3GSM World Congress 2004

Paul Otellini
President and Chief Operating Officer, Intel Corporation
Cannes, France
Feb. 25, 2004

PAUL OTELLINI: Good morning. What I would like to speak on this morning is a topic that I think may be a little bit controversial, but I hope interesting to you. It is communications as the next digital revolution.

This is what we do at Intel. We create digital revolutions.

The first one, of course, was in computing. That digital revolution generated well over a billion connected computers now throughout the world.

The second digital revolution had to do with consumer electronics, and this is the movement of entertainment in our lives from analog to digital. As movies, as music, as home entertainment goes digital, it also allows us to be able to move it around, to time shift it to make it advantageous to us as consumers and interesting to us.

That is the second digital revolution.

The third one is what's happening in front of us in communications. In this space, what I'd like to talk about is what Intel does best. What we do best is we innovate around a variety of architectures, and then we integrate. We integrate those architectures into silicon over time.

In that context, I very much agree with the fairly radical notion that's up here from the Financial Times that says the cell phone is becoming just another module that you fit into the corner of any other silicon device.

And that is, really, what I think we're all about. What we're talking about is transitions and volume dynamics. Just like the cell phone was to telephony, what it unleashed all of us to be able to communicate any time we want, anyplace we want, Wi-Fi was to computing. It allowed us to try to greatly expand where we do computing and how we do computing.

If we look at how Wi-Fi has really evolved, if you compare it to other broadband technologies, you see something very interesting. This compares ISDN in yellow, cable plus DSL deployment in green, and Wi-Fi in orange. And it looks at the deployment of those technologies in millions of units from time zero.

You can see by the fifth year of deployment, Wi-Fi is now two to three times larger in terms of installed base than anything we saw happen before it in terms of broadband penetration. And in fact, people like Pyramid Technology expect over 700 million users of Wi-Fi by 2007.

Why did this happen? Well, let's take a look at it. If you look at early broadband, early wired broadband, this is what you had to deal with. It was a fixed point in place. If you look today at modern wired broadband, this is what you deal with. It's the same thing. It's still a fixed point in place.

Let me contrast that to early wireless broadband. It was a Pringle can, antenna, aimed out of someone's window, two people communicating with each other across an alley in New York City.

And modern wireless broadband has gone quite radically different. You can now get Wi-Fi anywhere in this town. In Paddington, many train stations; Boeing is equipping its airplanes, through Connexion, with Wi-Fi; and even on camels in the desert in Egypt.

What's different? What's different about the wireless broadband explosion that's happened? There are some fundamental differences, and they have to do with what I'll call viral growth versus evolutionary growth.

Wi-Fi is, in fact, a viral-growth phenomenon. It's unregulated. It came up through the IEEE standards bodies. It is a bottoms-up technology, embraced by users one by one, then millions by millions because it made their life of computing easier and where they want.

Very importantly, it's one standard worldwide. And because it's one standard, and because it's based upon computer digital technologies, we're seeing a very, very rapid deployment in terms of integration, smaller devices, lower cost, lower power, and ever more integration to other products.

As we go forward, though, there's a very interesting challenge, and that's quite simply that one size of 802.11 doesn't fit all. And in fact, beyond that, I think what you have to think about is what people want to do with this technology and, in fact, mobile technologies in general.

Our customers collectively want to be all ways connected, in many different varieties of different types of devices, but they also want to be always connected in a consistent fashion, whether they're on vacation or on the road or in a car, in a train, in an auditorium like this.

Increasingly, that connection will move beyond voice, the persistent lowest common denominator that we will all need, and include a variety of data types: Entertainment, information, business, purchases, transactions, you name it.

Those data types are digital, and they are, we hope, the kinds of services that will fill up the capacity that's being built for the variety of broadband technologies that are being deployed now.

However, as I look at the environment, it's very clear to me that one standard around the world is not sufficient. What we need to think about as we deploy our networks and build our devices is what consumers are demanding.

They demand choice, and increasingly, they're going to demand coexistence, because they don't care what kind of network they're on, they just want to be connected, and they want to be able to get access to those services that are now starting to roll out,.

If you look at the variety of technologies that we're likely to deal with in the next few years, there are four fundamental ones. Interestingly enough, three of those have come through the IEEE standards body, and of course the other is 3G.

Wi-Fi is being deployed now. WiMAX, which is a broadband metro area network, rural area network deployment of Wi-Fi, if you will, which I'll talk about later on, is coming out next year. And then the next one is something that's called WiMedia.

Think of WiMedia as the next Bluetooth. This is an ultra wideband technology, 500 megabytes per second that, will replace Bluetooth for proximate personal-area networks to allow for very high bandwidth connections, so you can download video, for example, wirelessly to a PDR that may be in your house.

Consumers are going to want their devices to work on all these networks at the lowest possible cost going forward.

To do this, we need to turn to what's happening in digital electronics. If you look today inside of a typical handset or cell phone that's out there, it looks pretty much like the picture behind me. There are typically one or two radios, GSM, Bluetooth, increasingly GPRS or 3G. There are about 200 components on the boards inside that phone.

The performance of the phone is roughly equivalent in computer speak to about a 286, a product that we introduced a decade and a half ago. The device is certainly optimized for voice, but it's a very limited software environment.

What's interesting is the rate at which Moore's Law is going to change the insides of these devices and, therefore, the capabilities.

Let me just take one path inside that handset and look at radios. I'm going to show how 802.11 radios are evolving, but you can take the same example and apply it to any other kind of radio inside the phone.

This happens to be 802.11. In 2000, this is the smallest 802.11 device we could find anywhere in the world, and it happened to be made by a competitor of Intel's. It was Atheros.

It was 50 percent digital devices and 50 percent Analog Devices. There were five large chips, if you will, large-scale integration chips on this device. It was built at 0.25 micron CMOS, but there were some devices that were 0.5 micron silicon germanium. Fairly expensive proposition to build this.

If you fast forward to next year, what Intel is doing is taking that same low power 802.11g radio capability, reducing it down to two chips. It is now 80 percent digital, 20 percent analog, and it's on 0.13 micron.

To address that myriad of other devices that sit around the high performance silicon, we are moving to introduce MEMS technology into our product line. We can take it down to one chip on less than or equal to 90 nanometer silicon technology and have all of the radio protocols that consumers are going to want on that same chip.

Now, you can think about this as Moore's Law in action. More capability, lower cost, lower power, higher integration.

The kinds of devices that we will all walk around with will change as a result of that. And we don't have to wait until 2008 for this to happen.

With handsets in 2005 and 2006, so next year and the year after, are going to be radically different devices based upon this movement of Moore's Law. There will be multiple simultaneous radios built inside of each of these handsets.

We are going to bring what I'll call mobile high definition capabilities to the handsets as well, taking what we've learned in the computer and bringing them into the handsets in terms of audio capabilities and video capabilities.

Of course, to do the application processing and the radio processing, we need to take the performance up, and we'll move from a 286 class machine to an Intel® Pentium® III class machine, something that many of you probably still have in the computers inside your houses.

Most importantly, to enable this movement to data services, there will be increasing amounts of PC coexistence between the software applications that are developed in the PC World and easily port it over to the handset world.

Intel is very aggressively working with many of the independent software vendors in the world to be able to easily give them tools to move their applications from one environment to the other. And we've done this now on Java, on Windows, and recently announced that we're doing this on Palm, on Symbian, and Nucleus as well.

So as the world of Wi-Fi and PC applications deploy, they're intrinsically data-enabled applications into mobile space. Those same applications will be able to move very easily and very quickly over to handset space.

To give you some example of that, rather than talk more about the what-if technology, I'd like to bring out someone to maybe give us a view of what integration and innovation looks like in a handset. And let me introduce Sam Arditi who is the general manager of our cellular and handheld product group.

[Demo begins and ends.]

PAUL OTELLINI: We wanted to give you an example of the kinds of integration that's happening around us in the not-too-distant future. I said earlier that we're also looking at bringing PC capabilities into this environment. I'd like to show you now a demo here of a product that we code-named Carbonado, and what this is is it brings VGA-quality video to handheld devices.

So, you can probably see on the screen up here in a second, one side is showing you a game, and the other side is a video clip from a movie. Full VGA resolution coming into handhelds.

This is essentially a graphics accelerator chip today. It brings us PC-like 3-D gaming and video into the handheld environmental. It's 640 by 480 VGA resolution.

We will bring this to production first half of this year, 2004. It goes into the first PDAs in the second half of this year. We have a number of major OEM design wins that we're not prepared to announce today, but believe me, they've happened. And we'll be integrating this same video capability, video and graphics capability, into discrete silicon in 2005, into phones, and into the apps processor itself in 2006.

I think that really changes the kind of base capability that you can start thinking about deploying to your customers in terms of the visual effects that you may want to bring to your application base.

So what's next? If you take that same Moore's Law integration I talked about before in terms of the radio, the same kind of capability is also going to give us the ability to reduce other components on there.

There will be one high integration device for application, subsystem and memory, there will be another integration device for the radio subsystem, and a third for the power delivery.

Essentially, the silicon content is now no longer the limiter for the ergonomics or the form factor of the handheld. In fact, you can do almost any kind of form factor you want because the chips are so small that they can really be produced in a variety of different and, I think, very innovative kinds of capabilities.

The capability here for these handsets will be really quite stunning. It is simply broadband capability. All of your content, 10 gigabytes worth. You can do content consumption and even some limited content creation on these systems.

Now, it's not just the handsets that I think that Intel is focused on. It's also the infrastructure. This is an industry, in terms of the back-end, the central office requirements, which really has been built up around a number of proprietary standards over the years, over the hundred year history of this industry. And what we're seeing now is a very rapid evolution to a standards-based architecture, and it happens to be one based upon something called ATCA. ATCA is a standard that Intel helped develop. A number of companies in the industry are now adopting it.

What you see is a standard that was first completed in terms of base specification at the end of 2002.

By the end of 2003 Intel and a number of other companies were in production in ATCA blades, helping our customers improve their time to market, increase the cycle time, lower their cost and lower their footprints.

There have been a couple of notable examples here. One is NEC. NEC late last year introduced a point product that allowed them to take their development time down by two-thirds, increase performance by 10X and take the footprint down by 80 percent. That was a point product.

Two days ago, we were very happy to hear and work with Siemens mobile technology in their introduction of an entire new product line, which I've got an example of here. It's their next-generation telecom architecture. It's based entirely upon ATCA. So it looks like something you would see in a combination of a data center or a back office. And it's a standard blade architecture. Let me just pull one out here.

This happens to be a dual processor Xeon blade. The nice thing about this is it mixes computing and packet processing seamlessly. So depending on your requirements for compute, you can use things like the Xeon blades, or if you need to optimize your own packet processing, you can also put in network processor blades. They mix and match, it's very standard, and we're very excited about Siemens's announcement here. We think it's, indeed, an industry-shaping announcement.

Now, I'd like to kind of close on talking about filling up the gaps. We've talked about Wi-Fi, we've talked about telecom infrastructure, we talked about the evolution of handsets. What's next?

I think the next X factor is a technology that I talked about earlier called WiMAX. WiMAX is metro area network wireless broadband.

Intel is one of the leading companies in this. We are chairing the IEEE committee on WiMAX. It has gone from six companies in that forum a year ago to over 70 today. We see quite large adoption and rapid embracing of this technology.

We'll have our first silicon this year, 2004. There will be mainstream deployments in 2005 with CPEs, base stations, residential gateways and so forth.

In 2006, it's Intel's intention to make WiMAX an option in our Intel® Centrino™ mobile technology platform. We are also considering making 3G an option in that platform in that same time frame.

And by 2007, we'll integrate WiMAX into the silicon and the radios just like I showed you on that Moore's Law chart earlier.

I believe that this integration by Intel will lead to very rapid adoption, and let me give you an example of that.

This chart shows you, in blue, the total available market for notebooks in the world. And it shows 2001 and 2002 here. The green bar, little green bar, is the Wi-Fi deployment into notebooks in those same two years. I'll label this era BC, or Before Centrino.

After Centrino you see a dramatic difference. The green bars jump very rapidly in the last couple years to where Wi-Fi is now in the majority of notebooks that are shipping, not just with Intel but from any vendor that's out there.

We're going to see a similar inflection point with WiMAX as we start driving the integration of that into the technology. And in fact, we're projecting something like a 60 percent penetration of WiMAX into notebooks by the 2008 time frame.

So the same kind of curve that we saw for Wi-Fi I believe we'll see for WiMAX.

There is, as I said a minute ago, a lot of early momentum around this. There are thousands of pre-standard trials going on today with over 130 companies participating.

I talked earlier about the surge in WiMAX foreign membership, up to 70 people, and there now are in place a number of agreements between Intel and hardware OEMs to develop standard-based interoperable WiMAX customer premise equipment and take it to market in 2004. You can see some examples up here of people that are deploying in a variety of cities around the world.

But rather than just look at the chart, it might be interesting to see a video on one of the more interesting early deployments of WiMAX. Could you roll the video, please.

(Video plays and ends.)

PAUL OTELLINI: So, I tried to paint a picture of this next digital revolution. WiMAX, Wi-Fi, Moore's Law, high integration, high performance silicon are really going to change the way people get broadband and the way that people use broadband, and the way that broadband is deployed from a technology perspective.

And it's encompassed by a number of things. This is high performance apps processor; it's the all-digital radios; PC-like capabilities in terms of audio and video; I think, increasingly, global standards driven by a number of bodies; low-cost infrastructure; and of course all riding Moore's Law.

What we're talking about is a win for all three constituencies. For the consumers, of course; for the equipment manufacturers; but also for the carriers, because you're approaching an environment where you can see persistent standards-based products to deploy your applications and service on that will have a persistent architecture generation after generation that you can count on.

Our view that is in this space, the best is still to come.

Thank you very much.

* Other names and brands may be claimed as the property of others.