into. As mentioned historically, the purchase criteria in enterprise IT has been about efficiency, about either delivering business efficiency, so lower cost of operations and running your business, or efficiency in IT itself, running the IT operations more efficiently through deploying next-generation technology. That will certainly continue.
It's interesting, through deployment of next-generation technology, you get higher performance at lower cost, so you get significant reduction in total cost of ownership. What we've shown here on the pie chart is just one company, a Fortune 100 company. We went in and did an audit of their data centers and looked at all of their servers. What you can see here is that across all of their servers, 32% of them were over four years old. That 32% consumed 65% of the total power going into the servers, and yet is only contributing a mere 4% of the total compute capacity. This is just one company, but I can tell you it is not an anomaly. This is very consistent, very typical with what you'd see, and even typical with what you'd see inside of Intel IT.
Because of the beat rate of Moore's Law, every year, 20% of your servers are going to fall into that red category and are going to cost you more to run and maintain than to replace them with a new server. That trend will continue. Power is 25% of the total cost of a data center, CapEx plus OpEx. That's in the U.S. It's even higher outside the U.S. in emerging markets. There will be a continued focus and a continued value proposition from an efficiency perspective. As I noted, there are new drivers for purchase of IT infrastructure. For example, you have here BMW, so a car manufacturer. If you look at their IT organization, they used to be responsible for the support of the 95,000 BMW employees and their PCs and their network and all those things.
Today, that IT organization is also now responsible for managing a million connected cars, and that million connected cars will grow to 10 million connected cars in 2018. With that, over a terabyte of data will be transferred into the BMW IT data centers every single day, over a terabyte a day. It's a great example of how BMW has taken IT and created a new business opportunity where IT really is the business. This is just one example. You see this happening in all different segments, in all different verticals. The next big area is big data, we all talk about big data because it is big. It is on top of mind for all of us. Paul talked about the data explosion. You can see the graph there on the left. What is interesting is that there is traditional storage.
All this data has to be stored somewhere, there certainly is traditional storage, so NAS and SAN storage, that continues to grow at double digits as you can well imagine. With the advent of all this unstructured data driven from Web 2.0 applications, the need now is to process terabytes or even petabytes of data real time. It's a massive amount of data. There is a big trend and a big move of storage into distributed storage. The growth of distributed storage, we reflect here at over 30% CAGR. What's key is as this storage capacity continues to get built out, it's the affordability issue. Building out your storage used to be landing more and more drives. Well, this is a very expensive proposition because the utilization of those drives are quite low.
We used to all talk about server utilization and how we needed virtualization to drive up the utilization and get better total cost of ownership. That same phenomenon now is becoming very apparent in the storage space. What you see in both distributed storage and in NAS and SAN is you see lots of new technologies that are being deployed to help drive up the utilization of storage, things such as deduplication or thin provisioning, for distributed computing, erasure code. What is interesting about all these technologies is they're driving up the utilization of the drives by moving the computation onto the storage system. The computation of the storage system is going up dramatically, that's what's allowing us to transition this market off of lower-performing proprietary solutions onto the high-performing Xeon systems.
Today, estimated from IDC, we have over 80% of the total enterprise storage market as the industry realizes that there needs to be a shift from where the spend is going, from landing more and more drives at very low utilization into making that investment into a high compute environment, a high compute storage environment. The next area I want to talk about is networking, and Paul talked about this a little bit earlier. This is an industry that is transforming. With the explosion of data traffic, as you all well know, the service providers are faced with a very expensive continued build-out of the infrastructure. It is a very costly effort that is causing them to pause and look for alternatives. At the same time, they are looking for new business models that will allow them to deploy services on top of their networks.
The combination of the drive for new revenue streams, as well as recognizing that the current infrastructure build-out cannot be sustained, is driving a complete transformation in the industry. Just like back in the '90s, when servers moved from proprietary onto standard high-volume servers, commercially COTS, or standard high-volume servers, that same transition is now occurring in the networking space, in the telco infrastructure space. In doing so, the comms infrastructure is getting the benefit of some of those enterprise IT capabilities that IT has gotten over the years, things such as virtualization. Now virtualizing the network so you get far greater utilization out of them. The energy efficiency that comes with all of the server enterprise infrastructure. Those value propositions are being transferred into the comms space as well.
This year, we have a strategy represented on the slide here, we call it the four-in-one. Our strategy is to move all four of the comms workloads, the application layer, the control plane, packet processing, as well as signal processing, to move those workloads off of proprietary onto Xeon. This year, we launched our new platform that now encompasses three of the four workloads. We have added packet processing into our platform, and we have plans to move the fourth workload as well, signal processing, onto the platform as well. As Paul mentioned, there have been several public announcements of a move to this open standards infrastructure that allows them to get far greater utilization and lower cost of their infrastructure while delivering new services. Folks like Verizon, Korea Telecom, China Mobile, have all made public statements about their intent to move.
This is an area that is growing, obviously growing rapidly, thanks to all the increased data traffic, but particularly growing for us at over 30% CAGR, thanks to that transformation of the architecture. The next area is cloud. We all talk a lot about cloud computing. We talk about it a lot because it is such a compelling value proposition. When I was CIO, we spent 2010 and 2011 building a private cloud inside of Intel IT and have seen all the benefits of cloud computing, so lower cost of ownership, thanks to the automation and virtualization, and greater responsiveness to the business, thanks to the on-demand elements of the cloud. It is a tremendous value proposition. When you think about the public cloud, you tend to think about the four big guys.
It is true that back in 2009, those four big public cloud providers represented almost 75% of the entire public cloud build-out. If you look forward to 2012 this year, we project there'll be about a 3x growth total over those three years in the total capacity, public cloud infrastructure capacity, and at the same time, a significant diversification of the public cloud providers. We have many more customers that we're serving. That growth in players in the public cloud space, number one is coming from the emerging markets. China obviously is a big one with Baidu and Tencent and Taobao. Very strong build-out in the cloud in China, in the public cloud providers. It's also coming from telco. As I mentioned, the telco providers are looking to build their cloud solutions in order to enable them to deliver services rapidly on top of their networks.
Then there's just so many new services that fundamentally didn't exist just a couple of years ago. Twitter, iCloud, for those that are too cool for Facebook, you now have Instagram. All kinds of new services that you wouldn't have even imagined continue to get built out, which really is the virtuous cycle that we talk about. As you have more and more devices that are connected and you have new capabilities being built in the cloud, you enable innovation in services, which then fuels more devices to be connected, which drives greater build-out in the cloud, and so on the cycle goes. That's part of the overarching statement of why do we see such strong growth in the Data Center. Our technology is a differentiator for us.
As an example of that, the cloud segment is a segment that values compute density, which is all about energy-efficient computing. As I said, we've got leadership there. It's also an environment that values security. We've got security embedded into our hardware. The value proposition of our technology in the cloud is very clear to the extent that about five months before we actually launched our new Xeon E5 processor here in March, three of the top five cloud providers had already deployed the new systems into their environment. Very strong desire to get the new platforms in and start reaping the benefits from a total cost of ownership and from a performance perspective. Microservers is a great example of this hyper-segmentation of the Data Center space, a relatively new segment of servers driven by the emergence of the workloads around Web 2.0.
As I mentioned, the web hosting, content delivering network applications, lots of applications with heavy scale out. Today, we believe this market is just about 1% of the server market, so it remains quite low. We predict that it could get to 10% of the total server market by 2015. What I think is also important is when you think about that 10%, based on all the workload analysis that we've done, we believe that two-thirds of that 10% will continue to value the high compute capacity of Xeon. A third of it will do fine on an Atom-based server. We very rapidly deployed solutions, again, using our broad portfolio inside of Intel. We're now on our third generation of Atom-based processors to address this space. Being launched shortly, it's a dual-core Atom system on a chip at six watts that will be coming shortly.
We're also launching our first Xeon processors sub 20 watts. We have a 17-watt Ivy Bridge targeted at this segment as well. There's been a lot of noise about this segment and about alternative architectures in this segment. I think it's important to note that if you're going to deploy a server, it needs to have all the server features that are fundamentally required. Things such as 64-bit computing is a requirement. Virtualization technology, you're not going to deploy a server that you can't virtualize on. ECC, so memory protection. There's some fundamentals around servers that are core to our technology that you don't see in the alternative architectures, as well as just software compatibility being a huge advantage.
If you're going to invest in porting your application over, you have to have a pretty clear value proposition at the end of it, and we all know as well that data centers value standardization and simplification. It's all about standardizing to the extent that you can, and obviously, the data centers today are running on Intel architecture. We have many customers shipping already on Intel, both the Xeon and the Atom side, and I will say this is a great business for us. It's a great market. It's a great segment. When you think about microservers, it's all about high-density computing, so packing as many processors into a server form factor as possible. For instance, there's one server that's available that has 384 Atoms packed into a 10U.
You get the cost out by getting rid of power supplies and PCB board simplification and redundant voltage regulators, allowing you to put more of our content into that form factor. It's a great business. The last segment I want to talk about is high-performance computing. This is a segment that has tremendous growth thanks to tremendous demand for performance and, in particular, energy-efficient performance. As we all know, Moore's Law allows us to double performance every two years. If you look at the TOP500 supercomputers, the performance of the TOP500 computers doubles every single year. Every year, doubling of the performance inside of that space. As a data point against that, when you look forward, we project that the number of processors in next year's number one supercomputer will equal about 1% of the total Xeons that we shipped last year.
Very significant growth in total compute capacity every single year. We often associate high-performance computing with the big government labs projects, and that certainly is a big part of it. I think it's also important to recognize that high-performance computing has gone mainstream. Mainstream as far as to small and medium businesses, recognizing that you can get significant reduction in investments and faster time to market by replacing physical modeling with simulation. A couple examples of that here are Riddell, the athletic equipment manufacturers. They are using high-performance computing clusters to model concussions, head injuries, and design more secure equipment, more secure helmets. Another example here is a family-owned small business.
It's called ACE Clearwater, and it's a small business that is in the tooling business, and they have moved to a high-performance computing cluster to do their modeling of their metalwork, allowing them to significantly reduce their product development cost and reduce the cycle time. Just a couple examples to demonstrate that when we talk about high-performance computing, we're not just talking about the big labs. We're talking about mainstream computing being used down into small and medium business as well. We have unmatched assets for this space as well. We have, obviously, the energy efficiency. Most supercomputers, their first constraint is power delivery. Being able to deliver the most energy-efficient compute puts us in a great position to continue to win in the TOP500.
We have amazing software capabilities that allow our customers, our end users, to tune their applications to deliver greater parallelism and greater performance in an HPC environment. We also have systems ingredients. As Paul mentioned earlier, fabric is a critical architectural component in building scale-out systems. When you talk about the high end, it is even more critical. You have to have very low latency, very high bandwidth, very fast MPI in order to build these systems with double the performance every year. We made investments in both acquiring Cray's high-end proprietary fabric IP and the team, as well as InfiniBand assets from QLogic. Those teams will come together and will develop our fabric targeted at this space, targeted at high-performance computing.
Leveraging our manufacturing capacity, we have the ability to integrate that fabric over time and deliver even higher performance at even lower power. The other piece of the high-performance computing and supercomputing space is our new product that will go into production this year. We're very excited about it. We've been sampling it now for two years. We have software development vehicles that have been out for the past couple years. This is a Many Integrated Core. The first product is Knights Corner. This product allows us to deliver not just the processor capacity for the high-performance computing area, but also the accelerators. Prior to the Many Integrated Core product, there are some in the industry that have invested in porting their applications onto GPGPUs, graphics processing units. It is a significant investment to port to an alternative architecture.
You have to first port the code, and then you have to do extensive validation to make sure that it is still correct after the port, and then, of course, the tuning. With Many Integrated Core, we deliver to the industry a much simplified high-performing solution for accelerator space. You can see the quote at the bottom from Oak Ridge National Labs. This is in reference to their custom chemistry research code, millions of lines of code that would just be prohibitive to undergo the process of porting and revalidating, and with MIC, they can avoid all that. We see tremendous industry response, and we're quite excited about the near launch of this first product. I just want to say, I hope it's clear that the data center represents an outstanding growth opportunity for our company. It is a highly segmented space.
Many workloads, many optimized solutions against those workloads. It's why we have hundreds of SKUs and five different products to serve that space. All of the segments within the Data Center space are growing, and some of them are growing quite rapidly with the conversion off of proprietary onto Intel Architecture. We are on track to our commitment to you last year to double the business and double both revenue and units from 2011 to 2016. We continue to win in this space by leveraging the vast assets we have at Intel, leveraging the investments we make across Intel, serving everything from supercomputers to microservers. Thank you very much.
Anyway, being in the speaker slot after Diane once is an uphill battle per se. We try to make this up by being two of us, as you can see, and yes, two of us. Warm welcome from us for this presentation here. That is MIC.
I'm Mike Bell. I joined Intel, I have to say, 20 months ago, because everyone we work with is 20 years, 25 years. If I say almost two years, it's not quite the same ring. I joined from Palm, where I ran engineering for a few years, and before that, I was at Apple for about 17, working on a wide variety of things. My partner in managing this endeavor is Dr. Hermann Eul, who joined us as part of the Infineon wireless acquisition.
Thank you. Yes, my name is Hermann Eul, as you can see here, the designer of that foil did already an effort to Americanize my German name. If it was so easy with my pronunciation, I would be happy. Yes, I came over with the Infineon transaction. I still try to figure out whether Stacy counts me under the liabilities or under the assets on his balance sheet. I've been in the telecom industry for about, I think it's meantime, 22 years. Started even in infrastructure, joined semiconductors 16 years ago, and my differentiation has been I was with a customer before. That is always helpful. I've seen large systems and how large systems work. I've worked in wireline telecommunication, I've worked in consumer industry, I've worked even in security and smart cards, and most of my life I spent in the wireless industry.
We try to do our best to bring to you a good understanding on what we are doing and where we are heading to. When we say this, we believe that our momentum is growing. It is really growing.
We've had a fantastic start to the year. We've had a number of good wins and product introductions that we're going to walk you through. We think that going forward, the roadmap ahead for us is even better.
Now our know-how in multicom, and as Paul mentioned already, all the communication technologies, we will bring into this ecosystem and we will be bringing great solutions forthcoming years to this industry by bringing this all together and joining these unique capabilities that we have in this company. We are working in an exciting market.
As I said, we've gotten off to a good start this year. If you take a look at the years ahead, we have the opportunity to participate in a market where the units are measured in the billions. It's a high growth market. As you can see, it's accelerating out through 2016, as you can see in our chart here. We think we're positioned to take advantage of the growth in this space, and we think our products are going to be world-class and leadership products.
That's what we think. Actually, we are an incumbent in this market. As we can see from the right part of the slide, there is still a significant volume in 2G for the years ahead. Some people say that 2G may be the backbone of these ecosystems for the next 10, if not even 20 years. The volume of 3G continues to grow. You may not believe it, but we have seen this in 2G as well. The peak comes in this industry is about 10 years after the launch. Of course, we will also be there in LTE when this market comes with numbers. We are an incumbent in that industry, and we believe we have the right ingredients for this. Saying the wireless WAN technology, cellular, Wi-Fi technology, Internet history in that for many, many generations, power managements or imaging.
Yeah. It's not just about doing a chip and software. It's about having all the best pieces that go into making one of these devices. Of course, these devices are becoming handheld computers. It's about the connectivity, it's about the imaging, it's about the software and services, it's about the security, and it's about having all these pieces combined in a systems way so that all the pieces are tightly interconnected and play off each other to make a really great, compelling experience for the end user, which is really what it's all about.
Yeah. We take this systems view and build this on the best transistors. We are the only company in this space that has the opportunity that we can build this on the best transistors in the world to make the most performing solutions out of that. I would like to give you a few ideas on what that means to be in this space. Here, I take the opportunity to highlight our RF technology. We always took the RF technology from a system view, not inside out of the chip, rather from the entire system. What can we do to an RF technology to make this RF technology playing best in terms of your experience? When I say this, our solutions are the lowest in power consumption, and power consumption is your experience.
That contributes to battery lifetime. This contributes maybe also to a smaller battery size, and that means towards your experience on how neat and how compact the device is. The other solutions that we have are traditionally the smallest one. They take the lowest space on the PCB. This gives freedom for either building a smarter form factor for the devices you own or that delivers space for the customer to build in other capabilities in this all-encompassing devices. Performance, needless to say, engineering company always has the best performance. That is what we drive it to. That plays also back in terms of experience for you. If you have a high sensitivity receiver, your device is going to take longer phone calls or data transaction or emails in bad environment.
Everything contributes to our outstanding technology here, and we have generations of this already behind us. Some people say, I believe this by the way, we are the undisputed leader for RF in this field. Having said this, moving on. We are an incumbent in this industry. Here's numbers of customers that we are serving today where we have good relations to. Just give you a few data points, $2.5 billion 2G, 3G RF chips sold over lifetime. This is the number of devices of years. Starting with Intel, by the way, 1st of February 2011, about 15 months back. Since then, we have shipped $500 million chipsets into this industry. We are not new in this industry. $500 million is half a billion chipsets since we are Intel with that business. Cool.
Big number. Good start.
We are shipping LTE RF, and our complete LTE solution will also be ready for the market towards the end of the year. We believe they have the highest security built into our chipsets as well. Take this all, I make it even a little bit wider. Bluetooth, you are familiar with location services, 2G, 3G, LTE, new technologies coming, Wi-Fi, all that goes into this environment. What we want to do is to make that as easy as possible for you so that you do not have to bother. We bother really, because this is not easy. That is multiple transmitter in one device, being on multiple of frequency bands, and the situation hasn't become easier. With LTE, at the beginning of the year, it was 13 bands only for LTE, and it's still growing. It is a headache to get this done.
We contemplate this as a challenge, we contemplate this even as an opportunity for us. We have the best RF engineers, if somebody can make this challenge being comfortable for you, we believe we can do this. We will bring those solutions to the continuum of the devices and make a difference for you. Having said that, we are with the devices.
Back to Moore's Law. Paul mentioned that we were accelerating our efforts to make sure that our Atom chipsets and our mobile chipsets were going to be using leading edge process technology. I've had a couple people recently ask me, why is that important? I want to make sure that's really clear. We're going from 32 nanometer today to 22 nanometer next year and 14 nanometer the year after. It's sort of like magic. I mean, we have a great architecture to start with. We have a balanced system approach inside of our SoCs. We have the right building blocks, the right accelerators. What the process technology does is it gives us performance, better performance at better power with better size. It takes a fantastic engineering effort and makes it even better.
We get phenomenal advantages well beyond what we can just do from a building block level. We think this is a fundamental advantage that we have, because as Paul said, as we design those chips, we can factor the process technology into the way we assemble some of the building blocks. We get even better output when we're done. The other thing that's important to understand in this space is that just building a chip and putting it out there really isn't something that you can do and be successful. Intel has always had a really great software expertise in-house. In fact, over the years, there's been a large number of software engineers inside of Intel that have gone out and helped optimize things throughout the Windows.
I didn't know until I joined Intel that Intel is the number 2 contributor in the Linux ecosystem as well in terms of open source. Intel already has a long history of working with the Linux source base, which is by and large the basis for many of the smartphones on the market today. We've taken all that expertise and used it to our advantage in the Android space now. As many people know, our phone efforts right now are concentrated on Android, and we have hundreds, in fact, we have thousands of engineers right now optimizing Android to be the best version of Android on Intel architecture. This, again, is a fundamental advantage that not many other people have. Not only is just doing the chip not enough, just doing software isn't enough.
You have to actually think ahead of what do you want the end user product to be. When you design the chip, designing a general purpose processor in this space isn't enough. You have to really factor in what do we want this product to be able to enable from a user experience perspective and build that back into the product that you're building way up front in the early days of designing the part. We tend to look at this in four ways. We say there are four pillars of what we're trying to do.
We want to enable products that are engaging, that the person just can't put them down, that there's consistency in the way things work across the devices, that the devices are aware, that the devices aren't just passive things that sit in your pocket, but they can actually do things for you, and they can keep you up to date without you having to constantly check things and push buttons. It has to do all this in a secure fashion because, as I'm sure everyone knows, our entire lives are held on these devices these days. You're starting to see some of the security breaches out there with some of the ecosystems that are less secure than others. We think that we have some unique abilities to be able to secure this data in a way that no one else can.
Along with that, into our SoCs, we try to build capabilities that are unique. We've made a number of acquisitions, including some IP in the imaging space that allow us to deliver new capabilities to the user. In the phones that we've been showing and that our customers have been delivering, one good example, we have some unique IP that lets people take very high resolution pictures at a very high frame rate. Instead of having to say, do I want great pictures or lots of them? You don't have to make that decision. You can take as many pictures as you want almost as fast as you can push the button. On top of that, of course, we have 1080p video capture and playback standard with HDMI because what good is it if you can take high definition video but you can't show it?
We do all this with great battery life. We have in the demonstration platform we've been showing and in the customer platforms that they're shipping, we have 14 days of standby in a really slim package that's far less than 10 millimeters thick. Talking about performance, we knew we had a great part, and we had been benchmarking it and sharing it with people under non-disclosure. It's interesting. Not only do we benchmark well, either we win or come close to winning against other products that are 2 core, 4 core, much newer sort of products that have just come on the market. In many cases, not only do we win, but we do it in a more power efficient manner.
Our numbers are better, and we do it using less power, which is again, sort of the holy grail of what we're trying to do in this space. It's really fantastic that we've been showing these results, and now finally that we're shipping devices. Some of the third parties out there, such as AnandTech, have validated our results, and in some cases, gotten better numbers than we were saying we could achieve. It's been, again, over the past couple of months, this has come out as our products have rolled out, and we're very happy with the results. As Paul said, over the past couple of months, we have rolled out a number of partner introductions and actual devices.
Certainly, it started with the Google introduction in the fall, and it's led through a series of product introductions that have culminated in the past few weeks with us actually shipping devices in some of the geographies. I was over in India for the Lava launch, and it was phenomenal. Over the next Well, within the next quarter, we'll be introducing a product with our partner Orange in Europe as well. This is the device that you see here in my hand, and our momentum is building. The devices are all based on a reference platform that we put together. We decided about a year ago that rather than trying to show people PowerPoint to prove that we had a great product, we would go build a great product and let people see it for themselves, and the momentum has really built since then.
Many of these products people are introducing are variations of this reference platform we developed. One of the advantages of doing a reference platform, we actually built a platform that was so good that it went through carrier testing and was shippable, and it gave people a leg up to ship products very quickly because we did all of the difficult work and the tuning and some of the low-level work, and they really could concentrate on adding what they do best, either localization, UI, form factor changes, and it's worked to both of our advantages really well. Let me talk about the roadmap a little bit. As was mentioned earlier, the first product we have here is the Z2460, and I promise you it works much better than its name.
Ironically, the benchmarks that I just showed were all done with the 1.6 gigahertz variety of the part. The two gigahertz SKU of the part that we have should perform, well, it does perform even better than what we showed up there. Great part. We think it's a fantastic part for us right now to introduce a number of products over the coming year, but we're certainly not resting on our laurels. We have a whole roadmap of parts ahead of us. We're taking two branches, and I want to walk you through those. On the top of the slide you see behind me, you'll see a mid to high performance tier product family we're working on.
Later this year, we're coming out with the equally greatly named Z2580, which is also a 32 nanometer part, but we anticipate it should have about double the performance of the current part that we're showing and we're shipping at the moment. We think it's a great addition to our family. At the same time, we're hard at work on the 22 nanometer version that, again, we mentioned we'll be shipping next year. This is a really big deal for us because it's not just a technology shrink to 22 nanometers. It's a fundamental change. It's a brand new processor core that gives us new capabilities. It's state-of-the-art imaging. It's state-of-the-art graphics.
It's a fundamentally new, newly retooled part from the ground up that will actually give us and our partner companies the ability to do some unique things in software that we're not talking about right now, but we think it'll allow people to create a more immersive experience with the devices than is possible with the processors or the SoCs that are available today. Again, we're working on 14 nanometer as well. We're not talking a lot about that now, but believe me, we have a full effort going on there. We realize that a large part of the growth in this market isn't just in the mid and high tier, it's in the lower value segment as well.
We announced back at Mobile World Congress, I believe, a lower end part called the Z2000, which I say lower end, but still very high performance compared to what's on the market today. It's shipping later this year. We expect to clock it around a gigahertz, maybe a little bit more. It has HSPA+ as part of the platform, so the same great speed that are in the devices that are shipping now from our partners will be available there. Going forward, you see what we have here as labeled the 6331. It's a 22 nanometer part, but it's also our first integrated part. We realized that we need to bring down the component cost, the number of components on the board, the cost that's required to integrate this stuff into a small product.
This is our first foray into that venture, and we're well on the way with that part as well.
Cool. Did you get the hang of it?
There'll be a test later.
We will do the same for the tablets and make it a little bit shorter because the secret sauce to it is exactly the same, and that is what we will apply to tablets. What we currently see is there's 10 OEMs working with our solutions, we see more than 20 designs that they are currently preparing for the market. We are in this, we are positioned for being in leadership position here. Why do we think that is the case? We are convinced we have a clear performance leadership with the solutions that we provide. We have software optimization, and we have tons of years of experience on Windows, and we believe we can make the best Windows experience being available on those tablets. Of course, needless to say, we have the software compatibility.
What you are used to run continues to run. This is an extent and great advantage that we have. Of course, all our engineers work on innovation on IA, and this merit comes to the tablet world as well. Same situation as Mike described it in much more detail for the phones. The same pattern also for tablets, our solutions, we are going through the nodes of Moore's Law in twice the speed. 2012 is the year of the 32 nanometer-based solution. 2013, we will be on 22 nanometer, and for 2000, and we are heading towards a 40 nanometer generation. Same pattern, same audacious goals, same speed. We are moving on. When I say moving on, also organization-wise, we have moved on. We created the Mobile and Communications Group.
Mike and myself, we run this, that was constituted out of four different groups inside the company, NTG for tablets, UMG for the phones, MWG with the Wi-Fi experience, and IMC, that was the former Infineon wireless part. We put this all together. We feel now we have all the resources that we need in one place and in one team so that we can team this up and move towards the future and deliver great results to this market. In that sense, we got the best of both worlds, great silicon and great expertise in this market. The staff that we run, somebody did an exercise on counting the years, so you see, for the top 14 people on our staff, each of them has on average more than 10 years of experience in this industry. We know what we are doing.
Saying that, 2012 is the year where we continue to win designs. We have the momentum. We shared the momentum with you so far. We continue to build on this momentum and prepare for the future. 2014, about two years from now, we expect that we have, in every segment that we serve, leading products. From this on, it wouldn't be Intel if you couldn't expect audacious steps then going forward. We will let you know in due time when we can share with that more of this. This lets me summing this up.
Yeah, like Hermann said, we have the right technology, we have the right people. We are highly driven and motivated to make a serious dent in this space. We think our momentum is growing. We think we've gotten off to a good start this year. We are on plan and on track for where we want to be. The technology advantage we have with the process technology that we have inside of Intel, combined with just the fundamentally great architecture and the scary smart architects we have, is something I don't think anyone else can match. It's combining all of the pieces that Hermann showed you, the multi-coms, the software, the hardware, the radio technology, the application processor technology, all into one cohesive platform is why we're going to win in this space.
Bingo.
Thank you. I think I'm the magic chairs.
Thanks, guys. I think I'm the magic chairs. Thanks, guys. I think we're going to do a little Q&A session right now. I'll invite Paul, Diane, and Kirk actually to join us up on stage. While we're getting settled in-
You like it?
You'll see people with microphones. The IR team is spread across the room with microphones. They'll raise their hands, and if you would raise your hands if you have a question, they'll walk down the aisles, and we'll attempt to sweep back and forth and get as many of these questions answered as we can. All right. Good. With that, if we want to start, actually, let's start right down here. Vivek.
Thanks very much. I have two questions. Vivek Arya from Bank of America Merrill Lynch. Two questions. Paul, how would you assess Samsung as a competitor? I think a lot of the discussion has been Intel versus TSMC over the last many quarters, but how would you assess Samsung as a competitor and their ability as an IDM and also a hardware OEM to take Arm into more areas of the computing continuum? I have a follow-up question.
That wasn't enough. I view Samsung as a classic co-opetition partner in the industry. They're a very big customer of ours in PCs and growing very fast and a very good innovator. I think some of the best Ultrabooks that you're going to see this year are out of Samsung. From that perspective, a solid customer. They're a partner for ours in Tizen development. They're certainly in our design sights for our devices, I can't talk about design wins at this point, but you can think about them as the largest volume cellphone manufacturer in the world, it would be a very attractive customer for us. From that perspective, all those are on the cooperation side. Also on the cooperation side is the bulk of their semiconductor business, which is memories. They're the largest supplier of memories in the world.
Key, I think, to the advancement of memories and, to some extent, technology is they're also a partner in 450 development and a partner in the industry in EUV. That part of the semiconductor business of Samsung, which is the bulk of it, is a very strong cooperation kind of model. The area where there's some potential for conflict or competition is in their foundry business, and perhaps in their view as an IDM. Up to now, they've not really made the choice to optimize as IDM, the bulk of their foundry business is for someone else, which is non-optimized. How they evolve, I think, remains to be seen, and I don't see that really being an issue. I do see them being certainly one of the ones that can cross this chasm into high-value transistors.
They'll have a different set of challenges than we do, assuming their business model doesn't change and that memories remain an important part of their business, because the memory business will hit the EUV wall first. They're probably going to have to deal with that before we do. From that perspective, that's good that they're going to help debug that. On the logic side, our view is they may end up pushing Arm broadly or intrinsically in terms of their market. These guys' jobs is just to make sure that we build better chips for their systems business so that we win those units.
Vivek, one quick follow-up?
As a follow-up, at what market share is the mobile effort worth it? How should we track your market share goals, and progress?
Yeah
it makes?
I love that question. It reminds me of the question I got in the earnings call. Somebody said, "With Samsung and Apple having over half the market, is the other half of the market big enough for you?
I asked that question.
Oh, you're the one? Okay. Well, you're asking it again.
Ask again.
What a difference a year makes. A year ago, there wasn't anyone in this room that didn't work for Intel that thought we'd have a chance in this business, now you're saying, what's our market share goals? Is half the market big enough? Half the market's big enough and our goals are large.
Very good. Ruben? Let's hit the right side of the room over here. You have it? No? All right. Right down here. Yeah. Ross?
Hi, Paul. Ross Seymore from Deutsche Bank. In your presentation, you talked about that separation where Moore's Law and the advantage you have is going to be a rarer and rarer commodity. You've been talking about that for quite some time. Really, what creates that delta and that separation now going forward, and what sort of benchmarks can we look at financially, either from the benefits that it gives to Intel or the detriments it gives to the fabless business model to see the progression playing out as you envision?
Well, we've been talking about it getting harder. I don't think we've talked much about the distance getting wider until very recently. It was something we believe, but there weren't external data points to prove it out. Now there are enough sufficient data points externally to be able to see this playing out in real time. The difference depends on the market you're looking at. I think there's inherent advantage in the integrated device model, which I talked about. Go to what MIC was talking about and what Diane and Kirk talked about. In each of the nodes that we're talking about, in terms of our segments that we compete in, better transistors matter, and they matter more and more in a world where the world falls farther behind. We don't see ourselves moving to cost-based pricing.
We value base pricing. For the foreseeable future, we can use every wafer we can build and get paid twice, get paid for the foundry margin and get paid for the architectural margin, and that's a really good business. In some of these areas, though, where we're not the incumbents, you can't change the pricing model. We're not going to change the pricing model of apps, processors, and smartphones at any time soon. We have to be able to compete on performance at approximately the same price. If we have a lower cost, we have higher margins. We have more flexibility. That's where those advantages start playing out nicely, is in these new markets, more so than even in the existing markets that we have.
Okay. We have one over here. Then we'll come back over this side.
Yeah. Hi, Paul. Graham Tanaka from Tanaka Capital.
There's people on the audience, I think. Dice them.
The other guys can comment too if they would like. On the same subject, this is a pretty important thing. All of us know that Qualcomm has had problems getting enough 28-nanometer chips, and that's a direct data point that you're addressing here on the fabless model. At each cycle of tick-tock or whatever you want to reference, how much advantage do you think an Intel IDM would gain with each cycle? In other words, is this something that adds a certain % each cycle or of lower cost and higher performance?
It depends on how we spend the transistors. The answer is all the above. In some segments, MIC talked about the value phone segment. There, I think, is a very good example of how we've chosen to spend the transistors. We're going to deliver substantially more performance than anyone else in the value segment at a lower cost. That's because we can take the integration to a level that gives us a very good advantage. In high-performance computing, the MIC product line that Diane talked about simply couldn't be built on any other technology. You couldn't get that many cores, you couldn't get that kind of performance. We're choosing to use the dynamic range of this capability differently for each segment, I don't want to get pinned down to, we're only focused on this cost or this performance.
The aggregate of all these product lines in any segment will, I believe, increase our competitiveness on any axis relative to competition in each of these segments.
Right. All right, for the next question, we're going to have more time actually with Paul in the afternoon. If you have some questions for Kirk, Diane, Hermann with two Ns, or Mike. Great. Do we have one over here? Ruben, over on this side. Great.
Yeah. Thanks, guys. It's John Pitzer with Credit Suisse. This was going to be a question for Paul, but I guess I'll pulse it out to everybody else on the stage. When you look at all the bottoms-up growth opportunities, you guys have put a CAGR out there for the DCG group. I'm kind of curious, overall, what kind of growth rate do you think these opportunities support? Maybe Paul, from your perspective, when you look at the cost of doing business now, cost is moving higher, what kind of top-line growth rate do you think you need relative to the cost structure to generate still healthy returns or the returns you've seen over the last several years?
You want to start with Diane there on that one?
Yeah. I think you mentioned it, right? We are projecting a 15% CAGR over-- we said it last year as well, from 2011 to 2016, we're definitely on that beat rate. It's a great business. We have very healthy margins. Our operating margins are around 50%, and we look out into the future, and we believe we're still there. I think he was maybe addressing one of the other market segments.
Let me deal with the cost comment you made, John. I don't buy your assertion that our costs are going up. In fact, I showed you a chart that shows at least through 14 nanometers, cost per transistor is coming down, and that is different from I think much of the rest of the industry's view of the world today. Our capital costs are going up, and Stacy will talk about that in detail this afternoon. Remember, our capital costs are going up while the top-line revenue's gone up by $20 billion. The problem is we lived in the $30-something billion range as a company for a decade. We all got used to $5 billion to $7 billion of CapEx. When you see $10 billion or $12 billion, you say, "Oh, my God," you forget that our revenue's approaching $60 billion.
The size of our business has doubled. I think that it's even hard for me to get used to double-digit billion of CapEx per year or double-digit billion of R&D. We're a giant company now. We've grown substantially. Part of this is just the scale of the company. Stacy will show you data this afternoon, for those of you who hang around, that show that the capital model is not broken. In fact, it's still extremely healthy. Our unit costs go down. Our spending as a % of revenue stays within the boundary conditions that we've told you about.
All right. Great. Let's come back over here.
Thanks. It's Glen Yeung from Citigroup. This is not a question for Paul. It's for Hermann.
Bonus points for that, Glen.
Thank you. Hermann and/or Mike. Paul actually did talk about transistor technology and the increasing importance of that. I wonder to what extent that lead that you're building is evident to your customers in the handset market, and to what extent it's driving conversations that you're having with them and/or forcing them to have that conversation with you. That's question one. The second question is actually for Kirk. In talking about the Ultrabook cycle that we're coming into, how confident are you that we're going to see a typical notebook refresh, particularly now that we have the advent of tablets in the mix?
Let me take that one.
Yeah. I'd say our customers are very aware of the advantage we have on process technology, even more so with some of the news over the past, say, six months of people stumbling in this area. It's something where it's not only the process technology, but our long-term roadmap of the building blocks we're putting in and what we can do with that process technology has people very interested at the moment.
Maybe I can add another color to this. You have been maybe watching us trailing always in the technologies on the communication part. I would like to give you an idea on how, for example, RF transceivers are developed those days. You may know that we were the first having a CMOS RF. Today, nobody else can think of anything anymore. That was a revolution. We brought the first RF into a single integration, together with baseband and memory and power management. Today, that is standard. All of those integrations, all of those innovations had been triggered by semiconductor technology. This is not only bringing this technology into a new process node. The capabilities of Moore's Law forced the engineering community to do a new innovative architecture around it to be harvesting on Moore's Law capabilities.
That usually set us apart from the competition, in many cases, even more than 2 years.
All right.
I think relative to the refresh cycle, we have a lot of confidence. Going and saying we're on track to this goal for Ultrabook we took for the end of the year, I think it was an audacious goal. Getting 5 times the number of designs in the process is a big, big deal. Obviously, we're not against tablet. We have lots of tablets coming. If you look at the product that Paul showed, you detach the mechanical keyboard, you reorient it on Windows 8, you've got a pretty nice tablet, and you're not compromising anything that we continue to hear of the mechanical keyboard. I think there's going to be a tremendous amount of innovation that's going to drive a ton of excitement, and we have high confidence.
Let me just add to that, or tag onto it. I think in the absence of touch coming into mainstream computing through Windows 8, the potential for cannibalization would have gone up substantially. Now there's the potential for reverse cannibalization.
That didn't exist, it doesn't exist until touch comes on these machines. Touch is a fundamentally different kind of consumption criteria. Our view is, that's why I keep saying the end state's not obvious yet. Our view is that integrating these two capabilities will be very powerful.
I think you'll see the retailers pretty aggressive because it's a very easy differentiator of the old and the new from a PC perspective on touch.
Right. I think Shawn's been waiting patiently here.
Thanks a lot. Shawn Webster from Macquarie, hi. I have a question for just, I guess the first one on the continuing this thread on Ultrabooks. Kirk, you keep talking about the price point coming down as a catalyst for the volume shipments hitting your goals. Is there something that you need to do in terms of the pricing of your own processors in order to enable that? Diane, I had a question for you. In terms of the Romley, can you give us a quick update on Romley and what the ramp profile of that looks like for the balance of the year? On the back of what Cisco was talking about on enterprise spending last night, if you have any comments on the outlook for enterprise spending this year, that'd be helpful. Thank you.
Yeah. I think the simple answer to your question is no, there's nothing we need to do on our pricing to enable that. As we look forward from second-generation to third-generation Core, I think it's the momentum behind when more and more companies, as I said, mentioned 30 different touch designs coming. If I were to talk to you six months ago, it wouldn't have been that number. That's driving incremental capacity in the industry for 13-inch and greater touchscreens. That drives down that volume. If you look at prismatic batteries, each one of these things, that's why I went into the detail. This snowball is rolling now, and we're able to focus our $300 million Ultrabook fund more on the innovation that's coming next-generation around things like voice and gesture and sensors and all these kind of things to keep things going.
For this year, I think $699 is aggressive. Tactically, there's already local OEMs in China at $533. We're not encouraging that. It uses local components, people are able to get there. We think the experience we're delivering, people are going to be willing to pay for, and it only gets better in 2013.
Yeah. On the Romley ramp we launched in March, one of the data points that show the momentum behind the Romley platform is at our launch, we had twice the number of OEM systems ready to ship at launch than we did in the prior talk, the prior big platform launch, which was Nehalem. We had, as I mentioned earlier, over 400 designs ready to go, 100 of those being in the telecommunications space. Really representing that with Romley, it's not just a server platform, it's an entire server storage network, both enterprise IT and in the telecommunications space. We're showing very good adoption, very good momentum. Also represented by the number of cloud providers that were actually shipping Romley before launch.
We actually had five positions on the TOP500 list five months before launch, it's a great high-performance computing solution as well. There's a lot of marketing momentum that we've already demonstrated. The other question was around-
Cisco.
Cisco.
Why don't I'll take that. We don't see any change. Our view is the year is playing out just as we thought. The quarter played out last quarter as we thought. This quarter is playing out as we thought. The enterprise is good. It's not fantastic, but we don't see a change in that. I think John's comments were focused on Europe in particular. We haven't seen any change in Europe demand on the enterprise side. The early Ultrabook sales are not going to be driven by enterprise because of the enterprise refresh rates with the adoption of Windows 8 is going to be probably next year. I don't really see what he's seeing, and I'm not sure that what he's seeing is geographical versus competitive in his market.
Okay. Let's go back up to this corner here. Chris.
Thanks, guys. A quick question on Ultrabooks. It sounds like you think that when Ultrabooks start to come out, that the cost curve on notebooks will shift upward, i.e., the average selling price of a notebook will go up or people will be inclined to buy a higher price notebook. Is that true?
Well, I think, people will pay for experiences. We think, yes, at $699 and above to hit those curves, there could be that trend, but ultimately, we're going to fulfill the experience that people want to get. Touch is going to add a premium. I think a lot of the OEMs will have touch and non-touch keys, for example. $699 and above, I think, we'll hit our goals.
It's in the sweet spot.
Just to follow up on touch, does Intel have plans or is it possible for you guys to integrate touch into your chips or an SoC, or will you continue to outsource that and do a reference design?
No comments on what we're doing or integrating relative to touch. Certainly for this year, it's going to be discrete components with our partners.
One over on the far right.
Yeah. Thanks. It's Brett Simpson at Arete. I had a question about your strategy towards Apple. We can all see that they're building quite a big ecosystem around iOS. When it comes to Intel's position here, how important is it for Intel to build a strategic relationship with Apple around iOS, and how might you do this when Apple seem so wedded to their own silicon roadmap in that space?
Well, the decision to put iOS on a different architecture is uniquely Apple's decision. iOS is developed on x86, on Intel architecture. They rewrite it on Macs, then they port it over, right? The ability to take advantage of our architecture certainly exists, and they've been able to move architectures in the past. I think that any architectural change there has to do, going forward, will have to do with how this market evolves. To some extent, Windows 8 has a very strong position. With one OS, they've spanned tablets and PCs and maintained compatibility for the vast majority of the systems, which is a huge benefit to end users and applications and so forth. Yet they'll start their own ecosystem on applets for Metro. iOS today is limited to phones and tablets, and they've got the macOS up here. Those applications don't span.
Maybe they can create an environment to run a lot of the applets on the macOS, but having the same device span one operating system means that someone has to port. They either have to spend a lot of money moving all the Mac applications, which we helped them with the last time, and it's terribly expensive and a lot of work, or they have to spend a lot of money moving all the iOS stuff up. I think they have a strategic question that they have to answer, and Tim's the best person to answer that question, not me. Our job is to ensure that our silicon is so compelling on either side, in terms of running the Mac better or being a better iPad device, such that as they make those decisions, they can't ignore us.
All right. With that, I think we'll have to wrap up this first Q&A session, but there will be a couple more, so you'll have ample opportunity. We're going to break now for lunch, and even during the lunch period, you'll have some time for Q&A. As we walk out of the auditorium, you may remember where you checked in this morning in the cafe. That's where we'll be serving lunch. It should be set up on the table so we can avoid long lines. People will be able to sit down and immediately get fed. We've got about an hour set aside. The executive team will be circulating around the room. We encourage you to meet up with as many folks as you can. Thanks, and we'll see you in about an hour.
Ladies and gentlemen, please welcome Senior Vice President, General Manager, Software and Services Group, Chairman McAfee, Renée James.
Thanks. Okay, welcome back after lunch. We have, as you know, changed the format for the post-lunch section, so hopefully, you'll find it more lively and entertaining. I'm going to be joined on stage with Stuart McClure, who's the CTO of McAfee, in a minute. Before we get started, I'm going to give you literally one or two slides of context just to kind of say where we're at, remind you of a couple of key things around our strategy, and then we're going to move to the fireside chat. We will take questions at the end during the question section, and Stuart and I will still be in here for that. Starting with Intel and security, all of my predecessors this morning did a fabulous job of talking about security as it related to their portion of the business.
What we say is security is the third pillar of computing, and the reason why is because security is a key opportunity for differentiation for Intel and a key attribute on which consumers, IT managers, mobile users, all care about, value highly, and all the data says they'll pay more for. It becomes a pervasive technology underpinning across all of the different segments from one end of computing to the other, including the cloud. We have been working on security for a long time, and we've talked about this. In various different presentations over the years at the analyst conferences, we've told you about different features that we have in our silicon, different features that we have in our platform, and we started talking last year about security with the acquisition of McAfee.
It's from the silicon and the platform that creates the foundation that we build the enhanced co-developed solutions, the end-to-end solutions with the McAfee technology. 10 years we've been putting security technologies in our platforms. They are shipping in high volume. Many of these technologies have been shipping for over five to seven years. They're in our Core product line family and our Xeon family. They're in hundreds of millions of units, and that's all opportunity for us and McAfee to sell new services to the installed base, and then, of course, as those products continue to add new capabilities to serve into the new segments, including mobile. Our strategy is to create end-to-end solutions that solve new areas of security in a fundamentally new way.
Stuart will talk a little bit about what's evolved in the industry and why we think that our point of view about the future of security and how Intel's uniquely positioned to solve it through the combination of the silicon platform and services together is a great opportunity for us. There's three major areas of technology in security that most people are trying to solve for. Most products today, things that you think a lot about, like antivirus, are around execution control. That means stopping things that shouldn't happen from happening on your platform. McAfee, I think all of you know this, but I'm going to just repeat it one time for you. McAfee and antivirus and its competitors are very focused on execution control, but McAfee is also a multi-platform company. They have products for mobile. They have products for the data center.
They have products in the cloud. They have products for networking. Across that entire spectrum, they're going after execution control and now two other emerging areas, and we've made some subsequent acquisitions. One is in identity and attestation, which is a huge emerging area of security, and that's about knowing who the person is who says they are on the end of the device, on the other side of the website, what have you. On the other side, it's in recovery of the data and the systems after an unfortunate incident occurs. Intel has very unique, and exciting platform features and silicon features in all three areas but specifically in data recovery and in identity and attestation, some of which we've talked about this morning.
Combined with new services from McAfee, we create unique and differentiated security solutions that solve and prevent attacks from even happening that we couldn't before. In the area of execution control is the first product that we launched at the end of last year. This we call Deep Defender. It's a service. You subscribe to it from McAfee. It's for vPro-based Core systems from Intel, either installed ones or new ones, and you heard this morning how we're adding vPro to Ultrabook. This service prevents the execution of whole categories of malware that can't be prevented in software only. Stuart can talk a little bit more about some of the most recent worms that we would be able to stop on a machine that had vPro plus the new McAfee Deep Defender product on it. This is a fundamental breakthrough in security.
It's what we talked about when we said why did we buy this company? The creation of the solution that solves a new area of security in a different way. We believe that that's only going to grow and that all these other areas are going to be very similar, where it's the combination of the ability to use the hardware to actually stop the execution, use the software service to detect something malicious coming in. The hardware now can talk to the software in a unique way through the DeepSAFE technology layer, and we can say, "This doesn't look very good." The hardware now can speak to the service and say, "You know what? We won't execute it." It just doesn't infect your machine. That has never happened before.
We have embarked now into what I would say is the new era of security, which is hardware-assisted security. What you're going to see from us later this next part of the quarter and into the second half of this year and well into 2013 is a robust roadmap of new services that will be launched, and they'll be launched under the McAfee brand through McAfee that correspond to platform capabilities and features in our products, specifically around Ultrabook, around the Xeon family, around Core, and then coming to Atom later. With that, I wanted to give you just that little brief, kind of where are we, and invite Stuart up, and Larry's going to help us, and have a little discussion about where we're at. Thanks. Hi. Thank you. Thank you, Larry.
Water for you. Thank you, Renée. Welcome, Stuart.
Thank you.
Welcome, Stuart.
Thank you.
Renée, you were the executive who championed the McAfee acquisition and the one who made the deal happen. It's now a year later.
Yeah.
What's your assessment at this point?
Well, things are going very well. Measurements of well. We have just completed the first quarter, it was the largest bookings quarter in the history of McAfee. The business is doing well. We have a very robust roadmap of new services coming out. We've begun, in the last year, the innovation products. One of the assets of McAfee that we were very excited about is they have a robust R&D capability and research team. That, combined with our work in security, we've begun the, what I would call the invention products that will come out in the next several years. Things are going really well. We've had the employees. We have a really solid management team in place, the business is moving forward, growing very well in mobile, new area for us.
Growing very strongly in network security.
Stuart, you saw in the video some of the exponential growth we are seeing in security threats that are happening around the world. We're hearing about new threats constantly. I would assume there are a lot of threats that we don't hear about.
Yeah.
What's going to happen if our industry doesn't respond quickly enough to that landscape?
Well, yeah. Just to give you a little bit of background, we've, I don't know, probably about just 10 years ago, just one decade ago, for every one you heard about in the news or the media, you had at least 10 that you never heard about, never saw. Today, we're probably getting to about a 50/50. For every one you hear about in the news, you're probably the same number as never comes to light. The problem is big, and if you just do a simple search for security or hacking or hackers, you'll get a kind of a sense of the breadth of the problem. The real challenge, if we don't take this seriously, and we are with combining the hardware and software, which I'll talk about in a second.
If you don't, you can really impact folks' lives day to day in a substantive way. For me, as a technologist and as a security professional, and a passionate one at that, to me, this marriage of really pushing down security down into the core of the computing capabilities and hardware is truly the only way to solve this problem long term. I'm just thrilled to death to finally see this. We have been talking about this kind of solution for a long time in the industry, and now we're able to realize it.
Can you give me an example of a serious threat that people might not have known about that could have had a huge impact?
Sure. Hopefully, everybody's hanging on to their seats, but I'll give you a couple. Nothing too exciting.
Their computers.
Nothing too exciting. As you probably have seen over time, the physical world and the virtual world are coming together, which means that the physical computing environment, or I'm sorry, the virtual computing environment can affect the physical lives of people. Pretty good example, a number of years ago with a major transportation system, of course, controlled by computers, and we were called in to take a look at suspicious activity. When we got in there, sure enough, we found bad guys that had been in there for many, many months and had complete control over that particular transportation system. Could have done just about anything. Luckily, we caught them beforehand and was able to extract them and stop the activity. Think about just about everything that's being connected today.
If it affects your life or your health, that can also be impacted by the bad guys.
You touched on it, Renée, in your opening comments. You touched on it briefly here too, Stuart. Renée, what is that unique value proposition you see with the Intel McAfee combination, particularly as it relates to competitors?
There's a couple things. The first is that, of course, uniquely, we have been innovating at the silicon level for a long time. The combination of our ongoing and future innovation at silicon and at the platform level, at both levels, combined with the software assets, create the capability of this end-to-end solution. While a lot of our competitors will say they have secure features at one level or another, it might be at the platform level, or it might be at the silicon level, there's nobody that I can think of in the industry that has all the assets. None of our competitors have been working on it as long as we have in the combination of the two companies. I think that is a significant R&D advancement over where anybody would if they were starting today be.
We're innovating it at every level. The other thing about Intel is that because we cover every segment of computing, as Paul discussed this morning, we have really paid attention to how security flows through from the phone to the Data Center. McAfee has been working on cloud, network, endpoint from the same holistic range, which gives you across the entire continuum of computing security. None of our competitors are doing that.
What would you add to that, Stuart?
I would echo and probably put a couple exclamation points. The first one is co-design. I use this often. Everybody ever asks us. To me, that is the greatest value of what we're doing is that Intel has been actually remarkable very early on in building in security features into the technology once they've realized, okay, there are some gaps in how we're protecting these systems. They've built it in, but they've been atomic. They've been separate from a broader solution and tying the software in. For us, being at those early stages and being able to design it-
Yeah
with the long-term intent of a solution in mind, to me, is really just super exciting.
Renée, we have been criticized by some people in this room and others for our ability to integrate acquisitions in the past.
Yeah.
How is this acquisition going as far as integrating McAfee into Intel?
It's going very well. Evidenced, as I said, by them having continued strong growth as an independent business, in addition to the contributions to Intel. We started with Wind River and have continued with McAfee, really trying to learn from our prior experience and taking a different approach, which is the wholly owned subsidiary. McAfee's business, and was alluded to this morning, will continue to serve all platforms, will continue on their journey towards making the world a safer place. At the same time, the innovation and the co-development of secure solutions with Intel, we've really worked hard to focus on the integration around that. Around the two intellectual property R&D assets, how do we build solutions together, letting the company and its business continue to grow. I think that this model, it's difficult.
It requires a lot of work from all the organizations in Intel that support us, IT, and finance, et cetera, but it's working very well. It works well for the customers because customers want to see a security company totally focused on security, and they really are excited, and Stuart can talk about this, about what they see from the first solution that we're bringing out.
How does it feel from the McAfee side of the equation?
It's been actually quite nice because they've kept us pretty independent. At the same time, they've opened up these huge pipelines of opportunity for us to get in and talk, and really, like I said before, that co-design effort, getting in early and discussing all the great ideas. Intel's got great ideas, we've got great ideas, and then melding and merging them together and putting them on a roadmap that makes sense vis-a-vis broader solutions, right? From my perspective, and I know my colleagues, it's probably one of the most successful ways to do a major acquisition like this in a merger.
Renée, you talked about Deep Defender in your opening comments. Stuart, I'll ask you a question about, importantly, what are our customers saying about the product?
Right.
2, have you seen success with Deep Defender so far?
We have. We have seen success so far in many cases, in many different types of threats that have been coming out, but also with our customers, which ultimately, at the end of the day, they're the judge. We released Deep Defender and McAfee DeepSAFE this past Q4 in 2011.
For Enterprise.
For Enterprise, correct. For this first quarter, we've seen a lot of traction. We'll start to see continued traction this year and into really 2013 with a lot of the roadmap items that we have. I'm excited about what we have today because it's been stopping stuff left, right, and center that the software-only solutions just could not do. There's a real simple reason for this, is because when you have these pieces of malware and these bad guys that come on here with advanced persistent threats, they're taking advantage of everything in their arsenal. If they can take advantage of a core, let's call it a kernel, an operating system vulnerability, something that can easily be bypassed, then the only way really to be able to see it is to look from below or above, right?
To be outside of that world to see it properly. Think about it, if this world is already infected, if you put something on it, well, of course, that infection can manipulate what that product sees. If you look at it from above or below, you can see it objectively and independently, and you can prevent, and that's the key.
I know we don't name names of companies, do you have an example of where Deep Defender actually stopped something?
You mean an attack?
An attack, yeah.
Well, sure, yeah. There's many, but just recently with the Cridex worm, of course, that gained popularity really in January of this year, although we saw early samples last year. It's a simple little piece of malware. It's a root kit that once you get installed on your box, connects up to this nice big kind of cloud for hackers that you basically go in there, and you can pull all kinds of personal information, username, passwords, things of that nature, all meant for financial gain. They'll go in there, log in there every minute of every day, and get alerted of new accounts that come online. Then they go in and log into your account, and they try and extract money. This Cridex worm was a great root kit example.
Sure enough, day zero stopped that type of an attack, whereas others took weeks and weeks because they are really just focused on a blacklist model. The blacklist model means we have to see something before we can stop it. We have to know that this is bad. That's really the opposite of what we're doing with the McAfee DeepSAFE and Deep Defender. We're looking for maliciousness, not malicious.
Behavioral.
Yeah, exactly. More behavioral and more archetypal, if you will.
Renee, I know prominent on the minds of these folks in the audience is what is McAfee meaning to us, and what's it going to mean from a revenue and profit standpoint, both in the near term and the long term?
Well, near term, you can see it in our segment reporting. McAfee is contributing and growing well. As I said, they are continuing to acquire companies and innovate and grow that revenue stream independently. As we've been talking about, the platform features that we put into our silicon and into our platform-level products are differentiators for us. They are of high value to our customers. Even in the economic downturn, we saw that security was one of the only features and capabilities that people would pay more for and continue to value. We believe security is going to continue to be a high-value item, continue to be a high differentiation item at the platform level. We're going to get paid for that. Now we're going to get paid for the service that completes the solution.
In that case, we will be launching a series of enhanced services through McAfee. Those are going to be sold on a recurrent revenue stream with a three-year licensing model. We're looking forward to that combination as the forward-looking contribution.
Stuart, what about beyond financial? What other benefits do you see besides top line, bottom line?
In terms of-
In terms of the McAfee contribution to Intel.
Yeah. Top line and bottom line are going to be, I think, strong indicators for us. To me, it's changing the game of what we do in the industry. A lot of people know us as the antivirus vendor, and rightly they should. That is our history. We've extended far beyond the AV world, into network, into cloud, into mobile, into everything. It's a huge emerging part of our business. To the degree that we're going to obviously integrate and provide a lot of features and value add around protecting customers, to me, that's really the main benefit, is that we're going to be able to definitively protect and prevent against the most advanced threats in a very easily consumable and used way.
I think also this direction that we're heading is going to change the way the security industry raises the bar for what the minimal accepted approach is going to be. We have a head start on that. I expect, as innovation continues, that other companies will have to look to us and try to follow because this is a significant shift from cleaning up after you have the Cridex worm to preventing it from ever happening.
Yeah.
It's the fundamental change in the approach to security.
Yeah, I can't say this enough. We, as an industry, have been very reactive, and we've been criticized for that. You have to know the problem, you react. Somebody has to get hit. Really what we're doing is changing the game. We're being preventative. Proactive and preventative. That is changing the way the entire industry is thinking.
I like to think that that is a benefit to the customers of Intel. Intel platforms are safer.
Paul, Mike, and Hermann all talked about our progress and our aspirations in the phone segment. What does McAfee bring to the table? I'll start with you, Stuart. What does McAfee bring to the table when it comes to security in the phone space?
We have a number of solutions that are already out there for iOS and Android. We have near-term solutions that are very exciting that are even non-AV. Non-antivirus, non-protection of applications. For us, being kind of expanding and looking at every part of security is really important to us for mobile because we know that it's not just about bad stuff getting installed on there, but it's your actual usage, making sure that you're using it in a safe way, that you're the right person on that phone for.
The identity part.
Right. Exactly. Those things are key. We're moving far beyond AV.
What about more broadly? You touched on phones. We talk about the compute continuum. You talked about it being a pillar.
Security being one of the three pillars. What are we doing, and how important is that security solution across that entire spectrum, all the way from that handheld device all the way up to the cloud?
Well, as I've been saying, I think that in mobile, Stuart was going down this path, mobile security presents itself differently. AV is still important, but identity and the management of your personal data. There's a whole new area of security, which McAfee's well-positioned in. It's very cloud-based. McAfee, again, is well-positioned for that. Our platforms have performance, we have the capabilities in vPro and other things that will come to mobile that really give us an advantage of solving security for mobile in a new way. Mobile is an unknown. The way security is evolving in mobile is just really green field. I think that's a huge opportunity space for Intel, and McAfee as part of Intel. Across the continuum, again, security is top of mind. It is a core buying criteria and differentiator for any conversation.
We have really, I think, are positioned very well.
One other thing to add on that, we talk about the cloud and everybody connecting into the cloud, and just if you can stop and think about that for a second and kind of absorb it, everybody's pushing everything into the cloud.
Cars, TVs
where do you go? Target, right? The cloud. You go after data centers, the virtual world, the cloud, all the infrastructure that makes it up because, hey, why go and try to hack 1 million devices when I can hack 1 device and get 1 million accounts? It's both of those parts that we are focused on, not just the endpoint but also where everybody goes up into to share.
Is there a particular category within the continuum that we need to be or that you are focusing more attention on than any other?
Certainly, I think virtual in the cloud and everything, because it is.
Right
at the fulcrum, at the heart, is a big part of our initiatives. It's also, you cannot forget, security is this multi-line of defense problem. You have to have belts, suspenders, and duct tape and baling wire, right? You got to have everything to be secure. You really do, right? Because it only takes that 1 crack, just like the video.
Well, that's right. It could be in the network, it could be in the data center, you could have the greatest endpoint security, but if you haven't done the other end.
Yeah
You have to do the whole thing. That's kind of the point.
You have to do the whole thing, yeah. I think, if you said, well, is there just kind of a couple areas, I think those would be it, which is the mobile devices and tablets, et cetera, then it's the cloud and what's happening with everybody pushing it into it.
What about these emerging areas of intelligent systems like cars? We never thought about having to secure our car.
from a virus attack. I assume that's a very potentially serious threat that you're looking at.
Yeah.
Anything that's connected.
Anything. That's right.
Anything.
My appeal is think about anything that can be connected, and if it can be communicated into or if it can communicate out, it can be hacked, and it can be compromised. We know there are many teams around the world that are targeting those systems for attack, and we've all already seen a number of research efforts.
Can you share them?
that have showed up. Yeah, I can share many things, but I want you guys to sleep okay tonight, so I will limit that. Yes, there are a lot of really bad people out there that really do focus on this stuff constantly.
In the discussion around security, and we're just about out of time here, so maybe a minute or so from each of you. If there's a message for the folks out in the audience about the security landscape, the McAfee Intel response to it, what would be that key message that you would leave them with? I'll start with you, Stuart.
Well, I think the key message is this is a problem that's not going to go away anytime soon, it's a multilayered problem that we have to solve with a multilayered solution, it has to start at hardware and go all the way up into the cloud, in every layer in between. The more that we drive home that tight connectedness, and the simpler we make it to deploy and use, the more it will be used and deployed, and the more people will be protected. From my perspective, just never forget how many things are getting connected these days, how many things have radio antennas and are connected because those are the things that we need to think about solving and protecting today.
Renee?
My final thought would be, we will hear more competitive noise about security. It's very clear that other people will start talking about it more because we're talking about it more, and security is extraordinarily complicated. BK's coming up next to talk about complicated things. Security has its own levels of complication. Intel is, I would say this, security features are necessary but insufficient to really change the nature of security. It is critical to understand that Intel is not just putting in silicon features, not just adding secure features to our platforms. We're solving security in a new way as a solution, which is something we've never done before and no one else in the industry is doing today.
Okay. Renée, Stuart, thank you very much.
Thank you.
Thank you.
Renée James and Stuart McClure, who have specific questions for her. We now turn our attention, Renée mentioned it, we turn our attention now to manufacturing. Our next speaker is Brian Krzanich. Of course, he is Intel's Chief Operating Officer, and in that capacity, he runs our leading-edge factory network of fabs and factories around the world. He is here to, as Paul mentioned, explain what these two devices are on each edge of the stage. Please welcome our Chief Operating Officer, Brian Krzanich.
Thanks, Brian. Good afternoon. If you're wondering who this BK guy is, that's me, too. I kind of go by both names around the company, mainly because my last name. There's a couple reasons for it. Last name was always kind of hard to do. Yeah, I'll talk about these in a couple minutes, but I kind of want to get to the topic of my discussion today. It's really, in my mind, kind of the heart of what the Technology and Manufacturing Group, and really manufacturing at Intel, is about. Whether it's big or small, it's all about the details. When we talk about our leadership, when we talk about our leading-edge technologies, all of those things, the way we look at it is it's about managing those details.
It's about really delivering day in, day out, as Paul said, the 2 million 22 nanometer units a week. It's really a detail that we manage. What I'm going to talk to you about today is tiny transistors that have a big impact. We'll talk about these two models. I'll invite a guest up to help me talk through that. Then we're going to talk about our big investments, which immediately gets to our factories. We'll spend a little time about where is that capital going, how are we investing it, and why is it so important? Okay. As Paul said, you can't have a slide at Investor Day, or you can't have an Investor Day without a slide of Bob Moore.
Gordon Moore.
Gordon Moore, sorry.
Bob Noyce, Gordon Moore. Really, the focus there is we have this history of every two years, and when we talk about managing the details, it's about really planning those two-year increments and really just saying, "We will deliver that next set of innovations. We will deliver that next set of products that will deliver on the Moore's Law." We're talking about 22 nanometers, and that's what we're going to talk about today. I thought rather than just show another Gordon Moore foil and another set of graphs, we'd spend a little bit of time about what this 22 nanometers. What really is 22 nanometers? The first thing I think you have to do is you have to kind of put that size in perspective.
If you took 22 nanometer transistors and you took as many as you could fit on the head of a pin, you'd get to about 100 million transistors could fit on the head of a pin. What's interesting is, as we went to go calculate this, there was more variation in the head-of-a-pin size than clearly in our transistors. That's back to that managing the details. The guys who make pins don't really worry that much, I guess, about the exact size of the head. We do. It's a little bit more than 100 million. We actually were relatively conservative, but that starts to give you a feeling for the size that we're really talking about, 100 million.
100 million itself is a large number, but when you talk about stacking 100 million of those flat on the head of a pin and they're all fitting, that really starts to give you a feeling for just the geometries we're working with. That gives you size. We thought, well, let's get to the next thing. Let's talk about speed, because really these are switches, and we really want to talk about the speed at which these switches work. If you took the speed of a 22 nanometer transistor, they switch on and off, they can switch on and off about 100 billion times a second. 100 billion's a big number, but we thought, well, let's try and put this into more humanistic terms.
If you or I stood at a light switch and tried to turn it off as fast as we could, we would stand there for about 2,000 years before we did what one of our transistors, just one, can do in one second. That gives you a little feeling, right? 100 million of those can fit on the head of a pin, and it would take 2,000 years to do what one of them can do in one second. That really talks to Paul's foil about really the advancements in computing and what's really been delivered in our products. All of those things kind of come together when you start to take a look at some of these just really simple metrics.
Now I thought we'd go into kind of a little bit of the physics, and I promise at the end of the day, you would be able to take a quiz that says, "How does a transistor work, and what's different about these 3D transistors?" To help me, because I don't know if I could do this on my own, I'd like to invite up Mark Bohr, who's one of our chief scientists here at Intel, and really the creator of these technologies.
Hello, Brian.
Hey, Mark. Mark works in our technology and development group and is really one of the founders of these technologies through the years. The Gordon Moore foil there with the technologies over the years. Mark, can you really tell us what's different about 22 nanometers? What did we do there that was different than all of the years in the past?
Well, we made a big switch in going from a planar transistor structure, something we've been using and scaling for about 40 years now, to a new three-dimensional Tri-Gate structure, as shown here on the left.
Got it. Seems simple. Okay, these are pretty big. Paul kind of talked about just how big they are. The first thing we ought to do is probably talk about just the scale that we're talking about here, right? Let's walk over to this one. This is I think what's been built in the past, the first thing we did was, again, we thought, well, let's talk about that 100 million of these on the head of a pin, and if you built those up as big as this, how big would it scale for somebody like me? I would be about twice the size of the Earth if you scaled me up the same amount we've had to scale these transistors.
That kind of gives you a flavor for just how much scaling we had to do to get them on the stage here today and visible for you guys. Mark, I think this is what we've been using in the past, and maybe if you could explain how this works and kind of what's fundamental about a planar transistor.
This is a planar transistor structure, as I mentioned, what we've been using for more than 40 years now. This would be the surface of the silicon wafer. The stripe here down the middle, those are the silicon source and drain regions. The black structures here, those are the field oxide regions that isolate one transistor from the one next to it. On top, we have the gate electrode, and right at the bottom of the gate electrode, we have a very thin high-K dielectric, and that high-K dielectric is right above the transistor channel where all the action occurs. Of course, as Brian was mentioning, a transistor is essentially an electrical switch. It can turn on and off, of course, much faster than a human can make it go on and off, but I can try something here.
This is a planar transistor in the off state. There's a low voltage on the gate. That voltage repels any carriers from the channel region, which, as you remember, is right underneath the gate electrode here, and thus there's no current can flow from source to drain. When you apply a high voltage to the gate electrode, then stray carriers are drawn to the surface of the channel right underneath the electrode, a conducting channel is formed, and current can flow from source to drain. Again, you can switch on and off the transistor just by changing the voltage on the gate electrode from its on state to its off state.
It seems pretty simple, and these are really what we've been using for years and years and years in the past. I think that a lot of the innovations here were things like High-k/Metal Gate and all that just made this a much more better performing transistor, right? What were some of the key aspects about trying to drive the performance in the past here?
Well, of course, the way we have scaled in the past, simply make the gate length shorter, there's less distance from the source to the drain, make the gate oxide thinner, or, as we did at the 45 nanometer generation, convert it to a High-K material. Those types of scaling techniques worked quite well for 30-some years in providing devices that were not only smaller, but could switch faster, provide more current, and less power.
At 22 nanometers, you guys went to a different model, maybe we could walk across the stage here, hopefully without hurting ourselves, talk about this guy.
Right away, you see a big difference between the older planar style and this new 3D Tri-Gate transistor. Now the source-drain regions, instead of being planar with the surface of the wafer, they're actually tall and narrow silicon fins, and the gate electrode wraps around that fin. When the transistor is in its on state, there are actually three conducting channels that are formed on that fin or in the channel, on the left side, the right side, and that small top section of the fin. That's why we refer to this as a Tri-Gate, because of the three conducting surfaces on the fin. Right now, the transistor is in the off state. Again, the gate electrode repels any stray carriers in the channel region. No current can flow from source to drain.
When you switch the gate electrode to a high voltage, there is a conducting channel formed, current can flow from the source to the drain. I think as you can all intuitively see, when you have the gate electrode wrapping around that fin, the gate electrode has much better control over what's going on inside that channel. When it's in its on state, more current can flow. When it's in its off state, it has less leakage compared to the previous planar device. Also because of that improved control of the gate electrode over the channel region, this device, the Tri-Gate device, can operate at lower voltage than the previous planar transistor, and that has important performance and power advantages.
These are the fins, and that's why oftentimes you hear it called a FinFET or a fin field-effect transistor, right?
Yeah. FinFET is another name for what we call a Tri-Gate. Yes.
Right. Tri-Gate because of the three gates, and sometimes you'll also hear it called a 3D transistor because we pulled it up out of the silicon and made it vertical.
That's right.
three names, but really all the same device, right?
Yes.
Maybe we could spend a little bit of time here. You kind of explained why this is better and what a better switch is, but maybe we can talk about, as you went through this, really, how did you guys come about deciding to use the 3D transistor?
Yes. As you see in the graph behind me, that's a graph of a transistor gate delay on the vertical scale versus operating voltage on the horizontal scale. The yellow curve reflects the delay versus voltage characteristics of Intel's present 32 nanometer planar transistor, which on the advent of Tri-Gate, they were the highest performing transistors in volume production.
Had we used the planar transistor, we would've gotten that gray curve, and it would've just followed like normal, correct?
During the early development phase of 22 nanometer, we initially thought, well, let's keep extending planar devices. That's something we know how to do. The planar structure, we recognized, was really running out of steam, or we were really running into limits for how well we could scale that device and still get good performance and power. Normally, what you want to do on any new technology is take that yellow curve and push it down. In other words, provide faster or higher performance transistors. If we had followed through and done a 22 nanometer planar technology, we would've pushed that curve down, but maybe by only about 15%. It would've been a pretty modest improvement over the types of improvements we've provided in the past.
Clearly that's not what we did. Let's take a look at what actually occurred.
Again, during the early development phase of 22 nanometer, we were also exploring the alternate idea of these 3D Tri-Gate devices. Our early test devices, along with models and simulations, said, boy, these devices have much better low voltage operating characteristics. Under high voltage conditions, we can provide maybe about an 18% speed up over our 32 nanometer planar devices. The real big advantage is at low operating voltage. At 0.7 volts, we were seeing about a 37% improvement. That's an unprecedented improvement in gate delay at that low voltage.
As we heard Hermann and Mike talk earlier, Paul also, about our move into mobile devices, you guys saw this as a real opportunity for those kinds of devices, but I think it extends even beyond that, correct?
As this curve implies, it provides a big performance gain at low voltage, but that's not the only benefit you can derive by using these 3D Tri-Gate transistors. The other way you can use these transistors is maybe there are some circuits, some products, where you're not looking for a performance gain, but you're looking for lower power. These devices can operate about 200 millivolts lower voltage than our previous 32 nanometer planar devices and have the same performance. That voltage reduction has a big benefit in meeting active power. The combination of that lower operating voltage and the smaller size of the transistors provides about a 50% reduction in active power. That's a big advantage over both mobile products that value low power, but also high-performance servers that value energy-efficient computing.
This is physics. This isn't marketing, right? No matter who makes a planar transistor versus a Tri-Gate transistor, this is the way it's going to be, correct?
You're right. This is not marketing. Yeah.
I think that's important because as people go out there and say they're going to move down to lower geometries, and they think they can avoid the Tri-Gate transistor or 3D transistor, this rule will apply to them. They'll get that gray curve instead of that blue curve.
Yeah. This is physics. I think it's been widely recognized in our industry of the potential value of these 3D Tri-Gate or FinFET devices. Everybody would like to have Tri-Gate or FinFET technology today. It's tough. It's tough to get all the manufacturing details done right to deliver the performance and power that you really want. Intel has this technology at 22 nanometers. All of our competitors have announced their plans to use them at their coming 14-nanometer technology.
Right. Again, another advantage. Let's talk about that because at least from the manufacturing side, I've seen a big shift. Paul kind of talked about it a little bit this morning, about how things have changed in our industry from a technology standpoint. Maybe we can talk a little bit about just how that's changed.
Up until 10 years ago, we were just following normal scaling law, just taking the planar structure, reduce all the horizontal and vertical dimensions, and we would get what we wanted, a device that was smaller and faster. Around the 90-nanometer generation, we discovered, we recognized that traditional scaling techniques were no longer delivering all of the benefits that they previously had. We had to start inventing new things to continue scaling. The first thing we invented here at Intel was the use of silicon-germanium strained silicon. We were first to implement that on our 90-nanometer technology in 2003, and it was about three years later before our next competitor was able to copy that technique or use that technique on their own process. In 2007, we realized we had to come back to the issue of scaling the gate oxide.
Could no longer continue scaling SiO2, we had to invent and introduce High-k/Metal Gate materials. We did that, Intel, on the 45-nanometer technology, and it was about three and a half years later, before our next nearest competitor was able to start shipping their version of High-k/Metal Gate. Here we are in 2012, shipping 22-nanometer products with Tri-Gate transistors. Our competitors, based on their own announcements, appear to be about three to four years behind us, before they'll have Tri-Gate or its equivalent.
That's good. I think this is another important fundamental concept here when we talk about why is the industry changing and shifting, what's really different. It's really pretty simple, right? It went from a world of just scaling to a point where different parts of the transistor were getting so small, we had to change the material science and the physics within there. I always look at it as we now have to do two jobs. Your job became twice as hard. You have to figure out how to make it smaller, you also have to deal with the changes that occur as these segments of the transistor get so small they start to break down.
Yeah. We are in an era where we have to now continually invent new materials and new structures to continue scaling. I believe Intel is really leading our industry in doing that.
That's great. Paul really talked about the advantage of being an integrated device manufacturer or an IDM. I see it every day in our work, I think it would be neat to hear it from you, from the creator of the technologies and how you really use the IDM process.
I really see the value of having a research group, a process development team, manufacturing fabs, and circuit design teams all under the same umbrella, all working together in one company. Where we can not only select the right technology features, we know how to best co-optimize the process technology with circuit design to deliver compelling products. When in manufacturing, as it always occurs, there is always some minor hiccup, we have the ability to work together with the circuit design team to solve that problem right away without doing finger-pointing about whether it's a process problem or a circuit design problem. We fix that because we're working together and we're working for the same company.
I did notice you chose the manufacturing hiccup, though. The little hiccup.
Really brief. Short.
Mark gave a great description there. Paul kind of mentioned it, but I see it every day. I sit at a meeting twice a week that we have run for probably, I think now six or seven years called Output Max. You would think an output meeting, a meeting focused on output of the factories, we have a bunch of factory guys sitting in there, but it's amazing. Especially as we launch a new technology like 22 nanometers. We have some guys from Mark's team in there helping us with, "Well, here's how we created the technology. If you're seeing an issue, this is what it is." We have the designers from Dadi's team in there, "Well, here's how we design the product.
If we're seeing an issue, hey, we could change in the next stepping, and we could fix this." We have all of the supply chain in there, all the way from guys who buy the stuff to the guys who are out there marketing and selling it, saying, "Here's how we can adapt the whole supply chain to fix these issues." Everybody is optimizing for the end result, which is the most number of units at the lowest cost to our customers as quickly as possible. That's occurring every day at Intel, and I can't imagine it occurring any other way.
Right.
Mark, I hate to be the bearer of bad news, but maybe it was your comment about the manufacturing hiccups, for a while now, people have been kind of telling me that you're going to be out of a job in a couple of years, right? That the end of scaling is near, and as early as the last couple of weeks, there was somebody speculating that Moore's Law was dead. What do you think?
Well, I've been in this industry for more than 30 years, and the end of Moore's Law has always been 10 years away. Yes, it will always be 10 years away.
Okay. Let's talk about what that 10 years looks like for you right now, at least.
Okay. Well, of course, we have a 22 nanometer Ivy Bridge in volume production in three factories, and a fourth factory coming up later this year, all at our high yields. I know the 14 nanometer technology is in full development up in Oregon, have a large team of engineers working on that's on track for manufacturing readiness in the second half of next year. I'm personally spending most of my time now on the 10 nanometer generation, and I know we have solutions generation. We have a very talented components research group located up in Oregon that is exploring a range of very interesting ideas for seven and five nanometer technologies.
That's roughly a 10-year view out there. Is that any different than what we've had in the past?
That's pretty much it. For many years now, that's about our limit of visibility, about 10 years out.
Yeah. If we come back here next year, we'll still see another 10 years is your prediction.
I-
Just in general, our view
I expect we will, yeah.
Okay. Excellent. Mark, you've answered a huge number of questions. You've helped me explain how these transistors work. I really want to thank you for coming up here and helping us out.
Great. Thank you.
Thanks, Mark. It's really an honor to get Mark to take time out of his day working on 10 nanometers to come up here and really explain this to us. I think it adds a lot of value for you guys to hear from the people who create this technology, just what they did and how it got here. I want to get off the small for a second, for the rest of my presentation, in fact, and go to the big. That's when you start talking about our manufacturing scale. I think you've seen these kind of charts before. We show up. Here's where all our factories are. I want to spend a couple of minutes and just talk about the scale. Paul mentioned a little bit this morning about where we're putting 22 nanometers.
We're ramping it in Oregon, Arizona, and Israel. Israel's coming up quite nice. It's the last portion of our segment. We've actually made some efficiencies and actually been able to consolidate that supply chain and get it down to three factories. We're very proud of that. We have about 4 million sq ft of clean room that we're able to bring to bear to these technologies. The size and scope of that is about $36 billion of total investment. If you wanted to go build those buildings and fill them full of equipment, you're really talking about a large investment and effort. What you don't know is all the intellectual property or information that goes into even those construction of those buildings. We have 2 buildings, 2 clean rooms under construction right now: D1X up in Oregon and Fab 42 in Arizona.
They are slated to come online. Both of them are actually clean rooms right now. They're in various, what we call blowdown, which is basically blowing air through the filtration systems and cleaning them up. D1X will come up ready for install of equipment in Q4 of this year. Fab 42 will come up in Q1 of next year, right on schedule. I review their schedules every Friday morning at 7:00 A.M. They are within 5 days of schedule, which when you think about a 2-year project and roughly $2 billion in spending, they are within $10 million of budget. These things are being executed flawlessly. We're going to bring those to bear, those will be the latest generation. They are 450-millimeter capable as they stand.
We've designed them for 7 generations, for Mark's prediction of Moore's Law, that they will be able to manage that with their existing facilities infrastructure. That kind of gives you the size and the scale and what we're trying to bring to the manufacturing supply chain here. I want to talk a little bit about, I think last year Andy kind of showed you guys the cost of putting together a factory and a development team, then how much revenue you'd have to have. Paul kind of mentioned today the cost of factories. I thought I'd start with that and kind of talk about, well, why is this scale so important? Mark talked about how the technologies have shifted, and we're having to go down 2 vectors and the amount of innovation. People have talked about how this is getting more expensive.
Paul also showed that we're able to keep the cost per transistor down. I just want to show you what the result of all this has been. If you go back to 200-millimeter, you can see that it was a $3 billion-$5 billion revenue threshold. Paul showed you it's a billion dollars or less for the fab. You can just go through and figure out, "Oh, then I've got to go fill it full of equipment. I got to have a development team," all that kind of stuff. You can see why a lot of people, this gets to Paul's comment earlier this morning. Everybody had a fab, right? It really made sense. What happened is we got to the 300-millimeter, Paul said it was the 80s, 90s, and early 2000s.
The fabless model came into play, you can see quite a few people dropped off being able to afford a new clean room. It made sense. Well, let's consolidate into the fabless model. This makes sense. What's happening now is another transition that's occurring, that is that these things are getting even more expensive as the amount of innovation is occurring, as the square foot that you're required to produce a wafer is increasing. We're seeing 300 millimeter fabs become more expensive, that's what you're seeing is fewer and fewer players are able to afford these, there'll be further consolidation and transitions in our industry. You could probably guess what'll happen at 450. The line will probably move to the right.
That's why we believe that there's a real advantage in the Intel integrated device manufacturing and the scale that we're able to bring to these technologies. With that, I want to close my talk this morning, or this afternoon, just really tell you that what we're doing here is what we've done in the past and what we've always done, what we'll continue to do. That's really delivering in the details, whether it's in the small, bringing new innovation to transistors, or in the large, bringing the large manufacturing scale to production. We're going to continue to extend our technology leadership. You heard why we made decisions to go to the 3D transistor. It made complete sense. It's basic physics. Others may say we don't need it, but you saw the power and performance advantages it gives you. You can't walk away from that.
It's there. It's undeniable. We're going to continue to make investments. You saw our two factories. We're going to finish those constructions. We're going to fill them full of equipment. We have the scale to deliver, we really think that's going to lead us into the future with the products that Kirk and Diane and Hermann and Mike have all shown you today. With that, I'm done. I'd like to thank you for this presentation. I'd like to invite up the guy who's going to sell everything I'm going to build, that's Tom Kilroy, who's our Sales and Marketing Manager and Vice President. Tom.
Well, my job's safe. I've got a lot of years I had with some big sales to deliver on what Brian and Mark talked about. I'm actually here not to talk about products. I'm here to talk about markets, specifically emerging markets, which represents a big growth opportunity for the company. Before I do that, I'd like to spend 30 seconds just giving you a little read on how we see the quarter materializing. Paul mentioned this morning, the assumptions based on what we gave you for Q2 are playing out in the market pretty much the way we guided it. Our expectations for-- There's so much talk about Western Europe. There has been for over a year, our expectations are fairly modest, what we're seeing isn't surprising us.
The consumer segments remain somewhat sluggish in the mature markets, but the commercial and business, small, mid-size, large enterprise, playing out pretty strong, as are emerging markets. We feel that the number we gave for the second quarter is a good one, and we feel the number we gave for the year is a good one, based upon everything we're seeing. With that said, I want to talk about the emerging markets, which really has been a major growth driver of our business the last decade. As we saw that, we've made big investments to make sure it's a growth driver for the decade ahead. Let's get started. To begin with, you can see the color shades of the map, generally, the darker area is what we're talking about today, emerging markets.
That said, as I go through the presentation, I'll speak to the fact that even in that darker shade, there's a little bit of a different color based upon the maturity or the socioeconomic class that exists in those emerging countries. What I want to do is spend the front end of the presentation on market trends. Really, mostly third-party data. Not what Intel thinks, what Intel sees, but what third parties all around the world are seeing. I want to say or speak to the opportunity as we see it, how that's driving the go-to-market plans we have at Intel, then wrap up with our unique value. To start with, some 20 years ago, this picture of Jakarta is pretty typical of how we would see an emerging market when we hit the road, went out, and visited emerging markets.
Generally speaking, infrastructure was pretty poor, this could speak to many areas within China, Russia, Brazil, India. Limited technology adoption. Some was beginning, but overall limited. What was promising was this multidimensional growth that we saw 20 years ago. This is when I mentioned Intel was an early mover in investing. It was the promise that was ahead. If you go look at Jakarta today, the capital of Indonesia, it's actually the second-largest metropolis in the world, second only to Tokyo. Indonesia, as a country, for the last five years, has the highest TAM CAGR for PC growth. This is what I mean when I say, is Jakarta an emerging market or a market that's emerged? The answer is both, I'll explain what I mean there.
One indication of how much opportunity ahead is still in Jakarta is the PC penetration rate is still only 16%. Within Indonesia, the country of Indonesia, even though Jakarta is a little bit more advanced, huge opportunity ahead. This is a great example of what has fueled our growth. When we've talked for years and years about the importance of emerging markets, this has been a major focus for a reason. This growth CAGR, as shown by IDC, consistent 18% growth. This is nothing new. We've talked about it investor meetings, every presentation we give. Emerging markets is a critical part of our business model. I think many of you are probably asking the question, "Is this engine going to keep humming?" The best way to look at it is to find out what's behind this growth engine.
The number one thing is just the huge, massive population and the growth rate in the population to begin with. The other element that's quite interesting is the population is approximately 10 years younger than mature markets. You can see the rising middle class. There are more middle-class people in emerging markets than in mature markets. That's important because when we talk about computing today, we're talking about users. We're not talking about big iron and infrastructure. We're talking about individual people that want to get connected to each other through the internet. The numbers of people really start making a much bigger difference today than maybe it would have been five or 10 years ago.
The purchasing power of the middle class will actually surpass that, in the emerging markets, will surpass the middle-class purchasing power within the next three years, which is even more significant. It's not just the big numbers of people out there, but it's their purchasing power to go to work. You can see the staggering takeaway that in emerging markets, consumer spending is going to grow to $20 trillion, which is twice that of the U.S. today. Amazing opportunity here, but it's also a population that is really drawn to technology. When you measure drawn to technology, the way you do it is look at the race to online. The crossover happened in 2008. If you look at this population, the migration to online has grown a factor of 18x since 2000.
I mean, the numbers are big, but I think it's really important to step back and think about the significance of them. Think about China alone. There's a half a billion internet users in China. They long surpassed the U.S., 4 years ago, as far as connected to the net. The other thing that's quite interesting, if you see the chart to my right, is the internet penetration is broad. To my example about Jakarta, being the second-largest metropolis in the world and more mature, if you will, than many other Indonesian cities, it's broad. If you look at the last three years, you can see there's not that big of a variance. Getting online is the first way that people get connected and bring this opportunity that we see in front of us to reality. You say, "Okay, good. What are they doing?
What are they doing online?" The answer is, it's pretty advanced. In the U.S., we talk about social networking, Facebook and Twitter. The U.S. has been outpaced by Brazil, Indonesia, Russia, India. If you take a look at the percentage of internet users on social networking, the emerging markets are leading the way. It's not just the percentage of users. If you look at the time spent on social networking, eight out of the top 10 markets of time spent per person on social networking is emerging markets. If you look at the news, the media's tightly controlled or highly controlled in many markets. Picking up the newspaper, television isn't the way many people want to get the news in emerging markets. They want to go online and find out what's going around the world. You look at entertainment. You look at video viewership.
70% of YouTube traffic is outside the borders of the U.S. As you can see here, 18 of the top 20 online markets for video are in emerging markets. You see the pattern here, whether it's social networking, getting the news, being entertained. You see gaming, online gaming, again, China way outpacing the U.S. This is a market that is embracing technology. You might say, "Great. Well, how are they doing it? What type of device?" The question here is, myth or reality, are smartphones the primary on-ramp for most of these compute experiences? The answer is no. No, the PC is resoundingly, it is the dominant internet access device for emerging markets. Whether it's from the home, whether it's from work, whether it's from i-cafes, it's the PC by a long shot. It makes sense when you take a look at it.
When you look at the clear usage cases that drive this, I'm going to show you, it makes good sense as to why the PC is so much far out ahead. The point, though, is even if the PC is so far out ahead, the good news for the industry is there's a strong desire for multiple devices in emerging markets. If you look here and look at device preferences by usage, you see it changes. Phone, not surprisingly, most people use a phone for consumption activity. A PC, very much for social networking, as I talked about, banking, shopping, video consumption. If you look to the right part of the slide, it's a pretty interesting figure. It's just demonstrating how density is increasing. 82% of smartphone owners own a PC, and there's a percentage of those that don't that want to own a PC.
This is increasing every year. I think last year it was two years ago, it was 77%. People in emerging markets, much like mature markets, are striving for multiple devices. The other interesting stat I found was If you say smartphones really came to market around 2003, if you looked at the PC CAGR, the five years before the advent of the smartphone, it was 25%. I'm sorry, it was 21%. If you look at the five years after 2003, when the smartphone was introduced, it was 25. The PC is growing quite robustly while the smartphone isn't. This is a great opportunity. It's a proof point that density is increasing and a wonderful opportunity for the industry. It's obvious there's a huge thirst for technology. I now want to shift to, what about the opportunity?
To start with, I want to define a little bit more about the PC penetration dynamics and drill a little bit into that. Brazil is a great example of really distinct segmentation. If you look here, you can see there's the socioeconomic class A and B, then C through E. These definitions are fairly consistent based on income, education, goods ownership, and housing. Whether you're looking at this kind of class structure in Brazil or India or China, relatively consistently, the numbers and income might vary, but it's a good example of class A and B is very much like a mature market. You can see 81% penetration rate, whereas the class C through E is almost no penetration. Much more indicative of a true emerging market.
That's when I said within emerging markets, there's, whether you want to say tier 1 cities or socioeconomic class B, are very comparable to the mature markets that we have in the U.S. and Western Europe as well. For us, the A/B segment is very much one where we're going to get refresh. When the question came up this morning about who's going to be motivated to refresh the Ultrabooks, guess what? It's the A/B segment in Brazil, in China. These people have the income level and the desire, as you can see from the earlier charts, to get online and participate in the digital economy. On the other hand, the C and E is where this first-time buyer opportunity is. I want to tell you a little story.
I was in Brazil two weeks ago, our market insights group that works in my organization came up with this program where executives like myself travel all over the world. Instead of just doing big end-user visits, meeting with government officials, retail malls, go into a home of a first-time buyer. We did. We went into the home of a class C family based upon this definition here, walked in the home and spent, ended up being about 35 minutes with this family that just bought a PC. We met with the father. He's a security guard by night, so it was 2:00 P.M. He was home with his younger daughter. His older son was at work, his other older son was in school, his wife was working.
Not all that typical or atypical multi income because the older son was working. I asked him, I said, "Why a PC?" He started off by saying-- He was very humble and said, "I have a great life, very great life, but I have a tough life." He was saying, "It's tough making ends meet." I asked him, I said, "What was it that got you to the point of wanting to buy a PC?" He looked at his daughter, he said it was her. It was all translated, of course, from Portuguese, and he said it was her for education. That is not uncommon at all. Education is the number 1 driver for this in terms of motivations for emerging markets to buy their first PC. They want their student to perform better in school.
The other thing he said is his daughter had used a PC for years, as did his sons, but they had to go down the street to an internet cafe. He said he was worried about their security, so it was time for him to prioritize a PC for the home. I asked him, I said, "Do you have smartphones?" He didn't. They all had phones, but they were not smartphones. I was trying to understand the dynamic, and he kept moving back to the PC. It's a family PC. Interestingly enough, every single one of them was a Facebook user. The son was a big gamer at night, and of course, the daughter was using it for education. When I was trying to understand, would he buy a smartphone next? He said no, but he would buy another PC.
What's happened is it's in the home now, and they're enjoying it for many different usages. This is an example of a dynamic out there. The world is still so untapped as far as people that have not gotten a PC and speaks to the fact that the desirability is every bit as important as the affordability. I want to talk about affordability now, because for years, we've talked about how once you get in a range of that 3 to 4 weeks of income, okay, yearly income, you can get into the range of affordability of a PC becoming realistic. Brazil is an example where 5 years ago, the PC penetration rate was 18%. It's now the number 3 market, as Paul said, in the world, 37% penetration rate. If you look at the weeks of income to buy a PC, it's now 3.4.
Brazil, that green, has moved up, and it will continue to move up. There's other markets that are still huge. Look at the population of India. As Paul said, broadband will start moving that right. There's many markets out there where affordability is going to help fuel the growth of first-time purchases. Reality is, only one-third of that really makes up the emerging market TAM today. It's very important, but as I said, density, meaning multiple PCs in a family or refresh, is very big and important as well. Let's take a little bit more of a look at usage models for that first-time buyer. Again, the smartphone, largely used for voice, IMing, or quick snacking, if you will, on social networking or access. Generally, the tablet's fairly rare in emerging markets, the exception being education. For school, we're seeing more of that.
If you look at the PC, again, the consistent usages you find here are this window into the first-time buyer. Education for children, very strong. Best for family sharing. The family I visited had bought a desktop, very typical, because it's to share for the family. Advancement at work, productivity, and again, this pervasive desire on social networking is a theme everywhere. These are the usages that are primarily PC-oriented. The point of it is, whether it's a smartphone, a tablet, or a PC, in combination, all of this is driving just tremendous, robust growth for devices. If you look at this Gartner data, smartphones a 24% CAGR through 2016, tablets 53%, and the PC business itself, out over the next 4 years, projected by Gartner to be 18%.
We see this as great, where we're intercepting, as you've heard this morning and from Brian, that we're optimizing our architecture and our capabilities to go win in phone and tablet, and we've got this continued driver of our Core business, which has been so great for us in the past, and we think going forward, which is the emerging markets. This is a great catalyst for unit growth, but another question that often comes up is, are emerging markets buying a lower-end mix of products? Unit growth sounds great. What about ASP? You might be surprised to see that generally speaking, on the left would be all form factors, and this would be the mature market adoption rate of Core. This would be in retail.
The Core mix in mature markets is ahead, that's the green, it's ahead of the emerging markets, if you look at all form factors. I mentioned many first-time buyers buy desktop. You'll see that mix is a little bit lower. If you move over to notebook, it's a dead heat. You see that, again, this is retail, and adoption of Core. Not Celeron, not Pentium, they're moving right up to Core. This is something we've seen fairly consistently over the years. The emerging markets are drivers of our new technology, every bit as much as the Tier 1 markets in mature markets as well. The Class A and B that I showed you in Brazil or China, if you look at Shanghai, Beijing, behave like mature markets. By the way, first-time buyers don't like to buy the cheapest. We learned with netbook.
When we came out with netbook, we thought it would be TAM expansive in terms of reaching more first-time buyers. What we found out was the answer is no, because first-time buyers, it's a big deal for them. This gentleman that mentioned he's got a good life but a tough life, it was a stretch for him to get a PC. He was not in that category that we talked about. For him, he wanted to buy something that was going to be something he could be proud of and his family could enjoy. We've talked about units and devices. We've talked about the mix. There's a catalyst for growth that goes beyond that in emerging markets, and that's infrastructure.
If you look at just Data Center, or as Diane said this morning, traditional IT, clearly mature markets are still. When you look at the amount of dollars or annual IT spend, the mature markets are well ahead of emerging. The growth rate in emerging is significant, it's a 16% CAGR. It's growing at three times the rate of mature markets. As you move over to the right, the cloud dynamic is quite phenomenal because what's happening is, as cloud usage grows, local clouds or domestic cloud service providers are emerging. If you go look at Tencent, Baidu, Alibaba, those are three big players. You look at 700 million users or whatever the number is with Tencent. In China, these local players are growing, serving up content to the population in China.
These three companies, along with a number of others, as far as deploying servers and storage, are growing at a rate 3x that of the U.S. IP Data Centers that you're all very familiar with. 3x. It seems stunning, but then if you step back and say, there's a half a billion Internet users in China. What they're doing is when they're online, it's continuing to outstrip the capacity that's in place in China. My last visit to China, I was mentioning to Diane, I had a lunch meeting with a bunch of, it was 13 startup companies. This is sponsored by the government. It's a national program, and a few service providers. It's called Cloud Valley. It's their Silicon Valley as far as really deploying cloud.
Their vision is to build out a massive data center in Inner Mongolia because they're going to take advantage of basically free cooling. They're collaborating together. No problem with heat up there. The point of it is, we're not talking about hundreds of thousands of servers and storage. We're talking about millions of servers deployed. They're bringing technology companies like Intel in to consult and figure out how they go do this. When you look at the investment that's happening, it's happening in traditional IT, and it's happening in cloud as well. Another thing that's going on is government, and government has always been the big spender, but now it's even for a different reason. It's national security and economic competitiveness.
If you look at high-performance computing alone, you can see the number of TOP500 listings are up 2x in the last three years when you look at overall emerging markets. China's number two on the HPC 500 with 74 entries alone. It's more than that. It's not just national security. It's not just for economic competitiveness. When you look at smart cities and wonder what's going on, smart cities are about deploying technology in cities to make them more secure, to manage population, to manage traffic growth. If you go look at utilities, we've worked with a number of companies in Brazil to build the gas station of the future, which you go get gas and, of course, there's sensors everywhere, and the objective is to make it a secure experience, but enjoyable, where there's videos playing while you're pumping your gas.
You can do some shopping. These centers are deployed with sensors. I can give a number of examples, including in farming, but the point of it is beyond traditional compute, this intelligent deployment of computing is happening all over the world and in staggering volumes, especially in the emerging markets. Stunning growth from devices to infrastructure. What about Intel? What about our unique value? I want to take a quick look back at what we've done for decades and how, looking forward, we feel this is a clear competitive advantage to capture this growth. First of all, we've set up shop for decades, and it's paying off in the sense that the disproportionate investments we were making in our business to emerging markets 15 years ago, 10 years ago, is leading to not just high awareness for Intel, the brand, but preference for the brand as well.
Let's take a look closer as to what's behind that. If you take a look at the distribution network alone that's been built over the past 20-plus years. I was fortunate to run Intel's distribution business back in the '90s. In 1994, we started our box processor business, which was officially engaging the white box channel. We started that business not in New York or San Francisco or London, we started it in China. It was for a reason. We saw what was happening with these small shops, that people wanted to buy from local players. We went out and enabled all over the world a network that was, at that time, a white box channel, that today is much more than a white box channel. It was engagement over the course of the last 20 years.
This slide talks about just partnerships we have with this channel, which are small integrators. Today, they're not so much building white boxes. Some of them still are building desktops. As I mentioned, it's a big form factor in emerging markets still, but they're reselling M&C systems. Our engagement model is one of systematic training and engagement to advocate technology, Intel-based solutions. It also gives us unique visibility. When we mention we've got this unique visibility into what's going on in emerging markets, it's because of a network we've put in for the past 20-plus years. It gives us that unique visibility into what patterns are happening in sales out, what have you. If you move over to the right, increasingly, the compute experience is being enjoyed by consumers.
We've shifted from a lot of the enabling and training from some of these small shops to engage in the large format retail, large PC malls, whether it's 5,000 large format retails or 25,000 small retailers. We have an engagement model that's really built strong partnerships leading to a lot of trust and desire. When we do events in China or Brazil, all these thousands of players want to come and engage with us because we're kind of helping them map the forward-looking trends and what they should be investing in. Today, though, it's more than hardware. It's enabling and investing in solutions. If you look at the complete solution starting with broadband, we can look at that affordability model for a PC, but we can't get fooled. Because today, 80% of the world can't afford broadband.
This is something we saw was an inhibitor for amount of time. Paul mentioned this in India. We borrowed a page from the mobile phone industry here, where we had a program where we went out and we engaged the telcos. We certainly worked with government to help reduce taxes on broadband, but just by coming up with a model to have prepaid, where you can have prepaid bundles and preloaded content, makes a big difference. Because it's the total cost of ownership. If someone's on the hook for a monthly bill, it's affording not just the hardware, it's that as well. We've got programs to address the third billion. Two billion people are connected to the internet. We're going after that next billion, and this is one of the ways to do it.
If you move over to the software network, this is the moral equivalent of whoever's standing here in 10 years or 15 years looking back to say we were engaged with software, and the work that Renée's team does, it's optimizing applications for local software. These small communities in emerging markets care much more about the development of the local software ecosystem than they do what's happening back in the U.S. We've got a comprehensive program engaged with 2.3 million developers, and we're also investing. Intel Capital is not just out to put investment dollars in play to grow a portfolio, but it's around enablement, and close to $1 billion invested in emerging markets across 281 companies, 26 countries. All of this is helping develop an ecosystem that goes well beyond just the hardware.
I thought I'd conclude with just a quick view of a strategy that's been in action for 25-plus years. China alone, as I mentioned, our brand awareness is 93%, but the Intel favorability is 98%. If you go look at, we set up shop back in 1985. Today, we have 8,000 employees in China at 17 sites. If you go look at just the broad amount of distribution and retail presence and, as I mentioned earlier, the software engagement we have, the commitment to improving education is a big deal that we've done all over the world. Here, you see it a lot in the U.S., but we've done a ton in emerging markets as well. For the last eight years, we've been recognized in China by the Ministry of Education as far as being just a strong partner there.
These are the kind of investments that make the payoff when I talk about how important it is to really understand what's ahead and really develop that trust and sense of partnership in the local communities. We've really accrued great value the last two decades. Really, when we talk about why emerging markets are a growth engine ahead, it's for good reason. Our growth is certainly being fueled by many different drivers, and the good news is we've got phones and tablets as far as what's ahead of us, in addition to the great opportunity with our infrastructure business and the PC business. We have 25-plus years of history of where we've done engagement, and we've built real brand equity. That's why the Intel brand is what it is. It's not just because of what we've done in the United States.
The other, probably the most important takeaway, I think is emerging markets' desire to draw even or ahead is a reality to mature markets. This is how they think, whether it's businesses or cloud providers or consumers, they desire to be at where the mature market counterparts are, and in many cases, ahead. This gives us the reason why we think the emerging markets is a great growth engine for this company. Thank you.
All right. We'll have another round of Q&A. I'll invite Renée and Brian to join us up here. Again, we'll have the microphones in each of the aisles here, and if we can try and identify folks who didn't get a chance to ask questions the first time around, and then we'll come back around to those who did. Great to Uche here first.
Thanks. All right. I have two questions. Let me start over here.
Got you.
I've been in the darkness. That's fine, too. Just real quickly. Brian, you talked a lot about all the products, but I also remember, I think in the last couple of years, you've had charts to show improvements in cycle time as one of the key enablers of Intel investment. As product complexity increases, and you're seeing new products coming out in the smartphone area as well, are you able to still continue to deliver that improvement in cycle time and continue to drive high ROI with this level of investing?
Sure. Paul showed you the cost per transistor, and so when you talk about high ROI, that really delivers it, right? We're showing a continued decrease in our cost per transistor. That's really the fundamental of that return on the investment. Your first part of your question is, are we able to still drive the throughput time reductions? Yes. It kind of works like this. The mature technologies, we are continuing to make improvements in. 32 nanometers, 45 nanometers, all the way down. We still run some 130-nanometer technology, believe it or not, on 200-millimeter wafers. We are still making improvements in those throughput times.
In something like 22 nanometers, what we first focus on is getting it from how it comes out of the development factories, which is not at the same throughput times that we'd like our leading-edge technologies, getting it down to what we call our benchmark throughput time. That's what we're going through in really the first year of production, and then it'll go into that continuous improvement mode. Yes, we are able to get even the new technologies, even with those increased number of steps, down to those same throughput time levels.
In addition to that, because when I looked at the roadmap that you showed at the end, one of the future options you had, I guess, towards EUV. There's been talk in the industry as to when EUV will be manufacturable. Can you give us a sense as to, from the roadmap you've seen, when it will be absolutely necessary for Intel to insert EUV?
Yeah.
Then, small part that's still struggling to understand how you can still keep the cost per transistor down in the future. If you can explain that a little bit more, that'd be helpful. Thanks.
Well, if I explain how we keep the cost of transistors going down and not the other guys, that would be giving them the answer, right? I think it's really, again, designing from really the creation of the technology. We have an employee bonus on 14 nanometers with everybody in this company focused on the cost per transistor for 14 nanometer to get it on that curve. It's really, again, the designers, Mark and his team as they develop the technology, and the team that runs manufacturing and develops the supply chain, we can all make decisions and trade-offs to get to that cost per transistor, right? Dadi and his team can choose how many metal layers they want and how easy it is to design the process.
We can add metal layers and add to the cost of the production, but it lowers the die size and hence the cost per transistor because the die is smaller, and I don't have to build as many wafers. There's all kinds of trade-offs like that you can make that we make in a multiyear process to get to that cost per transistor. Our pay as a corporation, everybody in this corporation is aligned to that. It's very unique, and you get that with an IDM, I don't think you get that many other ways. Your other question was EUV. We believe we can continue to go. Mark would tell you that his 10 nanometer and his 7 nanometer, we believe we can continue on down there, whether or not EUV comes onto the industry table or not.
If I could predict exactly when EUV would come, I'd probably at least be out betting the stock market or something. It's not really clear to me right now when it'll be available.
Uche, we're going to spend a little bit more time on the cost per transistor question that you had in Stacy's keynote. He'll address that in some detail.
Maybe we'll
Yes. Stacy will give you the technical details.
Stacy just said that he'd go through the technical details.
It's marketing, so.
We've got one right here. Go ahead, David Wong.
Thanks very much. David Wong, Wells Fargo. Actually, still on the topic of EUV, can you give us some idea as to what the biggest issues are that the industry has to solve before we get to EUV? Does the cost of manufacturing actually drop when you get to EUV because you don't have to do all the complicated things you're doing now in order to properly?
You have two questions there. What are the problems to solve? I think some of them become obvious. The easy one right now is light source, getting a light source with enough power to get the throughput time up. Controlling the lens as it's all in a vacuum. The lenses are very big and getting the numerical aperture correct, then defects. We don't have a pellicle system defined yet for masks on EUV. Pellicles are something, for those of us who've been around for years and years, they are protection on the top of a mask that's outside the focal distance and has saved billions of die from defects. We don't have one of those for EUV yet. That could be a very big stumbling block.
If you want to know my personal opinion, the hardest part of EUV will be when we try and run it every day and generate 2 million units or more a week. It's going to be a very complex tool and a brand new sets of technologies, and it'll be just a huge task. When I think about it's going to be managing the details of how we do it. What was the second part of your question?
Does the cost drop when EUV-
I think it'll be interesting. It depends on when its intercept is and what its run rate is. There's talks of if you can get about 50 or 60 wafers an hour and it comes in early enough, then it could lower the cost. If it's lower than 50 to 60 wafers an hour or it comes in too late, you could have to introduce EUV and still do multi-patterning. I think it's all going to depend on time.
Right. I have $1 for anyone who has a non-EUV question.
I have five.
Only $1?
It's only worth $1 to me. Maybe worth more to you.
It might be worth five to me, though.
$1.
I'm excited for my dollar.
Okay.
Capital. I have one for Renée. Clearly, you guys have talked about speech and gestures, that awareness pillar is becoming more important.
I'm curious what you're doing with your organization to ensure that when I have an Ultrabook in 2013 and I talk to that Ultrabook, that the experience I have on IA is superior in speech because you have this publicly announced partnership with Nuance. This has to be about more than hardware. What are you doing alongside the silicon-
Oh, of course
with your partners at Nuance to make a difference?
I'm not at liberty to speak specifically about what we're doing with Nuance, but generically, I can talk about what we're doing around awareness. Awareness is more than just speech. Speech is a very important next I/O modality that we're working very hard on. Of course, we're doing performance work. It's algorithmic. The tuning of that work is the normal kinds of things that we would do. In addition to that, working on the integration of the user experience and how that shows up to the user with the other modalities. Awareness is also knowing where you are, the other things that come out of the sensors that Kirk talked about that are going into Ultrabooks and tablets and other. Working on context, location, how all of the sensors present data to services.
The user is using an Ultrabook and can have a similar, what I would call modern mobile experience as they might have on a smartphone around the output of the sensors as well as the new I/O modalities.
Gotcha.
All right, we've got three in a row.
You owe him $1 now.
Excuse me?
I owe him five.
One here, then we'll come back around and then to here.
Great, thanks. C.J. Muse with Barclays. I guess two questions for you, Brian. First one, as you think of moving to 14 nanometer, increasing number of critical layers, adoption of double patterning, how should we think about capital intensity as we go 22 to 14? Then the second question is, as you think about the lead time advantage in manufacturing versus pretty much all of your competitors, and how that is probably increasing from here, how should we think about the impact on 450? I would imagine that there's potential risk that could get delayed as folks spend more time on material science, et cetera, as opposed to looking to go to the wafer transition. Thanks.
BK, maybe we can have you hit the second part of that question and defer the first part to Stacy's keynote here.
Okay. That sounds good.
I'll bet it does.
I'll just.
Take the dollar.
We'll update your foil.
They're somewhat related. 450 is an interesting question, right? If you're struggling with Moore's Law, one way to deliver cost reduction to your customers or to your product line is through 450, right? 450 delivers a cost reduction about equivalent as a Moore's Law reduction. It's going to reduce your cost 1.7, 1.8x. I don't think it's going to be impacted by how fast people go. There's an interesting question about will EUV come before or after 450 and how those line up and how we line up the industry for that. I don't think Moore's Law will necessarily speed up or slow down 450. 450 will occur when we can get the consortia in New York all assembled there in the clean room. We can do the early development, people go and take that into production. I don't think it'll be impacted.
C.J., we'll come back to the first part of your question a little bit later. I think we had one waiting over here. Great.
Hi, this is Tom with question for you, Tom. You mentioned the fact that notebooks in general command a price that's very similar in emerging and developed markets. Sort of an indirect question associated with that. As you look at system-level pricing for notebooks and desktops, it seems to have leveled off at a $600, $700 price band. What is your best explanation for why that has happened, whether you think that's sustainable, and whether form factors like all-in-ones, et cetera, are a contributor to some of that in recent quarters or years?
I think part of it's the industry's fault for not innovating at the pace we collectively should be. I've been in the Sales and Marketing Group for most of my career here. I've never seen such a galvanization of the industry together to go make something happen like Ultrabook. The fact that price points have moved down and there's a lot of volume between $600 or $700 doesn't mean people won't pay for a value, okay, that's $100 or $150 or $200 more. They will. Look how many iPads have been sold for $700, $800. I think a lot of it is stalled because the PC needed a refresh. All-in-one, Kirk mentioned this morning, is a segment that's growing faster than the notebook segment, and there's some segmentation within all-in-one.
There's entry-level all-in-one, but then there's very robust all-in-ones with very interesting form factors, HD sound and everything. Consumers will pay for a great experience, and I think that is why we want to get Ultrabook into that sweet spot, but that innovation, and when we have touch coming into play and convertibles, I feel great about the opportunity to go sell Ultrabooks at $899 as much as I do about $699.
Okay. Down here.
Basically, it's James Magid. You talk about changes to all the platforms, the Ultrabooks being the most dramatic. To what degree is the Thunderbolt high-speed connection a standard part of these products?
I think we're just getting started on Thunderbolt, again, when you look at the whole platform, we got to talk about how do we go enable an experience? How do we enable an experience that we can have a sleek, thin form factor, ultra-thin, when you get back to ultra-responsive and something that really will enable a greater experience for consumers, there's a lot of industry momentum on Thunderbolt now. I think there hasn't been as much engagement as there is now and what you'll see over the next coming years.
Great. Let's go over to this corner, and then we'll move down to the middle.
Yeah. It's John Pitzer with Credit Suisse. Two quick questions. First one's for Tom. Tom, in your presentation, you talked about a CAGR for emerging markets of about 18%, and that's about 50% of the market. If I'm doing my math right, if the rest of the market's not growing, it gives you a full market growth of about a 9% CAGR, which seems significantly higher than sort of the consensus view of sort of mid-single digits or even low single-digit type unit growth for the PC market. Just wanted to get your thoughts on that dynamic.
Yeah, John, the data I showed was actual third party, but our view is very consistent with 18% going forward. We're in the stage right now, the last couple of years, where the mature market growth in consumer has been very low, and there's a number of reasons for that. You can talk about the backdrop of the economic situation in Western Europe, unemployment in the U.S. As the gentleman back there was pointing out, too, or my comment is, there hasn't been as much excitement and innovation to get consumers to pull their wallet out to go refresh their notebook, and the share of wallet has shifted to things like the iPad. The last real bump we had in mature markets for the consumer segment was Windows 7.
The Windows 7, I remember the last quarter of 2009, the first two quarters of 2010, I was way under calling my forecast because we underestimated how many consumers were embracing that new operating system and refreshing and upgrading. I think we've gone into this assumption to say the mature markets have gone to sleep and all the growth is going to happen in emerging. You saw the drivers behind emerging. It's our job in the industry to keep innovating and going beyond touch to voice and gesture and get that share of wallet back to the notebook in the mature markets as well.
Quickly for Brian, at 22 nanometers, I think you've announced three or four foundry partners or fabless companies that you're going to be doing foundry services for. There's clearly speculation as you stretch out your manufacturing lead that the foundry business model might be a way for you to exploit that lead economically. Historically, people thought of Intel as manufacturing limited SKUs but in very big volumes, and the foundry model is more about very many SKUs at somewhat less limited volumes. What kind of challenges or what investments do you have to make to exploit the foundry model, if that's going to be a route that you take going forward?
The first thing I'd break is that we're a limited number of SKUs, and large volume. We're large volume for sure. On a typical day, we run about between 6,000 and 8,000 SKUs. We are not a small SKU company. We are a large SKU company. When you take a look at the number of. You have mobile, and you have server, and you have desktop, and you have all different variations now as we've integrated in the graphics, and you can have GT1 and GT2 or GT0 and all the different bins of speed. We've been building a model more and more that's agnostic to the number of SKUs, probably over the last 10 years. We don't see
The limited foundy or even an expanded foundy model really changing that dramatically. We think we have the infrastructure to handle any increase in SKU count.
All right. In the middle here.
Hi, I'm Ambrish with BMO. Brian, question for you. Just trying to understand, you very eloquently pointed out three-year lead, now four years. In the same timeframe, the fabless companies, some of them have not done well, but there's quite a few fabless companies that have done remarkably well. What is it that underlying the assumption that when the lead gets to four years, is there an assumption that the other camp is standing still and there's no architecture-level innovations going on? Because Arm will come back and say, "Yeah, we have things like big.LITTLE on the mobile side and also in other applications." Just wanted your perspective on just trying to understand what is so different with a four-year lead as opposed to three over the last 15-odd years. Thanks.
By the way, it was Mark who so eloquently said it, so I should give him the credit of that. Additionally, Mark showed you that curve where the performance of the transistors with the 3D transistor is really targeted as they went through the development cycle. They saw that they could really shift the lower end of that curve, the low power curve. I think that's why I kind of made that comment about this is physics, right? You can't avoid this. You can't design around that necessarily. You can't do anything around that. It's a law of that transistor. So that 37% decrease, as we move forward and go beyond 22 nanometers, we've learned how to do those kinds of improvements and target the low end and the performance spectrum, and I think you'll continue to see that from us.
I think that's what's going to continue to differentiate us as we move forward. You're going to see the same kinds of improvements from Intel, but even more capability at the lower ends as we move forward. I think you're not going to be able to design around that after some point, and I think that's the point we're all trying to make.
All right. Got time for two more, I think. Got one back here.
Yeah. Brian, Dan Hutcheson with VLSI Research. I think you sold yourself short on Tri-Gate because you didn't talk about the benefit of the fully depleted versus if you tried to do planar and the impact on leakage, that sort of thing. I was wondering if you could talk about that a bit.
Yeah, we were trying to avoid getting into too much of a depth or detail.
Brian, if you could explain what fully depleted operation is before you jump in, just to baseline the audience understanding.
Now we're going to get
Yeah.
Yeah. Dan, I think this is a bigger conversation. Maybe you and I could take this as a one-on-one afterwards. We've done this a little bit, but the point that Dan's trying to make is that the solutions that others are going towards, and the advantage we have on leakage with this kind of device by pulling it up out of the silicon is quite large. We showed you the low power performance, but if you looked at something like battery life and truly your leakage rate when power is off, it's quite a bit lower on this device than a device in the silicon. That's just because you're able to pull this out and pinch it off from all three sides. That's basically the point that Dan's making.
How would you close the faucet?
Yeah. This is Dadi. I kind of make that comment, too. It is how well you can close the faucet, right? With this, you can come and pinch it from all three sides, and that is a very powerful capability.
Great, we will take one more. I think we had one down here. Trey?
Hi. Thank you, Shawn again. Hey, Tom, I was wondering if you could share your perspective on pricing. That was one of the things that really went well for Intel last year, for the microprocessors. I am mostly interested in what ended up going so well. Were you able to affect a pricing change in your stack? Was it just the mix of customers went up in all the segments? It seemed broad-based beyond just what was going on in the server area. Then, for the Software and Services Group, I was wondering if you could expand a bit on what the mix is between client and server and some of those segments that you talked about. Also your gross margin outlook for that business unit going forward from this level. Is it going to be flat, down, up, or what are the moving parts there? Thanks.
On the pricing, Shawn, it had more to do with, if you take consumers, look how sophisticated the usage models are. It used to be that you use your PC just for email or internet access. You look at video creation, social networking, people are using their PCs, their devices for things that are much more sophisticated. Therefore, when they go in the retail store, they are interested in buying something that they are going to get a better experience out of. When we went with the Core lineup, we put all our marketing behind Core, nothing on Pentium and Celeron, you go look at all the usage models, they are really geared around why a consumer would want a Core i5. There is a turbo feature in Core i5 that really brings to life a lot of the capabilities that are in primary usage models today.
Simply put, consumers were buying a richer mix. We didn't raise our prices. Consumers were buying up. You go look in the Data Center business, it behooved the companies, the cloud data centers, or the enterprises to buy a richer mix because they could spend more on the capital and save it on the back end in the OpEx. It's whether it was the business user or the consumer, they were buying a richer mix, and I think the case with Romley will be much like it was with Nehalem. I think it's a common case where the technology itself is selling itself, and we're positioning it for how the advantage works for business or consumer.
Great. Thanks, guys. I think we're out of time here. Our next stop is actually a brief break. It's again, out in the lobby. We'll have beverages and snacks outside. The executive team will be out there available for Q&A and conversation, and we'll rejoin here in about 15, 20 minutes with the final keynote of the day, Stacy Smith, and we'll have one more shot at Q&A here in the auditorium. Thanks, everyone.
Ladies and gentlemen.
Good afternoon, everybody. How do we turn these things off. Will I break something. It's up there. Hey, look at that. Work on the off button. I am a technologist. Good afternoon, everybody. This is now my fifth investor meeting to get to present to you as CFO. You may look back and reflect that I always try to start with a joke or maybe a little bit of an edgy story, because I do realize I'm the only thing that sits between you and cocktails. This year, Paul and I were talking about because of the important stuff I have to go through, evidently 30 or 40 topics of important stuff I have to go through, but particularly capital intensity, we thought maybe this wasn't the year to do a joke.
Instead, what I thought I'd do is I'd just take a couple of minutes exaggerating my credentials. I am a PhD in physics. I want you to know that. I don't know where Isaac is, but I would like you to refer to me as doctor when we're in the thing later. That PhD in physics is going to come in really handy when I start talking to you about FinFET and transistor density. I can't even say it, and things like that. Anyway. All right. Moving right along. It was interesting as I just sat here and watched the cadence of the day and tried to absorb it with fresh eyes. What was interesting to me is, last year, we stood in front of you, and we talked a lot about the plans that we had in different markets.
We talked about our plans in Ultrabooks. We talked about our plans in phones, our plans in tablets. We talked about the growth that we expected in the data center market. We talked about the fact that we did believe that emerging markets were going to be a significant driver. As I watched the Intel executives come up and speak today, what I was struck by is, and it really started with Paul, that the strategy is the same. We're not telling you a lot of new strategies. What we're showing you today is the progress that we've made to the things that we outlined last year. I was really struck by that, and I think that's actually the right approach for us. I think our strategies are the right strategies. I think we are making progress.
The one thing that was really new that has played out over the course of the last 12 months is this widening gap that we seem to have in terms of the value of our process technology, the value of our manufacturing advantage relative to others in the industry, and that's becoming really what looks to be a sustained and differentiated advantage. I'm going to spend a lot of my presentation talking about that, what are the implications of that, both from the standpoint of our financials, the products, the revenue, what we expect from a margin standpoint, but also hit very directly the capital intensity question that I know is lurking out there. Okay, what you're going to hear from me over the next 45 minutes is really what I just said.
I'm going to take the presentations that you've heard so far and try to translate them into what we're seeing in our financials and what we expect going forward. In particular, you're going to hear from me very directly that our capital cost per wafer is going up, but it's not impacting our gross margins negatively, and I'm going to take you through the math of why that is. The hints, you've kind of seen the hints over the course of the day. The hint there is that it's because of that lead that we have in manufacturing that translates both as a cost advantage and also extending our product lead across the different segments in which we play. I'll close this section, and it'll be the briefest part of the section, but it doesn't mean it's the least important part of the section.
It's just a little less meaty because there's not a lot of change there. I'm going to close today talking about our cash generation and then our priorities of what we do with that cash. Again, Paul really set the stage nicely on that. Let's jump right in with a discussion about our lead in process technology and manufacturing, what that means to capital intensity, and the implications on gross margin. Recognize that guy? It's Bob, again. He's the evil twin to Gordon. I actually am going to start this with a Moore's Law story.
A number of years back when I was running Europe, Middle East, and Africa, like Tom was talking about, they take the executives out into the emerging markets, and we get to do things to try to further Intel's brand, and I was given the opportunity to go onto CNN Türk. What they didn't tell me in advance was that it was a live call-in show. I'm on CNN Türk. They put the thing in my ear where the lady was going to translate the questions that were coming in in Turkish to English. The first question comes in. This ear is Turkish, this ear is Turkish. I'm not totally fluent in Turkish. We get to the end of the question, and I'm on live TV, everybody's looking at me expectantly.
I have no idea what the question was that was asked, I just talk about Moore's Law. I talk about the doubling of the transistors every 18 to 24 months, what it means in terms of lower cost, the ability to do lower power, and more importantly, the ability to continue to further performance. I get to the end, everybody looks entirely satisfied with the answer, and we go on to the next thing. You don't know what that person said. That's also true. That's also true, which is always the case when you're being translated, which we know from some of our customer meetings. My lesson out of that is that Moore's Law is always that safe place to go. As I'm talking about capital intensity, I'm starting with Moore's Law once again, because it's the safe place to go.
I am showing you something a little bit different on this slide, though, which isn't just the benefit of Moore's Law, it's some of the math behind Moore's Law. This is going to be important as we go forward. Again, as we go from process node to process node, we double the number of transistors. We've talked about the performance benefit we get, the cost benefit we get. There's some math behind that cost benefit that's important. Historically, as we go node to node, we see an increase in the capital cost per wafer, the capital cost per square inch of silicon, right? I'll use those terms synonymously, right? It goes up as you go from generation to generation. You have more mask layers, you might have multi-patterning, you have more expensive equipment, that cost per wafer goes up in terms of capital cost per wafer.
We get the scaling benefit, right? We get to pack on a whole bunch more transistors into a square inch of silicon, that scaling benefit more than offsets the capital cost increase as we go from node to node. We get that more performance, we get that lower power, because of that scaling benefit, we get the opportunity to put in more features and lower the cost. Those last two things are really the kind of driving force of Moore's Law for us over the years. I'm going to tell you why that continues to be true as we go forward. I'm just going to start, I want to go back to Andy's presentation from last year. He spelled out really the four operating guidelines and how we look at the investments that we make in our factories.
First, we get a huge ROI just in reducing those costs, right? Cost really for us is where we get the primary return of advancing Moore's Law and the increased product competitiveness as a bonus. I'm going to talk about that can be changing a little bit, where maybe we're getting paid in multiple ways, but that's been the historical part of the ROI in advancing Moore's Law. Second, because of that improvement in cost, performance, energy efficiency, we want to go as fast as we can, right? That two years isn't because we chose the two years. The two-year cadence that we have is because that's how fast we think the rest of the industry can keep up with us, in particular, the equipment manufacturers. We'd go faster if we could.
We'd get even an enhanced return on investment, two years is kind of the fastest we think we can go. Third is that we try to match our capacity with our unit volume over the horizon, I'll show you that data again in a minute. The fourth is kind of a caveat to that third, which is that we do bias our planning process to put in a little bit of extra capacity relative to demand, we do that very explicitly because the cost of being caught short dwarfs the cost of having a little bit of capacity available to respond to upsides. I think you're kind of seeing that play out right now in the industry. When you get caught short, it's devastating. You lose share, you lose the opportunity to win designs.
We always try to bias our process by having a little more capacity than we need. Those are still the guidelines of how we look at our capital investments, I do want to make two additional investments. You heard from Brian and from Paul that it's getting harder and more expensive to go from generation to generation, very importantly, there's fewer companies that are investing in trying to make that transition, right? That's the implication of that chart that Brian showed, that showed how the cost of a fab now is exceeding the revenue of many of the companies. More and more people are going to foundries. The implication of that is, if you go back 10 or 15 years, you had five, six, eight companies all trying to advance Moore's Law. Today, you only have a few.
That's resulting in an increase in our leadership, as our lead is extending, we're getting paid for that lead in ways other than just cost, and I'll show that in a bit. Then I want to spend some time on an addendum to Andy's fourth guideline, which is around protecting upsides, and talk about the other side of it. We do something very important in terms of how we design our process technology. Mark and the team build in something called forward reuse of equipment. When he's off looking at 10 nanometer, he's looking at how can he make sure that at 10 nanometer, the majority of the equipment that he buys the generation before that can be reused or rolled forward from one generation to the next.
That reuse of equipment protects a very important ability for us, which is the ability to quickly upgrade factories if we ever get a little bit ahead of the demand curve. That reuse, in essence, protects us against downsides in demand, and I'm going to show a case study of how that actually works in practice as I go through this presentation. Okay, enough of talking about what I'm going to talk about. Now let's just jump into the data. This is starting with units, and so what you're going to see here over the next three slides, I'm going to talk about the three things that drive our capital investment. The first is our unit growth, the second is feature integration, the third is complexity or capital intensity.
You can see on this chart that our units have grown nicely as we went from 65 nanometer to 45 to 32 nanometer, and that we're expecting unit growth again at 22 nm. You'll all get out your rulers and look, and what you're going to see is our expectations at 22 nanometer actually are pretty consistent with what we saw from 45 to 32. We're not banking on this big uptick in demand. It's just kind of a nice continuation of the unit growth that we've been spending. The important point here that you need to take away is, at a first approximation, this is what is driving our capital investment. Paul talked about it, that said, look, you have to think about the capital spending relative to the size of the business. It's unit growth, at a first approximation, that drives capital spending.
That was true at 45, it was true at 32, with one caveat that I'll talk about in a second, at 22, it is by far the largest driver of the CapEx that you're seeing us spend right now. The second CapEx driver is the integration of features where we take something, and we integrate it to that leading edge. We integrate it, in essence, onto the CPU. A great example of that is when we moved the graphics transistors to the leading edge, which for us started at 32 nanometer. You can see on the left-hand side of this chart how we've systematically incorporated graphics to the leading edge. At the 32 nanometer node, this was a significant driver of our capital spending.
In fact, you may recall on the call we were talking about as we were starting to raise CapEx, that at 32 nanometer, it was driving about as much of our incremental capital spending as unit growth was. We were taking the graphics transistors from N minus one and moving them to that leading edge. That's, in essence, done. Graphics won't be a big driver of incremental capital going forward because, in essence, for the core business, we've moved graphics entirely to the leading edge. You will see times in the future where we're going to find things where we say we want to integrate that to the leading edge. The reason we integrate is because we get lower cost, we get more performance, and we get paid for that value that we're bringing to the marketplace. It's not a driver at 22.
My guess is as we look at things that we want to integrate into SoCs out in the future, that could be a driver of capital going forward. Just one more thing, coming back to the integration of graphics. If you look at the right-hand side of this chart, and Tom talked about this, and you heard Paul talk about it as well, you can see that the improvement in our mix correlated almost exactly with the integration of graphics. I think there was a lot of things that drove that ASP benefit, but I think a lot of it was we integrated graphics. We got a ton more graphics performance in our product line, and we got paid for that. In many cases, we got paid for it because of the use model.
I think in some cases, we got paid for it because we could eliminate the need for a discrete graphics card, and so there was a bigger bill of materials out there for us to participate in. This is an important point. The third driver of capital, and these are kind of in order of how they're hitting us, is increase in complexity. We showed you this chart last year. It shows sq ft of wafer for different processes, which is a pretty good proxy for capital intensity. This is updated from what you saw last year, but it, in essence, shows the same trend. The question that Brian punted to me, when he was up of, are we expect to see an increase in capital intensity at 22 nanometer?
The answer is yes, we expect to see an increase in capital intensity as measured by capital dollars per wafer at 22 nanometer, and we expect it to go up a little bit more at 14 nanometer. The driver here is multi-patterning and lithography. That's what's caused the kink in this curve. We have more equipment, which takes up more space, that's why this is a good proxy, per wafer that we process through a factory. What's interesting here is that since we showed this data last year, you've heard from both foundries and their customers that they're seeing a similar trend. What's interesting is if you believe what at least we're seeing in the press and the disclosures of companies, they're seeing it at a much earlier node than we are. We don't really start seeing it till 22.
They appear to be seeing it at an earlier process node, they're seeing it at a much greater cost than what we're seeing, in terms of that increase in the capital intensity. There's one other distinction here that's important. Paul showed you how our transistor cost continues to come down at the historical trend. That means that we're finding incremental scaling benefits that are offsetting that increase in capital intensity. That's really important, I think that may be different from what others are seeing, I am going to come back to that point very specifically in a minute here. The combination of unit growth, functional integration, complexity, are driving an increase in capital. This chart shows capital spending from 2008 to 2012. You can see it's up. The primary driver there is unit growth.
It shows a breakout that we haven't given in the past, although I've hinted at it in the earnings call, which is the amount of our CapEx that is going into facilities versus equipment. You can see that in 2012, it's 40% of our CapEx spending is going in on space. The driver of that is still unit growth. Don't get confused by that space. We're building the space because of the unit growth that we've seen, that we expect at 22 nanometer, there's an implication to this chart that's pretty important, that implication is going to become clearer when we start talking about what's going on with depreciation, because you have to remember that the useful life of a factory is much longer than the useful life of a piece of equipment.
I think Brian gave you a really good hint to that when he did his presentation. He talked about the fact that when Mark's helping to design these big factories and we're building these big factories, we're building them to go out multiple generations. In fact, we're building factories today that are 450 millimeter capable. I'm now moving into the benefit that we get for advancing Moore's Law. Remember, as we go from node to node, we shrink the transistors. That allows us to do two things, add features, lower cost. That has been the driving force of Moore's Law over the year, and that brings our costs down over time. This chart just simply shows that cost benefit. It is our average die size.
For this chart, we take out Atom, you're not getting the skewing of what's happening to mix. You can look at the Core family in its entirety. It shows that we're getting the historical return on investment for advancing Moore's Law through 2011, that's an actual, that in 2012 and 2013, we're expecting the average die size to come down, again in 2012 and again in 2013, at that historical rate. A tidbit on this, I'm going to talk more when I talk about PCCG about the segmentation of our costs. The implications of this curve are if you're just looking on average across our Core business, I do expect to have an average lower unit cost next year than I do this year. That's the benefit of that declining die cost is it leads to lower costs.
I expect that to be true by segment, which I'll show you. I also expect that to be true for the average. Here's where we're getting into the other benefit of having this technology leadership. Again, it comes down to that benefit of Moore's Law, where we get to dedicate more transistors. I think Dadi and the team, under Paul's leadership, made a very smart decision a few years back of moving the graphics to the leading edge and getting a significant increase in graphics performance. By doing so, and just to put that in perspective, from 45 nanometer to 22 nanometer, we improved our graphics performance by about 20x over two generations. It was 10x at one and another 2x at this one. Just a stunning increase in graphics. You can see what that's meant to us in terms of the ASP.
Again, here, it's a simple concept. As our lead over the industry extends, we can do things in our products that become hard for others to match, and we're finding ways other than just cost to make sure that we're getting paid for that technology. In this case, we're getting paid in terms of mix. We haven't raised our price points, but we're selling a much richer mix than we used to sell because the competitiveness of those products is so high. This shift in the curve, by the way, is worth billions of dollars to us. It's a non-trivial economic impact. We continue to get that traditional cost benefit from advancing Moore's Law. We're also getting paid via a richer mix in our different businesses.
I'm now going to spend a couple of pages to just put our capital spending in context relative to our business levels. This is a chart that we've used with you in the past. This shows the capacity that we've put in place, the loadings of that capacity. This goes through a forecast for 2012, so that's new information for you. The first point is you can see that we continue to put in capacity, content with that long-term trend. Again, that's consistent with what we see in terms of unit growth. The second thing that you see here, it's consistent with what Andy showed you last year, that we bias our planning to have the ability to respond to upsides.
I'm not going to go through it again, but the math that Andy showed you last year still is absolutely valid, and that's all on the website. He went through very specifically the math of why it's overwhelmingly better for us economically to have the ability to respond to some upsides versus cutting things so tight and having factories that instead of running 85%-90% full are running 100% full. The third thing that you should take from this is that we're running in that sweet spot of utilization. I've used that term before on the call. When we're in that 85% to low 90s, for me, that's the sweet spot of utilizations. It says that the factories are running full enough that we're getting a great cost, but when Tom comes in with upside orders, we have the ability to respond.
When we start finding that we have progress in phones and tablets, again, we have the ability to respond. We don't constrain the market. The next way to put our capital spending into perspective is to look at capital spending relative to our business levels. These are done on the same scale, so you can see. You can see here, and Paul talked about this, just the growth in our business and the growth in CapEx. I also know some of you well enough to know that as soon as we're done here, you'll pull this slide, and you'll do the math yourself to figure out capital dollars over revenue. I'll just make it easy. What you see in 2012 is consistent with our longer-term trends, although it puts us at the high end of a 10-year trend.
We are running capital dollars relative to revenue levels that are a bit higher. By the way, that would be absolutely consistent with what I just showed you of capital intensity going up. When I think about 2013, I think it's also going to be at the high end of that historical range. I also know that this is the page where many of you are going to look for hints about next year's capital spending, and I know that from the conversations we've been having out in the hallway, that's what you all want. I'm going to eliminate the hint here, and I'm going to share some preliminary thoughts on what we're thinking for next year. This can change if our view on units changes or something else, but this is where we are today.
If you assume unit growth that's consistent with the past few years of unit growth, not a big uptick in units, just consistent unit growth, you bake in some increase in terms of capital dollars per wafer that I showed you. You offset that by the fact that we're getting a declining die size. That's that benefit of scaling. I'd expect capital spending next year to be at or slightly higher than what we're spending this year. I'm going to show you a couple of slides here of why I can do that and still generate really strong gross margins. Okay, I'm going to bring this section together with a couple of slides.
The first one, again, we've shown you this in the past, I've updated it now for 2013, it looks at depreciation as a % of revenue and depreciation as a % of cost of sales. What you see here is that depreciation as a % of both is on this nice downward trend. If you look at it from the start of the decade to today, it's been down and to the right. I've added our thoughts here on 2013, you can see that, again, for both of those metrics for 2013, depreciation as a % of cost, depreciation as a % of sales stays in that nice, healthy range. That should really help in terms of how is the capital intensity kind of playing through and impacting our gross margin.
The other way to look at this is just in terms of looking at kind of overall gross margins. What you can see here is that the impact of declining unit costs, even in a time where we're spending a bit more in capital, this great product portfolio where we're getting paid for things a little different than we have historically, have shifted our gross margins up pretty significantly since 2009. Later in the presentation, I'm going to give you some very direct thoughts on what I think gross margin looks like in 2013 based on what we're seeing today. I'll get there in a minute. What you should take away from this is that we are a capital-intensive business. We invest in capital. We get huge value out of it, and we can do it and maintain a very good business model.
Some would say an extraordinary business model. Before we leave this section, I think there's two other questions that I want to address head-on. The first is one that many of you were asking me in the halls, it's not a surprise, and it's just kind of help me with the math, and you were asking Brian the same question. Help me with this math of how capital dollars per wafer aren't resulting in an increase or a decrease in gross margin, an increase in costs and a decrease in gross margins. I'm going to take you back to the cost equation that we talked about at the beginning of this presentation, and I'm actually going to take you back to Moore's Law, and now I'm talking about Gordon Moore, not Bob Moore. It's important that we don't talk about his evil twin here.
Remember, historically, we see capital dollars per wafer going up. That's more than offset by the fact that we get the scaling benefit, i.e., we get to shrink the transistors. What we've shown you over the course of the day is that we are seeing a shift up in that historical trend. It's more capital dollars per wafer than what we've historically seen, right? We historically see an increase. We're seeing a bit more of an increase than what we've historically seen, primarily driven by multi-patterning at lithography. That's being offset by the fact that we're also getting more scaling benefit than we've historically seen, and I won't turn it on, but you really saw the magic of that when you saw what Mark was showing you on 22 nanometer. We get more of a shrink than we've historically gotten as we've gone from generation to generation.
Some companies, by the way, call that density, you can also say that we're seeing an improvement in the density of our process as we go from generation to generation. You have these two curves shifting. You have a higher capital dollars per wafer shift in that curve, offset by the ability to do a bigger shrink, and that keeps us on the historical trend of declining cost per transistor, which is the chart that Paul showed you earlier, right? I emphasized us there. I think that this really comes down to that advantage of the IDM model.
The fact that Bill Holt and Mark Bohr can work with Dadi and his team to look at the kinds of things, the trade-offs that they're talking about, saying, "Hey, are there things we can do where we actually add in maybe a bit more capital cost per wafer, but we get an incremental scaling benefit." Dadi, your assignment is to go make sure you get the benefit of that scaling benefit in terms of the products that you're delivering to the marketplace. That is a core advantage that I think if you're not an IDM, becomes very hard for somebody to mine. I think that this is why we're seeing something that's different than the rest of the industry.
Again, we're seeing an increase in capital intensity, but we're offsetting it by having more than the historical scaling benefit as we go from 32 to 22, then this data kind of extends out to 14 and 10 nanometer. What that should say is, hey, when you look out over the next four or five years, which is the time period we're looking at here, we expect that we can continue to bring down that cost per transistor at the historical trend. I think there's another question that's out there that I know there's at least one sell-side guy that keeps worrying about, which is, what if we're wrong about unit growth, right? Okay, I get all that.
You're investing, you kind of showed me the math of how you can keep your cost per transistor going down, but you're still investing $12 billion in CapEx. What if you wake up tomorrow, and none of those units come to you, that's going to be a big problem, right? Well, okay, yeah. It'll be a problem, but it's a problem that we can deal with. I want to hit head-on what we would do there. Before I do, I just want to take you back and remind you on that utilization graph. We're continuing to add capacity consistent with what we've done historically, which is consistent with unit growth, right? We're not getting ahead of that, and we're in that sweet spot of utilization. We do know that sometimes things happen.
It's worth looking at what actions we could take if capacity gets way ahead of demand. I'm going to use the financial crisis of 2008 as an illustration of the tools that we have at our disposal to respond to changes in demand, because I think this really illustrates it, and it's a great stress test of our business model of you had something that happened in a really precipitous fashion, how did we respond to that? We start here with revenue. You can see that in the lead-up to the financial crisis, our revenue was on this nice upward trend. It kind of went from $8 billion a quarter to $10 billion a quarter. Then you had Lehman fail. Over the next two quarters, our revenue plummeted down from $10 to $7 billion. It was pretty precipitous.
If you looked at units, it would look just like this. What's interesting then is the actions that we took to respond to that change in demand, the next three slides will show that to you. I'm really focusing here kind of the two quarters before Lehman failed, and then the four quarters after. The first thing you can see here, this shows utilization, this slide shows that we very quickly stopped running product through the factories. I can tell you, this was a scary thing, those of us that lived through it, right? None of us, I think Paul included, even with his slightly longer tenure than I have, had seen us taking utilization down to 30% in the factory networks. It was a pretty tough decision to make, but it was also a pretty obvious decision for us to make.
The reason that we made that decision really was for two driving factors. One, we know that coming out of a downturn, we don't want to have a bunch of inventory sitting in place on older generation products. We really excel when we come out of a downturn with our freshest, strongest product line. We didn't want to keep factories full, hoping that this would go away at the cost of pumping a bunch of older generation products into inventory that then wouldn't be at the leading edge when we came out of it. The second is, remember that forward reuse that we get where we can use 80% to 90% of our equipment from one generation to the next? We saw this as an opportunity to take 45-nanometer factories offline, upgrade them to 32 nanometer.
We could utilize almost all of the equipment that we had invested, we could actually then get to 32 nanometer even faster. What this chart shows is the impact of that. This is our quarterly CapEx spending. You can see the dotted line there was the plan that we had. We were on a path in 2009 to spend $8 billion in CapEx. By taking the utilization down, upgrading factories that we put in place for 45 nanometer to 32 nanometer, we actually reduced the CapEx down to $5 billion in 2009. More importantly, we accelerated 32 nanometer. That then is what has propelled our results in 2010 and into 2011, was the fact that we came out of the downturn with a bunch of 32-nanometer factories, a great product line, at the same time that other people pulled back. Okay?
The key here is that we have that ability to kind of forward upgrade capacity if we find that we get ahead of demand because something happens. This page shows the impact of the downturn and our response on our gross margins. We had two quarters where our gross margins were below the historical range, but by the second half of 2009, we were actually generating historically high gross margins. Again, that was a result of the strength of our product lineup coming out of the downturn, and the fact that we could quickly have full 32-nanometer factories as opposed to working through a bunch of inventory of 45-nanometer product. This was a result of a fast response.
It was a result of, I think, great operational execution out of the factory network and the product guys, a result of what the technology development organization does in terms of building in that forward reuse capability. This is an example of a worldwide financial crisis that kind of had a very dramatic impact to demand relative to supply. If we saw something that was more subtle, like just a protracted slowdown in Europe that leads a year from now, we realized we were building in a bit more capacity than what the unit demand actually supports, we'd use these same tools to respond to that kind of a change.
In fact, it was something not obvious to you all because it was on a fairly small scale, but even in something like the hard disk drive shortage, where all of a sudden our order desk kind of got quiet for a little while because the world couldn't get hard disk drives, Brian and his team found opportunities to take some older generation capacity offline and bring it up on 22 nanometer, use that equipment at least to 22 nanometer and offset a little bit of the 22 nanometer investment. Wasn't a lot, those are the kinds of tools that we would use to respond if we ever got ahead of demand. We'd just buy less equipment for a period of time, upgrade the factories from one process to the next. I'm now closed on the capital part of this.
I'm going to move into sharing some information about our different businesses. I'm going to go pretty fast here, again, because I started this with our strategy's pretty much the same. I'm not going to give you a bunch of new three-year targets. I'm just going to tell you how we've done and share some thoughts with you about some specific businesses. Again, we have these four businesses. We reorganized our segment reporting a year ago to kind of make it more match how we were thinking about the different markets. We have our PC business, our data center business, we have the other Intel architecture business, and then underlying that, we have our software and services businesses. At a high-level flyby, the PC business is a nice little business of about $35 billion in revenue last year. It generated almost $15 billion worth of operating income.
Our server business, as we actually predicted last year, crossed over $10 billion for the first time, it's generating about a 50% operating income margin, it was more than $5 billion of operating income. Other Intel architecture, where I'm actually going to spend a bit more time now that we actually have product shipping in the marketplace that maybe is a bit more relevant to you guys. This is the business that sells into the embedded, the phone, and the tablet market. Last year, that was $5 billion of revenue and generated a loss because we were making some pretty significant investments in these businesses, and I'll show you those investments in a little while. Then our Software and Services Group, through the acquisition of McAfee, had significant growth and was almost $2 billion of revenue.
Let's start with our biggest business, which is the PC Client Group business. As you heard and as you've seen, our PC Client Group business grew 17% from 2010 to 2011. That was a combination of both billings growth and ASP. We haven't given this breakout in the past. I think there's a perception out there that it was all ASP. The answer is we saw high single-digit unit growth. We saw an uptick in ASP, and that's how you get to 17% unit growth on a or revenue growth on a business that is that big. You can really see the themes of the day playing through in terms of the results of this business. We saw nice growth in units, primarily driven from emerging markets.
We saw a nice uptick in mix driven by the strength of our technology that resulted in an increase in ASPs. If you look at the investment line of this business, we kicked up the investments quite a bit as we started into Ultrabooks, a pretty significant increase in investments. You add all that up, and what you have is a big, fast-growing, and very profitable business. Here, I do want to dive in on costs, and this has historically been one of my very favorite slides. If you don't remember it, what it shows is for each of the segments of our business, it shows a cost trend over time. It breaks out between our performance, our mainstream, our value, our Atom, and it shows how costs have trended from 2008 to 2013 forecast.
You've heard us talk a lot over the course of today about how segmentation has benefited our mix and our ASPs. But behind the scenes, it's equally important, and maybe even more important, of how we've segmented our costs over the last several years. The beauty of this cost segmentation is that we can maintain a good gross margin regardless of which of these segments of the market is growing faster. I've told you in the past, I think the hardest thing for us to forecast right now, maybe it's the growth in phones and tablets because we're a new entrant, but behind that is really the mix that we're going to sell between the different segments of the market. It's just an inherently hard thing to call. By having a segmented cost, we can actually maintain gross margin regardless of which of the different segments is growing.
If it's the high end of the market, we get paid for having multiple CPU cores, multiple graphics cores. You saw that in terms of our mix over the last few years. If it's the low end of the market, we get paid by having a much lower cost structure that allows us to maintain the gross margin. The first takeaway from this is that we've achieved a nicely segmented cost model over the last several years. The second takeaway is, if you just look at these from left to right, you see that costs are coming down in every segment over time, right? There's some small interrupts in the trend, but generally, everything is from the left down to the right. The interrupts in the trend are decisions that we take.
If you look at the Atom segment in 2012, I think I said this to you last year, we actually expected that one to be up because we moved our Atom product family from single core to dual core based on how we wanted to compete in that particular market, right? We ooched that up a little bit, and I'm going to talk more about Atom in a minute. The performance segment, what you see there is the impact of multiple factories ramping on 22 nanometer. Those are Ivy Bridge products that are primarily in the performance segment. That causes a bit of an ooch up in cost, and we talked about that at the earnings call. Generally, though, down and to the right, which is exactly what we want.
The third takeaway here is more tactical, and that's that we expect our costs to come down from 2012 to 2013 across performance, mainstream, and Atom. Three of the four segments, you're going to see it coming down. In the value segment, what you're seeing is similar to what we did with Atom last year. We're adding in some graphics cores in this case, based on the competitive requirements of that particular market. We think that that'll be more of a flattish cost, maybe up a little bit next year based on how we're responding to the competitive impact of that particular market. You can see here how this cost segmentation, the ability to put different features in different segments of the market, really gives us an advantage around how we maintain our margins that we haven't historically had.
Looking forward for the PC Client Group, there continue to be two key drivers of revenue growth in this business. I think Tom did a great job of talking about emerging market growth. You got to remember, it's half of our consumption revenue today. It's been growing at a fast CAGR, and as Tom showed you, and as we've shown you in the past, you're now seeing that affordability is getting into the range that we're likely to see an increase in penetration rates. As Tom also showed you, the desirability of the technology that we sell is very high. That really is the unit driver in this space. I think the second driver of this over time is going to be Ultrabooks, whether it's the back half of this year or next year, not smart enough to know.
The level of innovation and excitement that we see around Ultrabooks, as you've heard from several of the executives standing up, is extraordinary. One of the beautiful things about my jobs, even though I am a PhD in physics, is people get to walk in and show me these devices that are coming to market six months from now, nine months from now. Often I say, "That's nice. It looks like another notebook computer. Well done." These things, you're going, "These are devices I want." You start coupling touch with them, the performance that we're bringing with Ivy Bridge, these are devices that I want. The fact that we have all of this innovation happening on the PC platform is an extraordinary change from where we were a year ago, where really the innovation on the PC platform was around driving BOM costs out.
Now we're seeing innovation that's striving to drive excitement up, and I think that's ultimately good for us and good for the industry. Then you add to that the technology leadership, which lowers our costs, the strength of our product portfolio, and it adds up to an expectation for next year of solid revenue growth in our largest business, and maintaining a continuation of the high operating income % that we've been achieving. A business that, again, is growing fast and is highly profitable. Okay. I'm shifting to the Data Center Group, and I'm going to spend the least amount of time here. It's not because I like them less. It's really because all the things we talked about last year are still true this year, so I don't have a lot new here. We talked last year about this expectation of a 15% revenue growth.
Sure enough, we came in at 17%. We talked about all these trends driving server units. Sure enough, we saw double-digit server unit growth last year. We talked about the fact that we were seeing a richer and richer mix based on the value that we were bringing to IT managers. Sure enough, we saw that. We talked about an expectation of a 50% operating income target over time. That's exactly what we achieved. We really pretty much did exactly what we said last year, and as we go forward, I think this may be the exact same slide I used with you last year. It's really the four trends that Diane talked about. It's this build-out of the cloud that's really being driven by all these devices computing, connecting to the Internet.
It's our expansion of our footprint into things like networking, communications, storage, out in time into fabric. It's the voracious demand that we see for high-performance computing, and it's the technology leadership that we have in the enterprise. We still expect a 15% growth CAGR. I was pretty explicit, Paul and I were pretty explicit on the call. We expect that we're kind of showing double digits in Q2 and back at something like this CAGR rate as we get into the back half of this year. Our expectations in terms of operating margin as we look out, it's hard to predict that five years in the future. As I look at next year, I still think we're in that 50% operating margin range. Very consistent with what we said last year.
Now I'm turning to other IAG, where probably a bit more change from where we were last year. This is a segment that, again, includes our Embedded, our Phone, and our Tablet businesses. Here, I'm going to be fairly explicit. I actually expect that this business will grow significantly next year. At $4 billion, and by the way, just so that there's no confusion, this is a 2012 forecast at $4 billion because we know netbooks were down a lot in 2012, so I'm starting with that as the baseline. 2012 to 2013, I expect significant growth in this business. At $4 billion, by the way, we're already one of the larger players in this business, and if you look at our Embedded business, if you look at our Intel Mobile Communications business, they're already among the largest players in their respective markets.
Our prediction is that we're not just large, but we're going to significantly outgrow the market. What's going to drive that is design wins in Embedded. I'm not going to show the chart this year, but last year, I showed this chart about how design wins precede Intelligent Systems Group. How design wins precede the actual revenue. Our design win momentum in that business is kind of uninterrupted, and Paul talked about that. We're winning designs, particularly with Atom cores, across a variety of segments. Gives me a lot of confidence in robust revenue growth next year. I think we're going to start to see phones kicking in, tablets kicking in, increases in our IMC business based on some of the design wins we've won. I'd expect that this business will grow from $4 billion to $5.5 billion next year.
Again, underlying all this is that transistor process technology leadership, right? That really is the driving force here. It's playing out in Atom and Embedded. It's playing out in terms of the performance that we can bring to the phone and the tablet market. You should also understand that we're making significant investments in this business. What this chart shows is our research and development spending specifically for the other Intel Architecture Group. You actually heard a lot of the themes over the course of the day, and it comes down to, as we were making progress in various markets, we made the very conscious decision to make some bets, to make some significant investments in this space. You heard from Mike and Hermann how we made the decision, form factor reference designs to the marketplace. We invested in that.
That was a significant investment. You heard from them how we increased, actually, you heard this a year ago from Dadi, how we increased our cadence in terms of the Atom processors intercepting 32, 22, 14 nanometer process technology now on an annual cadence. As we realized the opportunity that we had to distance ourselves from everybody else in the industry, we made some significant investments in that space. We made the decision to bring some lower-cost cores to the market, and I'll show you that in a minute, and they talked about that in terms of the bifurcation of the product line. We made investments there. We made similar investments in the tablet space, right? In tablets, we're taking some of the same pages from the playbook. We're moving into the space of doing form factor reference designs.
We're investing quite a bit to be timed to market with the best product to intercept the Windows 8 launch. These are all decisions that we made that caused an increase in investment in this business. One thing you should take away from this chart is that the scale that it takes to participate in this business is actually quite significant. If you don't have the ability to offset that scale with a significant business result, you actually can't afford to continue to make these investments. I look at this, and one of the predictions I have is you're not likely to see four, five, six players going after the silicon markets in phones and tablets. The economics just don't work for that. We're investing at scale today.
We're starting to see the fruits of that investment, but I think there's not a lot of companies that are going to have the wherewithal over time to make the kinds of investments it takes to be in this market. We have one advantage that others don't have, I want to talk about that on this next slide. We get the advantage that we can make these foundational investments and capabilities that benefit the entire company, then we can use those capabilities across our different businesses. What you see on this chart is the investments that we make in what I call the foundational R&D capacity. It's the $2 billion plus that we invest in our process technology development, $2 billion in shared processor and graphics cores. It's over $1 billion in software. You saw how we support every operating system.
We make significant investments to have that level of support and to have a great software capability. Let me give you just a couple of examples of how this foundational capability benefits all of our businesses. The first, I think probably one of the most advantageous to us, is that when Mark develops his process technology, when Brian builds his factories, they're building factories to run all of our products. We don't build separate factories for Atom. We don't have a separate process technology for Atom than we do for Xeon. It runs in the same factory that we run Xeon, we run Core i7. This not only gives us leverage on R&D, it also dramatically reduces our risk of entering new markets. I'll use the phone market as an example. We actually had plans that we would have designs shipping earlier than we did.
We've all heard what happened with Nokia. If you have to dedicate a factory because you've dedicated a process technology to that, your risk of being in that market goes sky high. We get R&D leverage, we also lower the risk of entering new markets. I'll give you another example. In the middle bucket there, the cores that we develop, we primarily develop them to serve a market, but then we find opportunities to use those cores in other markets. Diane hinted at this, but she didn't drive it home, I'll just build on what she said. In the micro server segment, what she did is she went in and she took a core, an Atom core that we had been developing for phones and tablets, and she said, "Wow, I can actually take this into the micro server segment.
I can put some RAS features around it. I can put 64-bit on it. I can get to market two years faster than I thought I could, I can actually have a solution in that market where I can make a lot more money, I can deliver a lot more performance than what I think a competitor can deliver." That's a very, very fundamental capability for us. We're getting leverage across these foundational R&D investments. I'll just make one other point here. This, by the way, is one of the main reasons that it wouldn't make sense for us to take an Arm license and get an Arm core and just go take Arm into our factories and make Arm perform better.
We wouldn't get a performance benefit, we wouldn't get a power benefit, and we would have to spend a lot more in R&D to make that work because we're not leveraging foundational investments across the company. By the way, when Paul talks about getting paid twice, this is one of the examples of that. We get to leverage these R&D investments that we make across the entirety of the company. The second type of synergy that we get is between the specific R&D projects that we do for our different businesses. Now I've added the top part of this chart that shows the R&D spending we do for our PC and server business and the R&D spending that we're doing for the other Intel Architecture businesses. Both of those buckets are about $2 billion.
Let me give you a couple of examples of how we transfer value across these top segments. For system on chip products that are going into the phone market, we've developed a pretty deep expertise over the years in terms of fine-grained power management. We've also invested a lot to get into very thin, power-efficient packages that enable these very small form factors. That's very appropriate for the phone market. As Dadi and his team were looking at how do we accelerate into Ultrabooks, Paul talked about how we've made this transition kind of remarkably in one year, those capabilities were extraordinarily important, and they actually accelerated our Ultrabook effort quite a bit because we had been developing these deep expertise in fine-grained power management and small form factor power efficient packaging that has great benefit now to our core business.
On the other side, in our core business, we have years of experience around performance. We do things around multithreading, we do things around out of order execution. Those capabilities now become very relevant to Mike and Hermann as they hit these power envelopes. How can they provide even more and more performance relative to our competitors? We have the ability to leverage the foundational investments that we make in R&D across multiple businesses, and we have the ability to take capabilities that we're developing for one business and lever it into the next. That's a pretty significant leverage for the business. It actually says that on a cash basis, my other IA business is generating a lot more incremental cash because I'd be making a lot of these investments anyway for that business. Just wanted to share that with you.
I'm going to move on and talk about cost for a second here. I showed you earlier how the average Atom cost trend was coming down. This is a different cut of cost data, and what this is showing is the lowest cost core that we have on various process technologies. What you see here is that at 65 nanometer, our lowest cost core was a Celeron. As we brought in the first Atom product, which was targeted at netbooks, you can see how we brought the cost down by almost half. That was the first Atom core. By the way, I think that was one of the things that we were really struggling to communicate to you of why we had such good margins on netbooks, even with that lower ASP, is because we also had a cost that was half the cost.
That cost becomes a really important segmentation benefit to us. We make nice progress as we go from 45 to 32 to keep this simple. I left that off the chart. But what I want to show here is where we get to with our lowest cost cores at 22 nanometer. The first thing you can see is that we're forecasting another significant reduction in cost. This is that other advantage of leading in manufacturing, is that we can dedicate that Moore's Law benefit to driving cost down. At 22 nanometer, in the product that's dedicated to the high-end smartphone segment, as we look at what competitors appear to be bringing into that market in that time period, and the kinds of performance levels that they're going to have to hit, we actually think that we're in a cost leadership position.
We may not be the cheapest, but we're going to be among the least expensive in terms of the cores that we're bringing to the marketplace. That becomes a pretty important benefit as we're breaking into the market. The other way that we dedicate Moore's Law to helping us from a cost standpoint is we bring down the cost of a core to broaden the size of the market that we're participating in. That's the playbook we're playing right now in embedded, by the way. As we bring Atom into embedded, we now start to expand the size or the percentage of the market that our products get to participate in. At 22 nanometer, you can see where we're targeting in terms of the low-end smartphone segment. It'll be a little less than half the cost of what we're targeting at the high-end smartphone segment at that core level.
Looking forward for the Other Intel Architecture Group, we expect that our intelligent systems business is on a path to be a $2 billion business this year based on the design win momentum that we're seeing. In fact, the designs that we've won by and large, we expect on the order of magnitude of 25% growth as we go from 2012 to 2013, and we expect that next year we're crossing over that billion-dollar mark in terms of operating income just for the embedded business. In phones and tablets, we're making significant investments, as I showed you. We're winning designs, and we're bringing our costs down.
I expect to see revenue growth next year for this segment of more than a billion and a half dollars, and we expect to cut the operating loss in half. I'm going to very briefly talk about Software & Services Group and then NAND, and then close with a couple of thoughts on dividend and cash. I'm not going to spend a lot of time here other than what Renée said. The integration has gone well of McAfee. The McAfee acquisition, coupled with our other software assets, gives us a business that will surpass $2 billion this year. And as we look forward and we look at just the cadence of that business growing, we think it's going to continue to grow in the low double digits.
More importantly, as Renée and Stuart showed you, we've gathered a unique set of software and hardware assets, which we think gives us an unrivaled capability in security. You're already seeing that in our core business. Over time, we're going to deploy that expertise into phones and tablets, which we believe gives us another value proposition to take out to our customers. The NAND business rolls into the Other Intel reporting segment. I will take a second on this one. It's a pretty good story. You can see the significant improvement in profitability over those years. It's really a result of the two-pronged strategy that Paul and I have been talking to you about now for four-plus years. First, we made the decision to shift as many of our shipments as possible into the higher-margin segments of this business, SSDs and compute NAND.
You can see the gold line there that we're, in essence, at 100% today. 100% of our shipments now are going into those segments versus the commodity segment of the market. Second, we've achieved process technology leadership in this industry. When you're competing against somebody like Samsung in this space, having that process technology leadership and cost leadership becomes critically important. To put the improvement of this business in perspective, 2008 was my first earnings call. I mean, my first investor meeting with you all. I got to stand up here and kick sand on my shoes about the billion-dollar loss. If you remember, that's when Paul stood up and said we will never let this happen again. Today, we're generating about half a billion dollars of operating income, and we have a business that's nicely cash flow positive.
A pretty amazing turnaround from this team in terms of this business result. You may have seen the recent announcements with Micron about how we've restructured the business. Simply put, all that does for us is it allows us to continue to participate in these higher-margin segments of the business, but it's taken even more risk out of this business for us as we look forward. This is a pretty important page. I'll spend a second on it. I'm now showing you my expectations for gross margin for next year. If you weren't paying attention, you are now. I expect that based on everything that we know today, we'll see a gross margin next year that continues to be in the high end of that historical range. You can see that nice growing as it went from 63% in 2011 to 64% in 2012.
Based on everything we know, we're continuing to be in that 60%-65% range next year. Some of the drivers of that you've seen over the course of this presentation. I expect that our unit costs come down a bit. I expect that we'll continue to generate healthy levels of business. What I haven't shown you, but it really hasn't changed from the trends you've seen from us over the decade, is I expect the startup costs will be higher next year. Consistent with prior trends, odd number year, which means startup costs go up. You add all that up and it keeps me in the healthy top part of that gross margin range.
By the way, when Andy saw this presentation, his comment to me was that, "Hey, that's a really good presentation, but the reality is you could just show this chart, and then we could all go get a cocktail." There's probably something to that, by the way. The other thing I want to take advantage of this page to talk to you about is something Paul hinted at, which is our spending as a % of revenue. I just want to reiterate, we're not giving guidance for 2013, but I do expect that we'll bring spending as a % of revenue down some in 2013. I think that we're investing at scale right now in that other IA business.
We'll make some increased investments, but I think we'll bring spending as a % of revenue down some, and over time, we're going to get back to that model that we've been talking to you about. The last section, and then it's Q&A and cocktails. Not going to spend a lot of time on this. You've seen it from Paul. It really talks about the growth that we've seen over the last several years. 2011 was a great year. 2010 was a great year. We're growing at a mid-20s CAGR over those two years. We put out expectations for this year of high single-digit growth, which will deliver another year of record revenue and record profits. What I'm not going to torture you with, but it's one of the things I like looking at, is just our growth relative to the people that we benchmark against, right?
It's the large companies, it's a lot of the tech companies. Against any of those indicators, our growth over the last three, five, even 10 years now is outgrowing those indicators, and that really is a testament to the execution across the 100,000 Intel employees and the management team that you've been talking to. Since we're a capital-intensive business, we look at return on invested capital as kind of our key metric of return that we're generating. You can see on this chart, there were three years where we're generating well over 20% ROIC. By the way, the graph shows 2012 based on consensus expectations, don't bother trying to do the back calculation and getting to what I'm expecting in terms of earnings.
I'll tell you based on what consensus is, I think we can probably do a little bit better than that, but we'll figure that out by the end of the year. Our goal is to be in the top 20% of the S&P 500, and you can see we're well above that goal on this. All right, moving to cash generation. Our business has generated over $130 billion of cash over the last 10 years. 2011 was a record at $21 billion worth of cash from operations, and 2012 was on track to be another record. Extraordinarily strong cash generation in our business. This takes the same cash from operations, and it compares it back to CapEx and the dividend. You can see that we've generated significantly more cash than we need for the investments in CapEx in every part of the business cycle.
I was actually having this conversation with some of you at lunch. I promised I'd come back to this. Let me just use 2009 and 2011 as ways to illustrate this. In that recession of 2009, we generated almost $7 billion of free cash flow, so cash from operations, less what we needed in CapEx. Yes, cash flow from operations was down, but as you saw earlier in the presentation, we actually reduced by $3 billion the amount of CapEx that we spent in 2009. You have this nice self-correcting in down parts of the cycle, which I think is a not well understood part of our business model. 2011, which was a very strong year, we generated over $10 billion of free cash flow. Cash flow from operations was over $20 billion, and we invested $10 billion in CapEx.
You can see the other side of that cycle where we're making some significant investments in a strong year. After investing in our business, our first priority, I think as you all well know, is in the dividend as a way to return cash to shareholders and generate shareholder return. This just shows that dividend increase over time. We're not done yet. This week, we announced another 7% increase in the dividend. When that goes into effect in the third quarter, we'll be paying out $0.90 per share. As a significant shareholder, it's something that I like to see. I love to get those checks. As we think about this going forward, and as I told you last year, we're dedicating a larger percent of our free cash flow over time to the dividend.
We took that up from about a third to about 40%. You can see that in the actions that we've taken over the last 18 months. Share repurchases are the other way that we return cash to shareholders. The way we think about this is we figure out a cash target that we want to be at. It's not an exact number, it's often a range. We say we want to have this much cash. We use the buyback as the way to modulate that cash balance, which also returns cash to our shareholders. You can see that in 2011, we generated a lot of free cash flow. We then invested in CapEx and paid out a healthy dividend. We still had a lot left over to do share repurchases.
We augmented that by bringing our cash balances down some. We augmented that even more by going out and borrowing some money. It was one of those rare opportunities and time where the weighted average cost of that debt was actually about half of what the dividend yield was on the day of the debt issuance. It gave us a nice opportunity to accelerate the buyback a bit last year. You can see the result of that is last year, we bought back over 600 million shares for about $14 billion. While we don't try to time the market, the timing of this actually was quite fortuitous. The average buyback price of those shares was about $22. It ended up being, I think, quite good timing.
To summarize this, my key messages are that our process, technology, and manufacturing leadership is a more valuable competitive advantage than ever before. Our investments are delivering these leadership products. We're getting paid in terms of lower costs. We're also getting paid in terms of just the competitiveness of those products that are allowing us to break into new markets, get paid differentially in some of our existing markets. We're growing, we're highly profitable, and we intend to maintain both of those and continue to generate cash to invest in our business and to return to our shareholders. With that, thank you very much. With that, thank you very much. I think we have Paul come up.
Yeah. We'll invite Paul up, we'll do a short Q&A session, that's all that's standing between all of you and a cocktail reception.
We restrict all questions to EUV, though.
Very good. Stacy did mention to me in passing that he thought it would be a crying shame if he didn't get a few questions that exercised his newly minted PhD in-
PhD, exactly.
FinFET.
Yes, FinFET.
FinFET. Head over. Great. Sit here?
Thanks, guys. I'll throw one out at Stacy, and then one at Paul. Stacy, at the risk of sounding like our dearly departed comrade, Tim Luke. I scritch up my hand.
A multi-part question.
One question in six parts, please.
Yeah.
Your gross margins this year are at the high end of the high end of the range. Can your gross margins next year exceed the gross margins this year? Could you talk about sort of the moving parts or what that depends on?
I just gave you a whole lot for 2013. I don't think I'm taking it any further than I expect it to continue to be in the high end of the range. I'm going to stop right there.
Strike one. All right, let's try Paul.
Paul may tell you about our gross margins next year.
Hope I don't strike out the side here. Paul, you guys talk about your manufacturing advantages, and I asked you a little about this on the call, but I'd like to expand on it. People have sort of pussyfooted around the foundry question. Are you guys going after the foundry business now more than you ever have before? Is this a possibility? And would you consider being a foundry for Arm processors?
As I said on the call, we are slowly moving into taking some foundry business for a variety of reasons. I describe it as a crawl, walk, run strategy. We're in crawl stage. None of the CapEx investments that Stacy talked about today for 2012 or 2013 include us moving into walk or run. If we decide to do that, or when and if we decide to do that, you'll see a change in our CapEx because that would have to happen ahead of any foundry announcements, point one. Just in terms of telegraphing rate of change, that gives you a good indication. Certainly, through the 22-nanometer generation and possibly into 14, I think that every wafer factory we can build, we can certainly fill with Intel architecture devices. That's our current view of the world.
As long as I can get paid twice the foundry margin and the architecture margin for that wafer, that's a better deal than just getting paid the foundry margin. That gives you some perspective. I do think there's a point in time when we could be selectively taking larger amounts of foundry business. I don't see us ever becoming a general purpose foundry, a la, say, TSMC or GlobalFoundries. I could see us being a selective foundry for strategic reasons or strategic customers. In that equation, the relationship with the customer would dictate what we decide to build or not build, to duck your ARM question. Who knows what it's going to be, and who knows who the customer's going to be.
At some point in time, I do believe that our ability to build factories that exceed our intrinsic demand and get paid for it on a value basis versus cost basis would certainly allow us to be much more aggressive in the foundry sector.
All right. We have one right here. Stacy.
I got Stacy Rasgon, Sanford C. Bernstein. I had a question going back a little earlier to one of the other presentations, looking at PC growth coming from emerging markets. You listed, I think it was one-third of the TAM was new purchases in emerging markets, and I found that surprising. That was actually lower than I would've thought. I was wondering how you rationalize that smaller portion of that TAM with your statement that you see an inflection point in PC growth, or at least in PC penetration in emerging markets coming from increasing affordability. How do those two things rationalize? I don't see how you see that same kind of inflection with the economics at the lower end of the emerging markets that drives a relatively low amount of that total TAM coming from new purchases versus replacements.
This is Tom. I had the slide. The one-third was today. If you look at the AB class and the amount of the install base that exists today in these Tier 1 cities and emerging markets, it drives a tremendous amount of refresh. As I said, we've been in China for 25 years. There's a huge amount of that is refresh. Increasingly, density is playing into it as well. Today, it's a third. The function of how fast markets like India move into that three to four weeks could drive that other factor up. Right now, it's reflective of the current stage.
What do you see, I guess within that?
Go ahead, Stacy. We'll repeat the question.
I can shout it. What do you see in terms of those first purchases, in terms of their outlook for replacement cycle, given that they're spending most of their attention?
What do you see in terms of replacement time frame for those first purchasers, given that they seem to be spending a good amount of their annual or of their income right now to stretch to buy higher-end PCs? What do you see in terms of both the replacement and the densification trends in the guys who already own PCs there? Do you see any changes or differences in the emerging markets versus more mature markets? Are they more willing to buy cheaper computers as a second device? Are they more willing to buy up as they replace or less willing to buy up as they replace-
You're talking about behavior in emerging markets.
Yes, in emerging markets versus mature, in terms of the replacement cycle of the first-time buyers and the replacement habits of the ones that they already own.
I think what happens in emerging markets, it's not so much a replacement. The asset lives on longer than three to four to five years. It stays in play, and there's the density where there's an additional purchase. That's a bit of a different pattern, whereas in the mature markets, the asset might get retired in favor of something new.
Got it. Because they're, I guess, adding or they're not replacing, does that lead to lower growth? Are they buying lower-end units as they add?
Stacy, I think you're maybe catching him offline, the other thing, to me, more importantly, isn't the refresh rate, it's the multiple PCs per household.
Right.
He didn't mention that, but that's what's happened in every country in the world, is that as the household income grows up, they add that second and third computer, just like we've done in the mature markets, and that far outweighs any kind of replacement rate discussion.
Let's go up here.
Yeah. JoAnne Feeney with Longbow Research. You've talked a lot about the advantages your manufacturing brings and how far ahead you are of the foundries. You've also talked about getting double margins because you do design and manufacturing. The question is this: What do you see for the pricing environment on the design side over the next few years, given the potential indirect competition from ARM through tablets or from your other x86 competitor? How much could you give up in margin on design and make it up on the manufacturing side, again, over the next few years? What kind of latitude do you have there?
Well, in PC space, I think that we're going to continue to get paid for the value associated with the install base, the legacy software, in addition to having substantial performance and price performance and power performance headroom. I don't see that equation changing in PC space, even with the incursion of lower-priced non-compatible, if I could say that, devices. We've seen that for years because Intel and AMD both have substantially lower-priced devices that we don't sell lots of, not by choice, but by demand. The customers are voting with their wallets to buy the $100 processors versus the $25 processor. Having a $25 Arm chip, I don't think changes that equation except that those chips are going to be harder to sell because of their inaccessibility to the legacy install base and compatibility of drivers.
In device space, where we are not the incumbent, as I said earlier, I don't expect us to change the value equation for pricing anytime soon in phones or tablets. Depending on the segment, smartphone chips or apps processors are $25-$30. Tablets, you probably can get $5 to $10 more per device. We've architected our product die sizes to be able to be highly profitable, higher performance, and price competitive in those segments. That is to say, I don't think there's an x86 value add associated with, let me say, the Android marketplace today. There may be over time as we are able to exploit feature sets in our silicon by virtue of us optimizing things like the Android operating system itself.
Do you see a stability in the ratio of the profits you derive from the designs versus manufacturing over the next few years, or do you see that shifting?
I don't look at it that way, nor would I comment if I did. To me, that's the beauty of the IDM model, is I don't care. I know that we can optimize for some segments for one versus the other, and we, in fact, do.
Let's go to right here. Vivek?
Yes. Hello. Vivek Arya, Bank of America. I wanted to actually ask that average selling price question from the other side, that the increase in ASPs over the last few years have been an important part of your growth. Now that your core mix is already at a very healthy 70% plus level, how do you think about ASP progression from here? I think when we have asked this question in the past, you mentioned that ASP should be benign. Given the amount of CapEx that you're putting in and you're saying you expect to be paid for it, but then you're saying ASPs will be benign, how do we sort of square those two things?
Well, first on the getting paid, the point I tried to make is we continue to get paid in terms of the traditional cost benefit, right? Again, that cost per transistor comes down, so I believe we're going to continue to get the historical benefit of progressing Moore's Law just based on the cost. My opinion on the market, and Paul can add his, I think Ultrabooks actually have an interesting opportunity to generate innovation and excitement that start to grow the TAM at the higher end of the market. Now, I'm not predicting that happens in the next six or nine months, but over time, that level of innovation, if it drives excitement, if it drives sales, it could drive a richer mix for us. It's not in my baseline projections, but I think it could happen.
Got it.
Do you want to add anything?
Well, the other thing I'd add is that we haven't really talked about this explicitly today, but it was implicit in Renée's conversation, is that for every segment that we're participating in, we are building software value add capabilities that will generate essentially an annuity income stream around our products. At the low end of the market, for example, phones, that annuity stream could be a much higher multiplier effect than it is at the higher end of the market, say, Xeon. To some extent, that model, to be able to have a silicon-based service capability in multiple dimensions, is what gives us much more margin insulation at the low end of the market and allows us, I think, a little bit more free rein to go after some of these segments.
All right. Just a quick follow-up. On Ultrabooks, I just want to challenge. I understand these are very innovative products, but what has been the sell-through data so far? Because people do say these are more expensive products. I understand you have a lot of design wins. These are early days, but what has been the sell-through data so far? What has that shown? How does the expectation that these would be 40% of the consumer mix then align with the view that it's really emerging markets that are really driving most of the PC growth, because these, I would imagine, are very expensive for emerging market consumers.
Well, first of all, emerging markets are not unilateral. There's bimodal, Tom talked about the tier 1, tier 2 cities in China, for example, buying as rich a mix as we get here in the U.S. In fact, slightly richer. We're not going for 100% of all the market. Second of all, the modeling here is essentially replacement price. We're not assuming that the PC price points all move up. They're still going to be $299, $399, $499, $599 slimmer notebooks, maybe not with the Ultrabook capability. Those will be based upon Intel Pentium and Intel Celeron devices, and those will be just fine. We think there is an aspirational quality to what's being done in the Core family around Ultrabooks that will accelerate what we've already done. We're not counting on everyone in the world spending an extra $100.
In fact, that would be crazy. What we've done is work the ecosystem costs, Kirk showed that, I think, quite well in his slide, such that $100 delta doesn't have to come out of Intel's ASP. I'm counting on price point replacement for the most part. I think there is a potential for a sell up inside of that Stacy talked about, but that's not built into our 40% number.
Let me just.
Right
I want to tie a couple of things you saw, which is, actually, I'll just tie it back to the chart that Tom showed, which is, he showed you that emerging market notebook retail, there is no difference in terms of the core mix between emerging markets and mature markets. To tie it back to Paul, we're just doing price point replacement. The question is, at the same price, at that $699 price point, is somebody going to want to come in and buy an Ultrabook? I think the answer is yes. Is there a possibility that there's a few more people that start buying up? If that's a yes, you get a richer mix.
All right. Let's go to Shawn Webster over here, then we'll do one more. Is there anyone who hasn't asked a question that has one? We have someone in the back there. Ruben. What do you want first?
Hi. Yeah, thanks. A couple of hopefully quick ones. In terms of the CapEx plan that you guys have laid out, without changing that, would that be enough for you to accommodate PC unit growth of 10%-15% for the next several years? That's my first question. I'm just trying to get a sense of the unit capacity that you're putting in place. Then the second one is on the capital intensity. You've tried to lay out why it makes sense in terms of the new capital intensity that you have for the last couple of years. Even when I look back at 2000, 2001, when you were making a 300-millimeter transition, that was the last time you had this same level. This just seems more sustained.
Is this the new world that we should expect even beyond 2013 in terms of the capital intensity, or should we expect it to die down to more historic levels at some point in the coming years? Thanks.
To your first question. Okay, ask the first question. Sorry.
The answer is yes.
Which is?
Are we putting in place capacity-
Yeah
over the next few years to be able to handle 10%-15% PC-
Oh
unit growth rate? Yes. Second question is yours.
I was going to say, there's a big difference between 10 and 15.
I understand that. We can tweak things.
Yeah.
The model assesses we can handle that.
For sure. I'll also take you back to the chart I showed that showed from 32 to 22 nanometer in terms of the unit growth expectation. If you just do the math on that chart, you're going to see it's in that range of what you just said in terms of what the historical unit growth is and what we're expecting for 22. You're on that trend. In terms of the capital intensity, the answer is that at 22, you're going to continue to see an increase in capital dollars per wafer, but we offset that by the fact that we get a scaling benefit. You're going to see an increase in capital dollars per wafer that's a little bit more capital intensity. We offset that by having an increase in the scaling benefit, and we can stay on the traditional cost curve.
The CapEx to COGS ratio should be constant over time at these new levels?
Yeah. What you would take from the chart I showed you where I showed revenue and CapEx is, we're at the high end of the range. We've been there a couple of years, and I gave you a strong hint that in 2013, I expect us to stay in that kind of a range.
All right, for our last question, we're going to go back up into this corner here.
Great. Thanks. It's Doug Freedman from RBC Capital Markets. One for you, Paul. When you look at integrating the graphics cores, can you give us a sense of what percentage of the value you were able to capture now that you've done that? Going forward, as you guys integrate more functionality, what percentage of the value are you targeting to capture in future integrations?
Well, before we integrated graphics into the processor, we were shipping
Graphics as a discrete chipset, if you will. It was integrated graphics, but it wasn't in the same package or on the same die. The overall value proposition has gone up slightly. We didn't lose anything by going from two chips to one, from a pricing standpoint. At the higher graphics SKUs, the higher-end graphics SKUs of our product lines, we're getting a premium as measured by the sell up to i5 and i7, as measured by the bill of material savings that our customers are getting from eliminating discrete graphics at the very high end. There was a second part that I-
Going for percentage that you hope to capture.
Other integration. We make that assessment. Dadi's guys have a great model to look at the cost of what we're integrating versus the value, sometimes we elect to share the value of that with the end user and our OEMs. Sometimes we elect to be able to capture that for ourselves, that really is a decision based upon competition, where we are in cycles, and those kinds of things. For the most part, things that we integrate, we want to get paid for.
Great. If I could, one for you, Stacy. In the past, you've shown your startup cost slide and how it's changed quarterly. Can you give us a sense of whether you're, in 2013, going to be able to stay within your gross margin target on a quarterly basis, whether we have startup cost accounting in the fourth quarter of this year?
How about that EUV question?
Yeah. You want me to give you gross margin by quarter for 2013?
Where is EUV when you need it?
I respectfully decline. I will say on startup costs, it's not going to look different than what you've seen in the past in terms of the shape. When you look at it as a % of revenue, it's right in line with what we've historically seen, which is why I didn't show the chart. It just didn't add anything from what you've seen over 10 years of history.
Great. All right. Thank you everyone for joining us. That wraps up the day. We'd love to have you join us, actually, across the street at the Robert Noyce building, where we'll have a reception. As you walk out of this building, you're welcome to either take a bus, but it's a very short walk, and there will be people escorting you across the street if you'd rather walk. If you do want to take a bus, buses are available where you checked in this morning in the cafe.