Intel Developer Forum, Spring 2003

Patrick Gelsinger
Senior Vice President, Chief Technology Officer, Intel Corporation
San Jose, Calif.
Feb. 21, 2003

PAT GELSINGER: It's just great to be here with you this morning. Thank you for a fantastic week of IDF. We started the week with a blizzard and snowstorm in the east. We end it with a bright, sunshiny day here in San Jose. Thank you for a fantastic week.

What we like to do is take a glimpse of the future, open up the doors of our minds of the research and labs and look forward and explore further the opportunities of convergence for the future.

If you think back to IDF a year ago, we started by this idea of extending and expanding Moore's Law. Expanding Moore's Law was using Moore's Law in new ways, areas and technologies that we never explored before; extending Moore's Law was continuing to move forward the nanotechnology and the abilities of our putting more and more transistors onto silicon. And with that, we spoke last fall about nanotechnologies and seeing how far we could take them.

What I'd like to do is take a look at what some of the luminaries said a year ago and look to see what Gordon and others had to say about this idea of extending and expanding Moore's Law.

(Video begins and ends.)

PAT GELSINGER: Let's stop the video right there. And we had seen Vint Cerf, Gordon Moore, and Bob Metcalfe in that video. What I'd like to do is invite Bob to join us via video. Bob Metcalfe was the inventor of Ethernet. Thirty years ago, he coined Metcalfe's Law. And I'd like to have a chat with him this morning.

So good morning, Bob. It's great that you took time out of your schedule to join us this morning. Thanks and welcome to the Intel Developer Forum.

BOB METCALFE: Good morning. It's a great honor to be here.

PAT GELSINGER: Good to see you again, Bob. I'm curious, Bob, when you invented Ethernet back in the '70s, did you have any idea what was going to happen with this little thing that you were working on at the time?

BOB METCALFE: Well, I wish I could tell you we had mapped it all out, the whole 30 years of it. But no, I have to say we were clueless. We were just simple country engineers networking some of the first PCs together and connecting them to this laser printer we had built down the hall. And I don't think we had any idea that Ethernet would over the next 30 years, would get to where it is today.

PAT GELSINGER: You know, Bob, when I was an engineer, just starting at Intel, I was working on the 286 chip. And it turns out that about 10 cubicles down the aisle were the engineers at Intel working on Ethernet chips. And, I walked down there and I had just no idea what they were working on and never spent the time to figure it out. But, man, what a phenomenal thing came from that initial work. You know, Gordon called Metcalfe's Law the other exponential. What do you think about that?

BOB METCALFE: I'm flattered that Gordon would even mention Metcalfe's Law in the same breath with Moore's Law. Moore's Law was a coordinating self fulfilling prophecy for the industry. It said you could count on progress like this, so let's all get ready for that progress, do our role in creating that progress. Moore's Law is a self fulfilling prophecy that led the industry in the right, exciting direction.

PAT GELSINGER: Precisely. Moore's Law talks about the capabilities for chips. Thus resulting in more performance per chip or more things being integrated to a chip.

BOB METCALFE: Metcalfe's Law has been enabled by Moore's Law. Metcalfe's Law, which says that big networks are valuable, the number of users, the number the value to the user of a network grows at a square. That's been enabled by Moore's Law in two ways. Moore's Law has given us more things to connect, and therefore more value. But Moore's Law has also made it possible to build smaller and faster Ethernets, three megabits, 10, 100, a gigabit per second, and now 802.11.

PAT GELSINGER: Metcalfe's Law has defined the value that emerges from the network. If you take the capabilities of Moore's Law and converge it with the network, how would you describe those coming together of Moore and Metcalfe?

BOB METCALFE: Well, should it be M plus M or should it be M times M or should it be M to the M? I'm not sure. But what's happening there is that the impact of Moore's Law is multiplied or exponentiated by the impact of what people have come to call Metcalfe's Law. That is, as the power of the nodes that you're networking goes up, well, clearly that power is amplified by connecting things together.

PAT GELSINGER: When you think about what you see going on today in technology, what excites you? What are the things that get your passion and really wow you again, Bob?

BOB METCALFE: Last report, there were about 8 billion microprocessors shipped every year. And I think the number is fewer than two percent of them are networked. Maybe fewer than one percent. So that's a lot of microprocessors that are currently being shipped. More than all the people on earth. And they're not being networked, which means they're not as valuable as they could be. So the whole notion of putting a radio on every chip promises to build great value in the future. And coming from Intel, with Intel's history and resources and commitment to building infrastructure, I think Radio Free Intel is the most exciting thing around these days.

PAT GELSINGER: Bob, now it's my turn to be flattered. But do you buy this idea of convergence, some sort of opening up of a totally new scale for the network and what communications and computing are going to mean to users in the future?

BOB METCALFE: Yes. Convergence, the bringing together of things that were previously separate. When we have separate telephone networks, separate television networks, separate data networks. As a temporary matter, that's okay. But long term, they need to converge. And as they then converge, instead of being additive, they'll be multiplicative. As those media are brought together, then what we can do with them will be much more valuable.

PAT GELSINGER: It seems to be an exponential win win. It's going to enable these products to be connected to each other. It's going to make them more valuable as they become connected, as these new usage models and users emerge.

BOB METCALFE: Well, we tend to think now that most of the networking in the world is among people. I think as we get the other 8 billion or more microprocessors connected, more and more of the traffic will be not directly between people, but between intelligent devices communicating on behalf of people. It's the next big thing.

PAT GELSINGER: Yes. It's exactly one of the things we'll be touching on today. Bob, one final question. We're here at the Intel Developer Forum, thousands of our closest friends. What kind of role do you see for Intel and for our developers in this new converged world?

BOB METCALFE: Well, Intel for a long time has understood a very important fact about the way technology proliferates, which is that the as long as you're careful to build a platform and to install infrastructure, the vitality, the unexpected comes from the users, the developers who use the platform and their users following that.

When we did the first Ethernet, we didn't expect what would happen, but we had been taught, as Intel well knows, to build things to be general purpose and as a platform that others can then use to build new things with, and then you just hold your breath and watch the unexpected happen. So I'm quite optimistic. I love these IDFs, as I think that's where all the action is, in technology development.

PAT GELSINGER: We're pretty excited about it as well. Exponential value, Moore and Metcalfe. You know, I think we have a huge, huge future in front of us. You know, thank you so much, Bob, for joining us this morning.

BOB METCALFE: Thanks for including me. I appreciate it.

PAT GELSINGER: Ladies and gentlemen, Bob Metcalfe, the inventor of Ethernet.

(Applause.)

PAT GELSINGER: You've heard Intel talk about this idea of convergence. You heard Craig Barrett open the conference talking about the idea of convergence. It's come up in other keynotes this week.

But I really want to know what we really mean when we use that phrase. What's this convergence thing really all about? And what I'd like to do is just take a closer look at it and try to dissect it just a bit with an idea of micro and macro convergence.

What do we mean when we say micro convergence? It's really the coming together of two technologies, two things, to produce a third. You can take a very, very simple example, like the integrated circuit and the abacus. They came together and we formed the calculator. Another example might be where we started with the microphone. We had the speakers, right. We brought them together. We had the headset.

Another example might be where we were exploring microwave technology. We had ovens, we were exploring microwave, and it's actually very, very interesting because there was an engineer who was working for Raytheon as part of the war effort in 1945 and he was working on a magnetron and seeing how he could miniaturize a magnetron. What he did was he had a chocolate bar in his pocket, and as he was working on the magnetron, the chocolate bar melted in his pocket. It ruined his shirt but more importantly, that led to the development of the microwave oven. Another example of convergence.

We see today the notion of convergence, just the PDA and a cell phone. In fact, for Christmas, I bought my wife one of the Treo PDA phones. She now has everything on it. Her calendar, her appointments and everything else, and she's the envy amongst our friends with her Treo phone.

What's really powerful about micro convergence is when we bring together these technologies into silicon. First we integrate them together, so we pull them together onto the same die. When we do that, it dramatically reduces the cost of those devices. So we're able to make them at much, much lower price points enabling all sorts of new capabilities, new users.

Micro convergence is this collapse, this bringing together of things at the smallest scale.

But we think there is something much, much more powerful that happened; that micro convergence, this bringing together two technologies to enable a third, leads way to macro convergence.

What we mean by that is sort of like what we said with Bob. He had no idea of this Ethernet, but it led to the explosion of networking overall.

We think about macro convergence, we think of these four elements: we see it enables new usage models, it enables new infrastructure, creates new business models, and ultimately creates new norms, behaviors, new societal changes as a result.

Let's take a simple example of that to try to explore this idea just a little bit further. We had the automobile, and the automobile was the coming together of two technologies. We had carts, and carts were used since the 2000 BC in India. We had these steam engines that were being developed in the 1700s. Then in 1769 was the first example of a car coming together and not very effectively, and Nicholas Otto developed the internal combustion engine in the 1850s and enabled really the first automobiles to emerge.

Today, over 40 million cars per year are being built, micro convergence, the coming together of carts and engines to build the first automobile.

Today those automobiles have become the American dream, worldwide you see this idea of the ability to own a car.

When I was just a young guy, I was working on a horse farm in Pennsylvania where I was growing up. One day there was a nearby auto place, and they had a car coming into their lot and they didn't have a good place to unload it so they brought it to the horse farm where we had good spots to unload horses. Off rolls this Corvette, T top, scoop doors. Pure lust. I wanted that car. That was my dream at the time was to own a Corvette just like that one.

And this idea of cars, this micro convergence gives way to macro convergence. We think about it. Cars have enabled us to go shopping, to go to fast foods, to take vacations. It enables a whole set of new usage models.

Additionally, it enables new infrastructure. We have the interstate highway system that emerged. We have support infrastructure, like gas stations. Auxiliary industries like oil and rubber.

Furthermore, we look and we see new business models emerging: transportation, delivery, rental businesses, trucking, disrupting the entire rail industry of the time.

Furthermore, new norms and behaviors in society emerge as well. Think about it. We saw the collapse of speed, distance, time, and direction, right? This gave rise to suburbs and cities, restructuring society entirely.

In fact, I have a 16 year old son and he just got his driver's license a few weeks ago. In many regards, that's the rite of passage into adulthood, this new responsibility, this new freedom of being able to go out with his car. And for me as a parent, it's a rather scary event as well.

So we see micro convergence giving way to macro convergence. Part of the reason we're so excited about this idea of convergence as it comes to our industry is, in fact, that we see that we're in the dawn of macro convergence, and what it will do for us in the future. We have Moore meeting Metcalfe, we have the PC, we have these native connected devices. We believe again this will give way to macro convergence. We see new usage models emerging. We see the ability for people to go online. We see new business models emerging, new ability to commerce, video conference, entertainment, new infrastructure emerging as well.

Finally, new norms and behaviors.

In fact, I had the other night, my son had a friend over, Kevin. He was visiting and we get home after going out to eat, playing some games, watching a movie and so on and he says, "Dad, can you take Kevin home so that Kevin and I can play games together?" And I looked at him sort of funny. He said, "Yeah, we don't have enough computers here at home, so we want Kevin to be on his computer so we can all be online together." Sort of a new idea of how, in fact, you play with your friends. New social norms and behaviors.

We believe that we're just on the cusp, the beginning of this idea of macro convergence. Many of the things you heard about this week we think just place us at the doorstep of realizing this next wave.

We've seen the digital home, we've seen these ideas of convergence in the enterprise. We've seen mobile wireless Internet clients. We think we're on the doorstep.

The last two years have been great progress, but we think the best is yet to come. The future is nearly unimaginable, the opportunities that will emerge.

What we wanted to do this morning was go back to the future. We wanted to try to project and extend our thinking, our minds, our vision and look at a couple of scenarios and what these macro converged worlds might look like into the future.

We're going to take two scenarios this morning. The first one, we wanted to look at what manufacturing and how it might be transformed based upon these convergence of technologies.

So if you could join me just for a second, we're going to go through a time warp here. We don't have William Shatner with us today, but the transporter is working so we're going to go out to the year 2015, and coming out of the early 2000s, the industry went in a period of incredible growth. Based on convergence, we had to go on the most aggressive capital spending spree that the industry has ever seen to keep up with demand.

We went to build 30 new fabs over that 10 year period, the fastest growth of the industry ever. And I'd like to have you join me in Fab 52, which is located in Chengdu, China, and it operates the first 450 millimeter fab operating with 11 nanometer devices. Welcome to the future. Diana, how are you today?

DIANA: Hi, Pat. It's nice to be here. So I understand you'd like to hear about my life as a fab worker in the converged future.

PAT GELSINGER: Exactly. Take us on the tour of the fab.

DIANA: Sure. Let's get started. So the first step is pass downs. When I come on I get pass downs from the previous shift and that tells me what they've been doing so I can take up where they left off.

You notice that the computer asked me verbally if I wanted the pass downs, and this is possible because there's speakers and microphones throughout the fab, and even throughout my suit.

An alternative to verbal is when I come onto my shift, all my pass downs, other relevant information is transmitted automatically wirelessly to my handheld. So that's another way I can get information and it's updated regularly throughout the day as necessary.

PAT GELSINGER: So we see the critical role that natural human interfaces play in this environment. Everything is sensory enabled and we see wireless communications occurring everywhere throughout the fab.

DIANA: Exactly. So there are sensors throughout the manufacturing facility, and they detect changes in environment.

(Computer speaking.)

DIANA: I think they've just detected there's a contaminant that's entered, and I think they're talking about you, Pat. It's okay. He's with me.

(Computer speaking.)

DIANA: We're going to be a little restricted today, but we'll work with it. So the previous shift was working with a problem with our wet etch station, and so I'm going to take up where they left off. So can you get me the procedure that the last shift was working on? So I can look through here, see where they were. I think I can understand what's going on. So it's really great. I get instance access anywhere in the fab to all the processes and procedures I need to do my job.

PAT GELSINGER: Okay. So enabling this is powerful processing, all this information, it's available anywhere throughout the fab, it's context aware, it knows where you are and how you're operating. And all of this is happening without your explicit directions on what to do.

DIANA: Right. It's just there when I need it. It makes it really easy. So let's pick up with this problem, see if I can figure out what's wrong with our wet etch station. I can use more data. Can I get some more data on that machine? Okay. I'd also like to know what effect this is having on the wafer. So can you get me the nearest test slot?

COMPUTER: Nearest slot located estimated arrival time four minutes.

DIANA: So I'll be able to check that out. These saturation numbers are really high. It looks like there's a flaw in the device. Can I get the machine schematic as well. Okay. You know, it's really hard to see on this small screen. Is there a bigger display?

COMPUTER: There is available display about 10 feet to your right.

PAT GELSINGER: Here we go. Oh, check it out. This will be much easier to see. So you can tell it told me I needed a new display. It could go find where it is because it's keeping track of where I am throughout the fab. It makes it really simple. Well, Diana, it looks like demand is booming so you better get back to work here in the fab.

DIANA: Okay.

PAT GELSINGER: Thank you very much.

(Applause).

PAT GELSINGER: So what we wanted to do was explore the use of these technologies and look at what some of the technologies that we are working on today and how those are enabling this wireless future. We saw sensor technologies, wireless technologies, and many other areas that would enable such an environment in the future. Let's explore the work we're doing today, some of the things that we as an industry are enabling, and how those might create this macro convergence of the future.

A year ago we launched our Radio Free Intel program. We were talking about how we were trying to bring communications to every chip and you heard Bob talk about that this morning.

What you see is we have a set of technologies, and when we first launched this we had dynamically reconfigurable radios, and we had silicon radio, MEMS, and intelligent roaming as part of our radios. In the fall we gave updates on those and demonstrated several of those and added to it smart antenna systems as well.

At the first IDF we held our first regulatory discussion, we had Bob Pepper, CTO of the FCC here. Increasingly as we move into spectrum and global deployment of these technologies it becomes very, very important for us as an industry to be actively participating in these policy debates.

What I'd like to do is give a specific update of some of the work we're doing to enable this next generation of wireless communication.

What we have this morning is a demonstration of our wireless development platform. We call this Ironsides. This is a platform for innovation. It allows us to explore new protocols, new research and smart antennas, software defined radios.

We want to take this platform and then influence research with it as well as use it to demonstrate to policy and decision makers, show them with real data the impacts on their regulatory policy directions.

We view this as a platform as well to validate our silicon architecture directions for building these next generation radios. Today for the first time, we're going to bring Ironsides live, we're going to give you a live demonstration of it.

What we have here is, we have the Ironsides platform. What Ironsides is, it's a compact PCA rack that we have here. We're able to put in place parallel baseband and PHY boards into it. And I have one of those here. So just a huge array processor that we've built inside of one of these devices. You see we have an RF node here, it's a little radio. We have this system set up here, this is our research platform.

We're going to connect it up to a commercial radio. This is the first operation that we've ever done publicly. So look, no wires, all wirelessly connected. Trust me, it's really working. We're going from our radio here, connecting and being able to transmit to our radio over here.

I'll come over to this side, and we're just going to do a little video connection between our two machines. And popping up over here on this screen is the video connection that you can see coming across our system.

So we're taking a commercial radio, being able to connect, transmit, receive across our Ironsides research platform. From this, we're able to show the research results of putting in place such a research platform today.

But we're taking this even further. Because what we want to do is we want to not only explore research of protocols. We also want to develop new architectures for how we deliver radios in the future.

Fundamentally, we think DSPs are running into a dead end of how they process and handle protocols for research platforms in the future. The result of that is we think because of memory latency, interdependencies inside of these wireless protocols, that we need to explore new architecture as well. That's what's shown on this display here, an array architecture or as we call it a radio adaptive architecture. We're exploring new architectures. We're also going to take these directly into silicon and build these into small things that we can integrate all the way into our chips in the future.

We're taking this to silicon for future potential production as well. So our first ever demonstration of the Ironsides research platform for Radio Free Intel. We believe from this, we'll be able to accelerate and deliver more rapidly the radios for the future.

We also described our work in ultra wideband band (UWB). We need high performance, low bandwidth, low power wireless communication.

What we've shown through this week is the idea of convergence of CE, PC, and bringing together these multiple islands into the future. What we'd like to show is some lab results of our UWB work, and when you look at UWB, the idea of UWB is you take a very, very small low-power signal, that essentially is noise. And you transmit a signal that's very, very broad across that signal, so you take a very, very low power signal and you cover a wide range of spectrum. The FCC said in their ruling from three to 10 gigahertz. Now, the problem with that is it's a very, very wide signal. And what we've explored is the idea of sub-banding, being able to carve that spectrum up into segments that you then operate across. And so this idea of sub-banding is the fundamental direction that we believe that will make UWB useful. And we're taking to the IEEE 802.15 standards committee that's been form, the whole architecture that's been formed for building sub-band.

This is an animation of real lab results that we're demonstrating in our research labs in Oregon.

Part of the benefit of sub-banding is imagine that the radio detects that there's a lot of noise in a certain portion of the spectrum. It can simply not operate in that segment of the spectrum.

Furthermore, and dynamically being able to reconfigure itself to operate and select the portions of spectrum it operates in.

We also believe that there's a great opportunity to come up with a universal PHY for UWB. And the basic idea here is that we see that there are many, many potential uses. We have the consumer electronics, mobile, PC segment, all of those needing the same high bandwidth, our goal was 500 megabits per second, low power, very inexpensive, short range connectivity.

Thus the architecture that we want to pursue is a universal PHY and MAC, which then enables many different software protocol stacks on top of it such as USB, UPnP and IP, Bluetooth, 1394, all of those upper layers being able to take advantage of a universal PHY layer for UWB.

This is what that we are pursuing with the ultra wideband and believe it will be a critical step in enabling the next generation of communications in the future, a common UWB radio platform for a wide range of CE, PC, and mobile uses in the future.

In our manufacturing scenario, we saw also that there's tremendous amounts of data flowing. We see that there's this networked amount of data, data being made available everywhere, we have new types of sensing information that's coming into this environment as well. We see that there's a critical need for high performance capable processing of the networking infrastructure as well. We've been doing research how we can take packet processing into small, upgradable devices. At ISSCC last week, we showed our first 90 nanometer technology, we call it our extreme processor, which is operating on dual VT 90 nanometer CMOS. And the important piece about this is, given that it's dual VT, we are able to have both the high performance as well as the low power capabilities being demonstrated on silicon.

We've taken that test technology, put it into a test board. Today, while still preliminary, this is operating in excess of five gigahertz today. We are demonstrating packet rates of over seven gigahertz per second, you know, being able to handle all sizes, small as well as large packets. As you can tell in our test packaging, very small piece of silicon that we believe will be integratable into the future. And as we continue debugging and designing this, we expect this to operate well in excess of ten gigahertz, being able to handle line rates well in excess of ten gigabits per second. Our goal is make TCP/IP an integrated part of our chipsets and platforms into the future.

We describe last year our work in sensor networks. We believe there is a powerful new computing platform that's going to emerge as we combine computing, communications, and sensing together into a single platform. What we've done is and what I have here today is our next generation sensor platform. And you sort of think about this as a reference design for sensor research. Sort of like a reference design or a concept PC, this is a reference design or a concept platform for sensor research.

We're taking this architecture, we're building silicon, which we have in fab this year, which will allow us to further reduce the cost, drive up the integration, decrease the power, and improve the performance.

We also described our work in building a software stack for this technology, a tiny OS and tiny Db, how you program it. We've launched two public trials, one is in the area of agricultural research. This is a vineyard that we've sensed in the region in Canada, as well as the habitat monitoring that we announced with the Great Duck Island.

We see there's many, many potential applications for sensors into the future. We see that they could be used in environmental monitoring, agriculture for process and manufacturing, for structural environments, for fire fighting and rescue. We'll be demonstrating and showing one more that we're announcing for the first time later in our keynote this morning. We believe this is a new paradigm, ubiquitous, proactive computing, being able to collect, deliver, sense information that was never possible before.

Another aspects of the demonstration that we showed was context awareness.

As we walked into the fab, it was able to detect my presence and take action on it. It knew where Diana was and be able to point her in certain directions. We believe this idea of enabling with context, enabling location aware computing, becomes a critical direction for the future.

Our Seattle research lab has been working this area and has launched an initiative context aware computing.

Now, at the bottom, you see that there are many potential sources of location information, GPS. Right, you have enhanced observed time of difference abilities. You can triangulate from axis points or wireless base stations. Maybe it's the last known location you were. Maybe it's aggregated information, you're on a train and it knows where it is, so you know where you are.

We then take all of those sources of location information and we fuse them together through fusion services that can then deliver to applications, location awareness.

Today, there's hardly any use of location aware application. Our goal through this work is to make location part of every application in the future, tracking assets, being able to build worldwide radios, 911 services, "Where's my child?"

A few months ago, I was just sitting down in a plane just before they close the door, you can still be chatting on your cell phone. And I get an urgent call from my daughter who's going to the airport to pick up one of her friends. She is panicked because she is lost. She has no idea where she is. She's gotten totally confused. And she's trying to explain where she is and have me help her get to the airport.

And, of course, you know how timing goes. About 30 seconds after the phone rings, I'm totally baffled on what she's trying to describe to me. The stewardess is breathing down my neck, saying, "You must turn off your cell phone now." You can imagine the sinking feeling in my stomach knowing I was not able to help my daughter who is now lost and very concerned. We want to bring location awareness, context aware computing to everything we do.

Another aspect to the demonstration that we saw was natural interactions. The idea of speech recognition has been the technology that's been two years away for the last 20 years. We've been working on this forever. And it's always bad and inadequate. What we've done, and this is work going on in our China lab, what we realize is that the actual speech recognition algorithms are actually quite good. But they work very poorly on a noisy signal. What you see is that the error rate for speech recognition is actually very good in very quiet, perfect environments. It becomes very poor when you move to noisier environments.

So what we are doing is to build better microphones. We want to improve the quality of the signal. What we have here today is an array microphone board that we've built. And so today, how many ears do you have? Two. How many microphones do you have in your PC? One, right?

What we want to do is enable array microphones to PCs we're going to have six years in the future. Based on this technology, it dramatically improves the quality of the signal. We have directional information. We're able to synthesize and eliminate erroneous noise and provide a high quality signal.

We believe at the platform level, this is the most powerful thing that can be done to enable high quality speech recognition into the future.

Louis Burns in his keynote on Wednesday talked about this capability being built into our platforms in 2004 with our ICH chipset that we shipped, and in high volume beginning in 2004.

We want to take our work even further, because when the human does speech recognition, they don't just listen. They don't close their eyes and just listen. They apply their visual recognition as well.

So our research work is to take statistical processing. Specifically, Bayesian networks, and that allows information to be done with data, add stochastic processing and add high quality audio with vision, lip reading, direction, sensing, and be able to bring those two together. We've been working in this area. We have an open vision library we have available to the research community. When I add vision to speech, to audio, I get dramatic improvements in the data rate, and this is all measured data from our research lab in China. We believe this will be the next step in natural human interactions as well.

So we've looked through a number of specific technologies, work that's going on today, advancements of our research work that we've been describing to you over the last year that we believe will fundamentally enable environments like we just saw in our manufacturing scenario.

But manufacturing is great, and it's sort of the life blood of what we do at Intel. But let's explore another aspect, another potential use and user for the future. What I want to do is explore how these technologies might apply in a home or a health care environment.

To do that, I'd like to invite Eric Dishman to the stage, and Eric heads the proactive health research activity for Intel. Eric, if you could join us.

ERIC DISHMAN: Good morning, Pat.

PAT GELSINGER: Good morning, Eric. So, Eric, I understand you're a social scientist. What's a social scientist doing at a company like Intel?

ERIC DISHMAN: Actually, I get that question a lot. Intel has the largest social scientist team in the industry, and I'd rather spend my time doing social science research where you're going to impact billions of people as opposed to writing an obscure article no one reads.

PAT GELSINGER: That's pretty powerful. So what are you working on?

ERIC DISHMAN: There are two taboo topics I want to talk about tonight. My mom always told me never talk about aging or illness in public and I'm actually going to talk about both of them today. And it's partly because for our industry to ignore the realities of what's happening with the aging population would mean we're missing this huge opportunity to look at a big macro convergent trend.

So I wanted to start with a survey today of our audience. Raise your hand if you were born between 1946 and 1964. So you are the baby boomers officially, those born between '46 and '64.

How many of you are already experiencing the challenge of you or someone in your household taking care of an aging parent today? All right. If we fast forward to IDF 10 years from now and I asked you that question, how many of you think you're going to have to deal with this issue of helping to care for an aging parent over the next 10 years? You start see a lot more hands go up.

PAT GELSINGER: Maybe half the audience.

ERIC DISHMAN: Let me ask you this question. In your own retirement, whatever year that maybe, how many of you actually want to go into a nursing home? That's actually true of this survey we do internationally as well, that almost nobody wants to go into a nursing home.

So what I want to share with you is while our industry was paying attention to the Y2K problem in 2000, there was this huge population shift, and Joel Cohen's quote captures it nicely. In 2000, this was the moment where we crossed over, where for the rest of human history there will be more old people on the planet than young people.

PAT GELSINGER: So we had the technology Y2K, but the social Y2K was probably even more important.

ERIC DISHMAN: That's right. We are all sort of sidetracked by the technology. And this has huge implications for convergence. Let's look at this graphically. This is a map of the global population in 2002.

This is the people who are over the age of 60. And most of this is yellow or orange, meaning zero to 9 percent or 10 to 15 percent of the world population in those different countries are over the age of 60. What's fascinating here is that there's no blue or purple. You do notice that Europe has lots of pink.

PAT GELSINGER: We have a lot of Europeans here today. You're not saying they're old, are you?

ERIC DISHMAN: No, I'm saying they're sophisticated and wise and leading the charge for the age wave. What's interesting is you fast forward to 2050 and suddenly purple and blue are all over the map, and purple and blue mean 25 to 29 or 30 percent or more of the population in these countries are over the age of 60. This is going to have a huge impact on our future.

PAT GELSINGER: It's a stunning demographic shift. But 2050 is a long way away. How about something in at least my career?

ERIC DISHMAN: So actually, even just 10 years from now, so if we're in IDF 2013, half of the work force in places like Japan, the United States, and most of Western Europe will be working a second full time job to try to take care of an aging parent. That has huge implications for productivity and it's driving a bunch of new uses and needs for technology to help with those problems.

So what I want to think about is if you're in this context of this massive worldwide age wave, you as the audience are going to be caregivers. Half of you are going to be taking on this role in your life.

You're going to be the first generation that actually spends more dollars and more years taking care of your parents than you have of your own children. That's a huge thing that most people don't realize that's coming to a home near you.

PAT GELSINGER: That's stunning. I think about caring for college, but now we have to think about caring for ageing and for our parents as well. That's a stunning shift in our thinking.

ERIC DISHMAN: That's right. And the second issue is in your own retirement, you are going to be a different kind of elder than what we have today. You'll be much more proactive about your own health. In fact, we're already seeing you as the first generation who are using the Internet to get health care technology information. You're using home diagnostics to start taking care of your own health.

We're going to see that explode over the coming decades. But partly fueled by micro convergence and partly fueled by the macro convergent trends going on here. What this means is you will be the first generation using the digital home as a platform for wellness and for aging in place.

PAT GELSINGER: Louis talked about the digital home on Wednesday, but he's talking about entertainment and media and all of those types of devices. The digital home for health care?

ERIC DISHMAN: It's pretty much the same infrastructure but a whole different set of usages and value propositions for it.

So what we're talking about is several trends coming together. All of that technology that Louis talked about for the digital home is the first trend that we need in place. The second is this huge worldwide aging demographic we just talked about.

The third is this explosion of medical diagnostic devices. If you look at diseases like diabetes today, you already start to see lots of these on the market, but those are poised to explode for all kinds of other conditions in the coming years.

And the fourth is actually the huge challenge that countries worldwide have dealing with the cost of health care. In the United States last year we spent US$1.3 trillion on health care. That's about 15 percent of our gross domestic product.

PAT GELSINGER: And if those demographics occur, what percentage could this be?

ERIC DISHMAN: The optimists say 20 percent of the GDP and the pessimists say 30, even 40 percent of our gross domestic product.

PAT GELSINGER: So a full third of everything we do goes into health care.

ERIC DISHMAN: That's exactly right. Because of that, the pressure on health care is to say we need to impact people's behavior at home, at work, out in their lives, not just when they come to the clinic and hospital. So the home is becoming a site for health care to happen.

So it's this convergence of these technologies and trends, we're going to say there's a whole new set of value propositions for that digital or communications technology that's going to help us with primitive medicine, the early detection of disease, and to help us maintain our independence and stay in our own homes, since as you all saw, none of you actually want to go into a nursing home.

I want to give you an example really quick from some current field work we've been doing. This is Milford Rice rise and what we see is Milford and other consumers we're studying today already using today's rather unconnected, unwieldy technologies to drive these kinds of value propositions.

He's 69 years old and he actually has a heart condition. He's trying to use his treadmill every day, he's trying to take his blood pressure every day, and actually he types in all of that data and shares it with his mother who is 90 and also uses it and she e mails it back to him every single day. Because they found they can bug each other and support each other in this goal to have a healthier lifestyle, which is a hard thing to do alone.

The fact that he is doing this today and his mother is doing this today is in spite of the technology.

Let's look at what the digital home in the future might look like and how you might simplify this in many of the same ways as you talked about with the manufacturing scenario where you have a simple at a glance interface. In this case we have an image of mom's fitness coach that shows during most of the week the activity was quite good, but towards the end of the week hasn't really had much exercise. That's a good time for Milford to call and say "Hey, mom, I'll go out and do a mile today if you do a mile today."

PAT GELSINGER: Are there other examples, other things that would be different in the digital home of the future?

ERIC DISHMAN: In some ways it might look the same, but what's in the background is important. So we want people to be able to use the phone as they always have, but the phone should be smart enough to know through sensor networks if you're unconscious or if you've fallen and should dial itself to get you help.

Or the television. This is a major place where people spend lots of time. This should not only be a digital entertainment device, it should be a tool for prompting, coaching healthy behaviors. Even the chair is a valuable place to have information.

PAT GELSINGER: So the same sensors we've been talking about become part of the furniture, the fabric of our home.

ERIC DISHMAN: Exactly. We want to know your weight and track that over time but we also want to know is mom sitting in her favorite chair. And we've done experiments where we can know mom is home sitting in her chair and it's a magic kind of connectivity we don't actually have today.

PAT GELSINGER: What's Intel doing to make these visions come to reality?

ERIC DISHMAN: There's really kind of three main things, and the thirst is collaboration. Because this is a huge effort that would take many industries coming together to actually make this home health care push happen.

So the first thing we've done is we have launched a new initiative called CAST it's the Center for Aging Services Technology, and I'm actually currently the chair of that. And the goal here is to bring together big players from the health care industry, from the high technology industry, and actually from the aging services industry.

So we've got some people like GE, Honeywell, Motorola, you might have heard of a few of these, but also Merck, Bayer and other players.

PAT GELSINGER: So people we don't usually thinking about attending IDF become collaborators in the future.

ERIC DISHMAN: That's right. The second thing we're doing is funding and working with university grants and researchers all around the world because what we want is a thousand flowers to bloom. The more people looking at this digital home infrastructure for health, the better.

The third is working with agencies and associations, whether they be government or private agencies and associations. And the challenge here is actually to name the big three. The big three for this age wave are cancer, cardiovascular conditions, and cognitive decline. Those three alone, the cost of dealing with those in just the United States is a half a trillion dollars a year.

PAT GELSINGER: So that's almost half of the total cost of our health care in those three areas.

ERIC DISHMAN: That's right. And this is partly why we've chosen these. Because if we can work with these three and actually use digital home technologies to do early home detection or do treatment, you are going to have huge cost savings.

The first we have worked with is the Alzheimer's association and it was great to get this quote from their CEO. And they're excited because for the first time they are going to start spending some of their research dollars to do digital research to improve the lives of people with Alzheimer's.

All that is to build collaboration. The second thing we are doing and you mentioned at the opening I am a social scientists. We have a team of social scientists actually studying hundreds of households around the world, studying both today's boomer households and elder households to figure out not only what one generation ahead technologies have to be, but what one generation ahead users will actually need for those technologies.

PAT GELSINGER: Now I know why a social scientist is at Intel.

ERIC DISHMAN: That's right. How our silicon can make the biggest and best impact in these households.

PAT GELSINGER: What are we finding?

ERIC DISHMAN: I want to show you two photos from the Alzheimer's work, which is the most challenging work that I have ever done because it is such a devastating and costly disease.

PAT GELSINGER: These are real photos from real people?

ERIC DISHMAN: And this is common in Alzheimer's households, where they have plastered the whole home with notes reminding the person with Alzheimer's how to lock the door at night because they don't actually remember who is a stranger and who is not or to actually remember to take a tub bath because, you know, it's dangerous for them to do a shower.

Those kinds of, if you will, smart home sticky notes are already around. And it's a ripe environment for the kinds of contextually aware technologies we have been looking at.

The last thing, if we built the collaboration f we have studied real people around the world and understood their needs, we actually have to start building tomorrow's technologies today. These are photos from our lab where we have the smart home lab and the sensor networks that you referred to where we're literally having the phone, the television, the PDA, everything is connected and every point can be a touch point for people to interact with. And, in fact, I brought one of those technologies for contextual awareness today. I went one forward too many.

PAT GELSINGER: What we are doing is I have been part of a context aware demonstration for my whole keynote so far. If we bring up the animation we've actually been tracking my location throughout the keynote this morning.

ERIC DISHMAN: There you are.

PAT GELSINGER: Here I am, walking around the stage here. So far, I have traveled nine miles this morning. It was pretty exciting practice session and you can see the system is following me as I move, you know, this will be, like, from one room to another throughout the home, right. It's able to follow my location and be able to monitor my activity how far have I traveled, how much activity have I done throughout the home that morning, that day, which rooms and paths have you followed throughout the household. What you see is it's able to locate me and precisely monitor my behavior.

ERIC DISHMAN: In fact, what we want to be able to do is know not only what room are you in and what distance you have traveled, but which device is closest to you, much like the manufacturing scenario where you could use whatever screen. A medication reminder or any kind of information can follow you around and you don't have to rely on having your PDA with you. Pat, you actually need to take your allergy pill. I wanted to remind you to take that.

PAT GELSINGER: Reminding. That's a huge piece of health care, establishing compliance.

ERIC DISHMAN: That's exactly right. Most of the people in this age group are taking 10 medications a day. That's hard to do for anybody, let alone someone who is starting to have some sort of cognitive failure.

PAT GELSINGER: Through this, we have seen this is a tremendous amount of computation required to enable this. We had four infrared cameras as part of this. We're gathering 70 megabits per second of information. We're processing that, taxing our fastest processors today with stereo capability.

And with this, you can see just the computational requirements just to follow context, not necessarily to deal with all the statistical implications that would come from that data.

You know, so what are the future capabilities, what kind of things do you think we will be sensing next?

ERIC DISHMAN: On top of this basic platform for doing context awareness, we're actually looking at next generation capabilities like gait analysis, actually being able to look at your walk and average stride length may be an early indicator of a disease like Alzheimer's or Parkinson's. We are also trying to look at technology usage. If mom has been using the oven for 15 years and we know it takes her ten seconds to operate and suddenly she's standing there for ten minutes, that's a window on a possible cognitive problem she's having that may help us know of a problem years before today's diagnosis can actually catch that.

The final is actually kind of a pathways analysis, trying to figure out, again, if mom typically goes from the kitchen or from the bedroom to the kitchen for breakfast in the morning but is suddenly wandering around different parts of the home, again, that may be an early warning where we can catch things and address them early on.

PAT GELSINGER: What are the key technologies to make this happen?

ERIC DISHMAN: The first is communications. So all of the things that you talked about with Radio Free Intel we assume as a given. We need wireless, robust, secure networking, and everything is on the network, including the sensors embedded in the chairs.

The second is actually interfaces. So making the television, the telephone, the lamp, but also PCs and PDAs intuitive and easy to use interfaces. And also the sensors as the interface between the human body and the environment to know what's going on contextually within the system.

The last is computation. You mentioned Bayesian networks before were needing all kinds of statistical approaches that we're pulling out of Intel Labs and using to turn all of this data that nobody wants to get this data into any useful information, we need inference engine to power all of that.

PAT GELSINGER: This is very, very exciting, new users, new uses, the use of sensing, computation, the stunning amount of MIPs this is going to require.

One of the things that troubles me and nags me about this, Eric, is, you know, what about the privacy implications? I mean, giving up all of that personal information, aren't you concerned about this intruding on people's privacy?

ERIC DISHMAN: No. You have to understand that this technology is about them regaining and maintaining the privacy of their own home. I mean, you look at the photos of the people that I showed you, you are talking about somebody who may be forced to move out of their home in 50 years and to go into an apartment or who may be suffering through a disease that could have been caught with a sensor network earlier in their life.

These technologies are about giving them back privacy, choice, and autonomy to live independently from wherever they want. And that's an exciting project to be working on.

PAT GELSINGER: Wow. So what you have seen, ladies and gentlemen, this morning is an example of macro convergence in health care and how we can take these underlying technologies, micro convergence technologies and potentially enable macro convergence for home, health care of the future. Eric, thank you very much.

ERIC DISHMAN: Thanks, Pat.

(Applause.)

PAT GELSINGER: I'm very excited about this work. It just sends tingles up my spine as I consider how we can be changing society in the future. I was pondering, is there more? Can we go further, even deeper into convergence?

In fact, that's exactly what we're working on and we're pursuing with precision biology research. And what I'd like to do now is have you join me in welcoming Dr. Andy Berlin, the director of biotechnology research at Intel.

(Applause.)

ANDY BERLIN: Good morning, Pat.

PAT GELSINGER: Just like I asked Eric, what's a social scientist doing at Intel, what's a biology researcher doing at Intel?

ANDY BERLIN: Well, I'm here, and my team is here because our precision biology labs here in Santa Clara are taking convergence to a new level. What we're doing is taking Intel's nanotechnologies, our industry's technologies, and combining them with biology and medicine to make it possible to use chips in new ways.

PAT GELSINGER: So, you know, the nanotechnology stuff that Sunlin Chou and I were talking about at the last IDF, you are using that in biology?

ANDY BERLIN: That's right. We're taking Intel's nanotechnologies and working with the medical research community to make it possible to diagnose disease and improve people's health by using chips.

Let me give you a few examples. You can imagine taking that distributed sensor scenario we just saw over there and take it one step further. Let's look at how this might evolve.

So imagine supposing that people are wearing the sensors. Maybe they are embedded in our clothing, our socks. When you were out walking your nine miles this morning I think I did about five of those with you your socks can tell you if are you about to get a blister.

You can take that one step further. There's work that we are involved with at the University of Rochester at the Center for Future Health that actually is building a bathroom mirror that looks back at you. So you can have your mirror watching your face and your body and telling you, for instance, if you've got an early onset of skin cancer, doing early disease detection.

As you start to bring Intel's nanotechnologies into the picture, that's where our lab comes in. And that if we can build a smart Band Aid.

PAT GELSINGER: So "Intel Inside" Band Aids?

ANDY BERLIN: Yes, that's right. Our Intel outside, depending on how you think about it.

PAT GELSINGER: So, Andy, can you be a little bit more specific? You know, what does Intel really have to bring to this?

ANDY BERLIN: We're actually thinking that you can take Intel silicon and our optical devices, as you look around, almost everything that Intel builds is actually smaller, substantially smaller, than the biological molecules that are of medical interest.

PAT GELSINGER: So our devices are smaller than all the biological structures that you would really like to research?

ANDY BERLIN: That's right. They're smaller than cells, comparable in size scale to individual molecules of DNA, smaller than most proteins. And we're building these things by the billions.

PAT GELSINGER: Uh huh.

ANDY BERLIN: Imagine if we could use arrays of them to, for instance, if a biological molecule comes by, the behavior of almost everything we build changes. So you can start to use these as kind of little voltmeters, little probes that reach out and touch these molecules and tell us something about them.

PAT GELSINGER: So we're building billions of little precision biology amp meters. Can you be a bit more specific? Silicon technology useful for biology?

ANDY BERLIN: Well, silicon technology, both our electronic devices and our optical devices. It's part of the reason that Intel is kind of the best place in the world to do this kind of research. Because it's literally like being in a huge high tech shopping mall. So you can go down the hall and there's the laser lab. And if you need a particular thing done to your molecules, particular kind of excitation, there are people who will just do it for you. And you can go down the other hall and there's people who know how to take silicon and do all sorts of magical things to it.

Let me give you one example of our multi stage shopping trip. We stopped at one of your technologies, actually. Do you recognize this slide?

PAT GELSINGER: Sure, I do. This is our FIB technology, focused ion beam. Craig mentioned it on Tuesday.

We first did this on the 386 chip. We didn't want to have to tape out the whole chip, so we just edited the FIB. So we went and FIBed the mask and went to the 486 and said, as we were trying to do some debugging, "We'll just edit the chip itself." We edited the wafer, cut some lines.

We had a problem with the Pentium Pro, because we went to the FIB chip, couldn't actually probe the lines anymore. We sort of used the FIB to bore through the back side of the silicon and get access to the transistors from the back side as well. So what are you doing with my FIB tool, anyway?

ANDY BERLIN: Well, what your FIB tool did, and particularly the piece that had to do with getting it from the back side, is it gave us the capability to sculpt silicon into any shape we needed very quickly. I can wander down the hall and say I need silicon that looks exactly like this so I can use it to move a piece of blood around on this chip.

And, you know, my friend Mike will just go off and go build these things. And it's great. So we actually took that same technology you had and used it to construct Intel's first microfluidic chips where we're moving around biological samples, fluids, on chips.

PAT GELSINGER: So what are these kind of channels used for?

ANDY BERLIN: Well, we want to use them ultimately to get to the point where we can kind of line molecules up almost single file and kind of march them past our devices, both our optical devices and our transistors, and use them to figure out what's going by, count them up, and figure things out about people's health.

Let me give you an example. What's shown here is individual molecules, single molecules, of DNA, moving through a piece of Intel silicon here in Santa Clara.

PAT GELSINGER: So what I see, the white dots here are literally DNA moving through Intel silicon channels.

ANDY BERLIN: That's right. They're individual DNA molecules attached to little marker to make them visible.

PAT GELSINGER: So, what else can you do with them?

ANDY BERLIN: Well, what you see here is not terribly well controlled. We're just showing that we can actually flow fluid through the channels. We're actually taking that to the next level and creating a set of building blocks for both manipulating these molecules and figuring out what they are. What can we figure out about them, what's the DNA sequence, tell us something about the protein content of someone's blood.

Let me give you another example. This is showing a single molecule of DNA attached to a silicon chip. And what we're going to do is come in with a laser beam. I told you I stopped in multiple stores in your mall, and what we can do is grab ahold of the end of that piece of DNA with the laser, stretch it out and then let it go and slingshot it across the chip.

So we've actually here at Intel using our technologies built the smallest slingshot. It's a single molecule slingshot for DNA, and you see the bottom beam moving on the ship, and when the slingshot gives away, we slingshot a single molecule of DNA.

PAT GELSINGER: I'm paying you to build the world's smallest slingshot?

ANDY BERLIN: And it's the best job at Intel.

PAT GELSINGER: How is this useful? How is this work producing results that the industry can use?

ANDY BERLIN: This is just the first step. This is getting to the point where we can separate the molecules and move them around, pump them around on chips in new ways.

Our work, this particular building block which is the first of several you're going to see coming out from Intel over the next year just gets to the point we can move the individual molecules in three dimensions to where we need them to be.

You're going to see further results coming out as we work with medical researchers to take this to the next level as well and start to actually tell what you those molecules are and start to say things about people's health.

PAT GELSINGER: That's excellent. Breakthrough results in how we bring these together. What's the goal, Andy?

ANDY BERLIN: Well, my dream, never mind the goal of the project, my real dream here, the thing that gets me to work every day is I'd really like to find a way to build a chip that can analyze someone's blood and tell them two, three, four years earlier than is possible today that they've got the early onset of a disease like cancer at a stage when it's really treatable.

That's kind of my dream that I really want to do.

The actual official goal of the project is something far more business oriented and essentially we'd like to find a way to turn health care, which is the largest segment of the economy, into a chip based industry.

PAT GELSINGER: So when we launched our extending and expanding Moore's Law, we thought about fluidics and so on like that, but moving it into biology is really stunning. And what you've seen is convergence at a whole new level. Convergence that brings together technologies, in this case lasing, sensing, fluidics at a whole new level, potentially entirely restructuring the health care industry of the future, the single largest segment of our entire economy. Andy, this is just thrilling work. Thank you so much for joining us today.

ANDY BERLIN: Thank you.

(Applause).

PAT GELSINGER: What I hope we've shown today is convergence in a whole new way and in a whole new light.

We've talked about micro convergence, bringing together these building blocks of computing, of communication, and interfaces, and how our research work at Intel is enabling micro convergence, extending and expanding Moore's Law to a level that I certainly don't think most of you ever considered possible.

Even more importantly and more exciting is the potential that that will have for us as an industry, when we realize collectively the macro convergence that will result from macro convergence at these levels, the new usage models, the new infrastructure, business models, and the impacts and changes on society.

Thank you very much for our time today, and I enjoy the time that we spend. Most importantly, I invite you to work closely, join with us in enabling macro convergence to a whole new level.

Thank you very much.

(Applause)