Workstation Leadership Forum 99

Paul Otellini
Burlingame, Calif., USA
July 1, 1999

PAUL OTELLINI: Good morning. And welcome to day two of the Workstation Leadership Forum.

You know, I think that it's very impossible, darned near impossible nowadays, to pick up a newspaper and not read about yet another industry or yet another company being disrupted or changed by the Internet. At the same time, you read about six or seven more that are being born in the Internet.

And it is important sometimes to go beyond looking at the Internet as a force of nature and remember that it's ultimately a technology. And as a technology, it is subject to what Clayton Christensen at Harvard describes as the law of disruptive technologies.

Now, Dr. Christensen's theory is very interesting. He's looked at a number of industries over the last 50 to 100 years. In each of these industries, he plots improvement in their given core technology along a given axis.

And that improvement in technology is, by definition, equal to what the industry needs to maintain the mainstream of its customer base.

But then he went further and he looked at industries that had been displaced. And time after time, found that in all of these industries, there was a common pattern. And the common pattern was that there was a new technology that emerged, typically well under the radar screens or well under the profit screens of the entrenched companies, but was moving at a much faster rate of technology change.

And he calls this disruptive technology.

In industry after industry, the disruptive technology displaced the entrenched ones.

It started with Henry Ford's motor company that employed mass production techniques to essentially displace all the hand-built boutique manufacturers before it.

He looked at more modern examples like the mini mills in the steel industry that completely displaced all the entrenched high capital participants by starting out building rebar and moving up into sheet stainless steel, the highest margin product in the steel industry, operating at a much lower capital cost, much higher turn rate, reusing technology and product.

And I think in our own industry that is, computers, we've seen this in the last 15 or 20 years with the evolution of the personal computer disrupting other products before it, dedicated word processors or typewriters before those. In industry after industry.

And what I wanted to evolve today is a story about the disruptive technology of the Internet and how it's going to impact all of your businesses, all of our businesses going forward.

A few years ago, Intel began talking about the Internet and saying that it would change everything, it would change the way we work, the way we play, and the way we live.

And a few years ago, that was a stretch. Most people kind of shook their heads in disbelief. And it's amazing now, over that intervening period, it's now an accepted cliché that the Internet is, indeed, changing everything, and in every aspect.

But our vision throughout this period really has remained the same. Intel is focused on helping build a world of a billion connected computers, served by tens of millions of servers, generating trillions of dollars of E-commerce in the not-too-distant future.

But as we look at this vision, sometimes I think it's hard to remember where we came from.

50 years ago, the ENIAC was the state-of-the-art computer. It weighed over 30 tons and it filled the room the size of a tennis court.

25 years ago, the state-of-the-art computer was the Altair. It was based upon the Intel 8080. It ran at two Megahertz, cost $400 million and had 256 bytes of RAM. But the product I think all of us became most familiar with in terms of computing was the IBM PC in 1981, based upon the Intel 8088, this was a 64,000 transistor product running at 4.77 MHz, as compared to the nine and a half million transistor Pentium® III Xeons™ that are powering the workstations we're going to walk you through this morning, running at 550 MHz.

I think that along the way, along the 50 years from the ENIAC to today's computers, what's really happened is that computing and computers has become ubiquitous. And in the process of creating that ubiquity, it also created the infrastructure for the Internet to exist and begin changing our lives.

Now, because the Internet was based upon the infrastructure, we have the benefit today of having over 90 percent of all of the Internet access devices in the world being based upon our architecture. And, more importantly, from our view, going forward, all of the servers, 85 percent of all the servers that ship in any given year today, are now also based upon Intel Architecture, many running on NT.

Most of these servers are now shipping into Internet and intranet applications year in and year out, continuing to build this infrastructure towards the millions of servers for the E-commerce future that we all see.

As the infrastructure gets deployed around the world and, more importantly, inside of your companies, it's enabling new classes of work. I think that Craig and Bill talked yesterday about collaboration. I'd like to talk a little bit deeper about collaboration in some of the industries that traditionally use workstations and show how the Internet is impacting those industries.

In electronic design automation, typically an island of development inside a company, probably the most dramatic change that's happened in some time is that there's now companies out there creating Internet design centers, virtual design centers on the Web. One of them is called OrCAD, which was acquired by Cadence last week. But these companies are out there to harness the ability of people to work outside of a fixed set of buildings or cubicles to help collaborate and participate in a broader design environment.

So you've seen the similar activity in digital content creation and mechanical design. Yesterday, you saw the example with Dassault and Chrysler, where you designed something once, you developed the models once, but you can reuse it many times to not just build products, but also to produce the advertising and merchandising for your corporation.

Collaboration goes beyond working people inside your company. Increasingly, it is important to work with your suppliers and with your customers going forward.

Now, the flip side of all of this is that the Internet also creates a channel for your businesses, not just your business, but for your customer's business. Because it creates a very ubiquitous channel, it means that competition for you and for all of us is nothing more than a click away.

Now, that implies that you have to move faster going forward.

Our view is that, as the enterprise is developed and we deploy computer technology, productivity has increased.

And, in fact, Alan Greenspan gave this speech two or three weeks ago where he warned the American economy would not benefit at the same rate of change from the deployment of computer productivity as it has in the last few years.

With all respect to Dr. Greenspan, I think he was dead wrong. I don't think he understood the multiplicative effect of what happens when Moore's law meets Metcalf's law. And when you start getting the compound effect of the networks being deployed, you move from simple productivity to organizational efficiency. And that's really what I want to talk about as I go through the various demos here.

But rather than have me tell about you this, let's take a look at what some of our customers using Intel Architecture and NT have to say about it.

VIDEO: The transmission in traditional based workstations to Intel-based workstations was really easier than I expected it to be. We were able to get more collaboration between our facilities, we were able to run multi-tasking applications, and we were able to (inaudible) in particular as a multi-threaded application -- applications which we were never able to do before.

When we opened up Blur, there was no question whatsoever what we were going to do. We were going to be Intel-based workstations from the ground up. Everything we want to use, every piece of software we want to use, every piece of hardware we want to use is on Intel-based workstations.

1998, Intel was the leading platform that we delivered for our new customers, at least in America, so I think that speaks for itself, the capability, performance, and price affordability has really come into its own in our space. One of the nice things we see about Intel-based workstations is they're scalable, so you can enter the market, whether you're in your home or whether in your business, with the type of technology, the speed, and push that meets your needs. And you're not locked into a single environment.

One of the avenues we're beginning to explore using Intel-based workstations, which we would never explore with the traditional workstation, is virtual prototyping. This gives us the ability to test designs, test machine that saves us in the long run. We don't have to cut any metal. We don't have to go through a build-test procedure, multiple iterations, build, test, build, test… Based on the new workstation processors, Pentium III Xeon, they're so fast, stuff that took days to render before now renders in minutes.

Historically, graphic intensive applications were measured by how many coffee breaks an engineer might take. Because as soon as you hit that button, you needed to take a coffee break. Now if they are taking a coffee break generally, they're out there to brag about the design they just completed because the that is so much improved.

PAUL OTELLINI: I think what those companies are telling us is that keeping pace is not just a function of collaboration. It has to go beyond that. And the expectations of our companies, of our employees, of our product cycles really are increasing every day in terms of time to market or demands. And that increased expectation requires higher levels of performance in all aspects of computing and in productivity.

Perhaps, historically, the highest performance requirements were in the supercomputer industry. NCSA is a company that was established in 1985. And their purpose in life is to develop a facility for researchers to do applications work on some of the big problems of the cosmology, weather forecasting, weapons research. Those types of things have only been done, traditionally, on supercomputers. But they have been terribly expensive in the past and not particularly scalable. You bought one, you used it until you bought another one.

What NCSA has done is they've taken and clustered together a series of dual Pentium II and III Xeon processor machines from a number of vendors to create, effectively, the world's largest supercomputer. The one they have running today has 160 dual processing workstations clustered together. And by the end of next year, they intend to double the capacity of that, running over 512 microprocessors inside of this, to do more and more of the problem-solving that they're focused on.

They selected Intel Architecture and NT not just because it was cost-effective, but because it was scalable. They're able to scale this thing up sequentially over time, they found it to be very stable and very high performance in the most performance-sensitivity computing environment in the world.

Moving farther down the spectrum, though, there's a number of industries where supercomputing modeling can help but has been prohibitively expensive, industries like those that are involved in safety and simulation for planes and automobile crashes and so forth. There's a company that's a spinout of Lawrence Livermore, called LSTC, which is the Livermore Software Technology Center. And what they do is focus on finite element analysis. And they now, with Intel Architecture and NT, can do unique modeling and simulations of crashes for automotive companies over time.

Now, the automotive industry, up to the point that it had this technology, was doing over 2500 real car crashes with real dummies every month. The estimated savings from LSTC moving to the simulation models has not only increased productivity, but over $350 million a year of savings for the automotive industry.

And LSTC, which began not on Intel Architecture, is now seeing that over 75 percent of all of its new business is based on IA and NT.

What I think both of these examples demonstrate is the value of standards. Industries which in the past couldn't afford supercomputing capabilities or couldn't afford the high-end modeling capabilities or the virtual prototyping capabilities can now do that real time in a cost-effective fashion. And that's simply because the economics of scale of the PC industry are being brought to bear through workstations on the kind of problems that were only done at the very high end or too unaffordable to do before.

We've all seen this before. It's economies of scale drive value. That value drives the volume target for the software industry, the software industry writes increasingly to that target, benefiting the entire ecosystem and creating more customer value over time, essentially becoming a disruptive force relative to other architectures that are out there. And what we're seeing in industry after industry that use workstations is end users are picking up on this and it's converging.

This looks at mechanical design, digital content creation, electronic design automation and finance. And in all cases, the red parts of the bars are Intel Architecture on NT, and the purple parts of the bar are the rest of the Unix vendors that are out there.

And you can see that the rate of growth in every industry and the majority of the volume in every industry is effectively now or about to move very quickly to Intel Architecture on NT.

Users understand the economics and the dynamics of this business.

I think that, going forward, this class of workstations gives your company, gives your environment, gives your employees, a productivity advantage. We've always focused on performance. We now have the leading performance in workstation. We've always focused on having the best price-performance. We're maintaining the price-performance advantage in the industry.

But it's not just in processors that we've focused. We also tend to focus on improving the entire compute platform.

By the fourth quarter of this year, the entire platform will be refreshed and you can buy workstations from a number of OEMs, essentially, giving you 4X the capacity in memory bandwidth, 2X the capacity in graphics, and AGP Pro capability for greatly enhanced graphics capabilities.

But more importantly than the hardware is, increasingly, we see a choice of software moving over to the architecture running on NT.

Over 95 percent of all digital content creation applications are now running on Intel Architecture. In the last year, over 50 percent growth in new software packages in electronic design automation to over 350 being deployed running on Intel Architecture. And in mechanical design, vendor after vendor is now seeing above 60 percent of all of its new licenses or all of its new sales all going out on Intel Architecture.

The software community understands what the hardware community is doing and is changing its development habits to take advantage of this.

I'd like to give you an example of what is now being done on Intel Architecture in the area of digital content creation.

Let's take a look at this. (Video.)

PAUL OTELLINI: Now, those spectacular clips were created by a company -- technology by a company called Avid Technology. And many of you may not know who Avid is. But for people in the film and entertainment industry, Avid's a very important company. They represent over 85 percent of all the prime time television shows, most of the movies, and much of the music that we all are entertained by.

And to tell you a little bit more about the technology that you just saw, I'd like to introduce Bill Miller, who's the CEO of Avid. (Applause.)

BILL MILLER: Good morning, Paul.

PAUL OTELLINI: How are you?

BILL MILLER: Great. Good to see you.

PAUL OTELLINI: That is spectacular technology.

BILL MILLER: It's pretty amazing what they do with the systems.

PAUL OTELLINI: I've never met someone who has won an Oscar before. What's it feel like.

BILL MILLER: A rare treat. Unfortunately, I can't claim any personal credit. But the guys at Avid have done a sensational job of building this technology.

PAUL OTELLINI: That's obvious. How does the technology work?

BILL MILLER: We started with the simple editing application, bringing computer technology to the editing process to improve productivity and creative options. We're now going way beyond that. And I'd like to get an opportunity to show you just a little bit of that if we step back here.

PAUL OTELLINI: Absolutely.

BILL MILLER: We'll go back here. This product is our product we call Symphony. We're going to show you version 2.0. Which is going to be released pretty shortly here.

And I'd like to introduce you to Steve Tomitch. He's going to help us through the demonstration.

We'll give you a flavor for what we're doing now to take this technology even farther. We are the disruptive technology in this business.

So, Steve, if you can just take us through a little bit of the editing stuff and then we'll move on to the finishing things.

STEVE TOMItCH: This is the basic starting point, this interface. This is pretty much the industry standard interface. It really did revolutionize the whole game. On the far left monitor you can see over there, I have my bin for my raw material. We'll be taking shots from there and moving them to my editing window and assembling shows. It's as easy as dragging and dropping some material.

I can pull them over there and play them back instantly. I have instant access to my material, lots of changing capabilities. When I'm working, I'm working at the highest quality picture, full bandwidth, broadcast-quality picture, and two streams of it. So if I wanted to, let's say, do a realtime dissolve, I can just go ahead and queue that up. And hit "play" and you'll see two streams flowing through realtime dissolve between those shots.

Okay. That really did revolutionize the --

BILL MILLER: This is the way it began by putting the pictures together. And then you take the instructions from one of our systems, go to a much more expensive room to change the colors, to do the finishing touches that are all required.

With the advances in the platform technology and improvement in our software, we can now replace all that. We can deliver a system, this one, Symphony 2.0, that will enable you to finish a whole network television program right here on this system.

There's a lot of features to that. Steve's only just going to show you a couple of quick ones. But it's really a powerful capability.

Steve, go ahead.

STEVE TOMITCH: One of our new tools is color correction, an engine built into Symphony. Let me show you a couple of shots that are going to show a little color problem. Let's play the first shot.

So we have bicyclist against a blue sky and against a green sky. The green is not going to work. What I want to do is go into the color correction engine, bring that up and I can see in context all of my shots, all my material. I've got these tools down here that allow me a lot of different ways to manipulate the color space. And importantly, there's a one-touch tool, too, called Natural Map. So what I can do is I can say this is my bad color, this green up here. And this is my good color, the blue up here. I can just say replace the green with the blue, hit match color, and voila.

It's much closer match now.

BILL MILLER: It changes every frame.

STEVE TOMITCH: Yes, all in one fell swoop.

The other thing, color correction, also another thing important in finishing, which is the ability to do multiple versions. Let me play a snippet for you here. This is a trailer from a feature film. And we'll queue it up just right and hit "play."

(Video playing.)

STEVE TOMITCH: Real quick, just a quick example there. IT'S a lot of star power, high production value, great-looking picture right here and sound. That's only one version. That's the wide screen version. Normally, to make HD version, the film version, the broadcast version, the cable version, those are all different processes that happen out the editorial process and they're expensive and time consuming. We bring a lot of that into Symphony. Right now I have queued up my wide screen version. I can with a click of a button look at my broadcast version and a broadcast aspect ratio. I've produced that, it's ready to go, and I've collapsed that whole process right down here. .

PAUL OTELLINI: Amazing. What does something like this cost?

BILL MILLER: Something a little north of a hundred thousand dollars, which in this business is a pretty big cost saving, because it replaces maybe half a million to a million dollars worth of equipment.

PAUL OTELLINI: That's great. If you're making movies, it makes a lot of sense. Not all of us have movie production in the mind, and not all of us want to do mainstream kinds of things.

A lot of companies want video for Web broadcasting or employee training or those kinds of things. Do you have tools that can meet the needs of smaller companies?

BILL MILLER: Absolutely. As the Internet gets increasing bandwidth and corporate networks get the capacity to move video around, we think media is going to be an extremely important part of corporate education, government communications. And we're working with the Intel Architecture to Intel an application which will ship this fall that's ideally suited for that application.

If we step over here, I can show that one to you as well.

This is an application we call Show Biz Producer, it'll sell on a turnkey basis all in for under $10,000 on an IBM Intellistation. And it's designed especially for corporations and government agencies and education that want to use media to communicate their message across networks or on CD-ROMs. We've got an interesting example, a realworld example here from an outfit called Durkee Foods.

They had a problem that plant tours in their Marshmallow Fluff plant, believe it or not, were disrupting operations, so they got an idea to create a virtual tool they could put up on the Web that would enable people to see how they made the Marshmallow Fluff.

And they executed that, and we've got that right here.

I'd like to introduce you to Allen Baldwin. He's going to tell you a little bit about how they put it together real quickly.

PAUL OTELLINI: Go ahead, Allen.

ALLEN BALDWIN: As Bill mentioned, they needed a way to get video to the Web. So what they did was they took a very affordable DV camera, camcorder, spent a couple of hours videotaping the process. And you can see some of the clips playing here. And Show Biz Producer has DV hardware as part of that turnkey system. So the 1394 standard can be fed directly into Show Biz and you can see in the upper right here, this is a bin. A bin is a place to store individual clips. And I can double click on any one of these. And there's a nice, tasty shot.

PAUL OTELLINI: Looks great.

ALLEN BALDWIN: I can scan through very quickly. I can play the footage, I can see what's there.

Now, I'll show you very quickly how you can pull together all this and actually put together a nice video project.

I start with a clean slate. Empty time line, empty player. If I wanted to, for example, see the top two rows and how they played together, I'll just lasso those, drag them, drop them. You see how quickly it builds an actual cut that I can then hit play and take a look at.

Show Biz Producer lets you add effects like dissolves, wipes, graphics, you can import graphics from other applications. And I'll show you the final --

PAUL OTELLINI: So you can do all this work in-house now.

ALLEN BALDWIN: Absolutely. The final actually looks like this. And you see the use of graphics. You can add CD music.

But we don't stop here.

If you remember, as Bill mentioned, the original intent is to get this to the Web site. And if I pop over to Explorer, and see, here's the Marshmallow Fluff site. And if we wanted to actually go back, if we really wanted to know what a Fluffernutter was, we could take the virtual tour.

(Video.)

ALLEN BALDWIN: It's the same movie that was created in producer. And the key here is that Producer can export in a variety of digital formats, including streaming media. So mission accomplished with Show Biz Producer.

PAUL OTELLINI: That's great. We'll have to go measure the hits on Fluffernutter. This is truly spectacular technology. Thanks.

BILL MILLER: We're excited about it. We think media is going to be the next big thing on the Internet. We're looking forward to providing the tools. And with the broad range of capability that's available on Intel Architecture, we can move all the way from your next Hollywood production when you decide to give up the chip business, Paul, down to serving corporations.

PAUL OTELLINI: I think we'll do a bunny suit tour of the fab.

BILL MILLER: That's terrific.

PAUL OTELLINI: Thank you very much. (Applause.)

PAUL OTELLINI: The picture I tried to paint for you so far is kind of two forces merging together. One is this convergence in the workstation arena based on Intel Architecture and NT. And the other is the force of the Internet. And the combination of those two forces really can be an opportunity to unleash the power of the platform here to do new and better things.

Let me give you an example of that.

In the world of the Internet, there's clearly an explosion of data going on. The average size of a decision support database is growing from over a terabyte by a factor of 24, well over the factor of six terabytes over the next few years. The average number of people that are accessing decision support software databases is growing by a factor of 42 to almost 100,000 in the next year. This means that you've got increasingly large amounts of time accessing increasingly large amounts of data. And it generates a need for analytical tools to be able to take advantage of that data.

The promise of the Internet really is all about push or focused marketing.

The reason that companies like Amazon are so successful isn't just that they are discount and quick, available books. But they also can tailor their marketing precisely to your needs. They know what kind of books and CDs you buy, they use that data to market outbound back to you.

Much of the Internet valuation of companies today happens because of that promise of outbound marketing. The promise of outbound marketing is only delivered with analytical tools going into databases that have collected all kinds of data about where you've been, where you travel, what you're like, et cetera, et cetera.

And I think that, going forward, looking at this data in kind of the classic fashion, rows and Cs of numbers is going to be increasingly impossible as the databases explode. And it generates the need for data visualization, generates the need for lower-cost data mining, all of which is now much more readily available on machines like the ones I've shown you today.

There's clearly some benefits to data visualization. You can do things smarter. You can take advantage of these databases that are deployed around the world, put them on your workers' desks not just in engineering, but also in marketing and finance, in purchasing, and many of the areas of your companies that are prone to analytical works.

And once you have this capability deployed, you have an instantaneous increase in productivity. And you get some other advantages, because, again, you can reuse this data not just in engineering, but also in management, potentially even in sales.

To give you an example of how some companies are doing that, I'd like to bring out Sam Abuzalaf, who is vice president, development, of Terrex Corporation, to show us what they're doing in this area. (Applause.)

PAUL OTELLINI: Hi, Sam.

SAM ABUZALAF: Hi, Paul. How are you?

PAUL OTELLINI: Good.

SAM ABUZALAF: Excuse me. A little hoarse this morning. My mother told me silence was golden, and I never really appreciated it.

PAUL OTELLINI: What does your company do?

SAM ABUZALAF: We're basically in multiple businesses as far as software development is concerned. But what we're going to show today is visualization of geospatial data.

PAUL OTELLINI: Okay.

SAM ABUZALAF: The kind of things we're going to show today five years ago was actually a price tag of $5 million, taking man years to accomplish. Because of the Intel Architecture/NT architecture we're able to accomplish the same tasks at below $5,000.

One of the local wine-makers here in the Napa Valley is Robert Mondavi, who is one of our customers.

PAUL OTELLINI: Know them well.

SAM ABUZALAF: Good wines.

But they use our technology to actually be able to visualize the vineyard that they have themselves. So in this case, they use our technology to visualize wine growth, early soil erosion, soil diseases, and that sort of thing.

And, actually, interesting parts, this is utilized by both the engineers and the scientists and the sales and marketing and those decision people as well.

Let's minimalize this and see how really easy this is to do.

Wine makers have been storing data in databases for many, many years. And what we do is we actually access all of this data directly either (inaudible) or out of databases. So in this case, what we have here is a 2D map. And in this map, what we're going to do -- by the way, this is how you organize all this data. It's visual. It's (inaudible) and click and unclick, whatever that you want to use or not use.

So in this case, we can go out and select on this vector line. So we actually can import data sources like elevation models, vectors, images, maps, all of the things that actually give you what you see on the screen here today.

As we select on this, this tells you it's a body of water, it's a lake. And for Napa Valley specific applications, as we're demonstrating. This is a merlot wine. It gives you growth factors, ambient temperatures.

PAUL OTELLINI: The database is acting realtime with the chip?

SAM ABUZALAF: Exactly. You're reading out of the database. You can add new data and send it back for editing later.

So, anyway, as do you this, the trick is how do you build this.

Well, we do this in realtime, in 3D, it requires more hand modeling, no programming, no scripting. You simply go through an interview process.

And the interview walks you through the different steps that you would need to do. And you simply answer the questions that are asked. And then when you get done, you simply go in and click on "build off." It's that easy.

PAUL OTELLINI: And it builds the information with the database realtime.

SAM ABUZALAF: Exactly. It takes the 2D data and represents it in a very unique and different way.

In this case, we incorporated the technology use of billboards. And these billboards are blank on purpose. And so what you actually have here is the virtual terrain, if you would. But it's not just about terrain. It's about --

PAUL OTELLINI: These are information Windows about the terrain?

SAM ABUZALAF: That's exactly right. But on top of the terrain, which is your most critical foundation, you are now showing in polygons different legends for the maps. This is merlot. This is cabernet.

The scientists might want to represent scientific data. The businessperson, as we also, might want to show this in financial output. And we will show that next.

Next thing do you is actually bring up the finished product. And this is the entire terrain scape, if you would.

This is really what we're all about. We're actually rendering in realtime, and we're paging in data in realtime.

And as we get closer to the actual vineyard itself, you will see some interesting methodologies that have been incorporated here.

Let's come down and look at this. And these billboards now can take data from any Windows document, engineering document, spreadsheet, attribute sheet of any number of commercial software components.

Let's come up here and take a look at. This I want to explain one other thing. Actually, two things. We are in multiple industries, flight simulation, urban development, defense commercial side of the house, wine making, transportation, utilities, and all sorts of things.

But I think as the GIS and the CAD markets come together, you're going to need a window that allows you to visualize these vast amounts of data. And there's really no better way to see it than in a three-dimensional perspective in realtime.

So to close this, basically, what we typically do is give you real cutesy things like going through a sunset.

PAUL OTELLINI: You just moved from production to sales.

SAM ABUZALAF: There you go.

PAUL OTELLINI: Okay. Thanks, Sam.

SAM ABUZALAF: Thank you very much.

PAUL OTELLINI: Bye-bye. (Applause.)

PAUL OTELLINI: So what I've tried to demonstrate in the last few minutes is that workstations are no longer just for engineering departments of your companies or architectural departments of your companies. They really are moving out into other parts of the corporation. Because workstations are nothing more than being all about speed and efficiency, which is really what most of us in the knowledge part of the industry, you know, are focused on in terms of workers.

You want people to work faster, you want them to work for effectively, have better access to data, and you want them to be able to do more and find more things out about that data.

These machines give you all of the advantages that you have with workstations, but the performance and the flexibility and the ability to communicate inside the enterprise seamlessly that you get with PCs.

The other way to look at this is it's not just for the large databases, the geophysical ones or the decision support software databases. Last week in the Wall Street Journal, there was an article that they titled the "Deluge of Data." And in that article, they looked at just the average data warehouse size in U.S. companies in '96 through '99. It was a very astounding trend there. What they discovered was that 68 percent of all of databases in 1996, data warehouses in 1996 were less than a hundred gigabytes in size.

By 1999, over 50 percent of the databases were over 500 gigabytes in size. And if you take the average size of the database, '96 to '99, it grew by two orders of magnitude.

So the explosion of data is not just in the data mining. It's not just in the geophysical. It's not just in the people who are doing decision support software. It really is in every aspect of every company, because we're creating these databases for every element of our work.

Now, one promise of the Internet that really is being delivered is that it delivers data. It delivers data to your company, it delivers data to your employees' desktop.

And in a very simplistic model, data enables businesses to make better, faster, more timely realtime decisions going forward.

But increasingly, data needs analysis. And the industry is enabling this kind of analysis by developing new classes of software that weren't available to knowledge workers before because it was just too expensive, the way simulations were too expensive for many companies before in the engineering environment.

To give you an example of another company doing innovative work in this area, I'd like to bring out Doug Cogswell, who is CEO of Visual Insights, which is a venture company under the Lucent Corporation.

Doug, good morning.

DOUG COGSWELL: Good morning, Paul.

PAUL OTELLINI: I've been talking about data visualization and how this can enable better decision-making and so forth. I understand your company has products in this area.

DOUG COGSWELL: Exactly. Visual Insights is a Bell Labs spinoff. And we're helping customers cut through the mountains of data they're finding these days. Basically, we find insights to the data that lead to action that lead to better value in the areas of category and product analysis, customer relationship management and E-commerce.

In essence, we have I think a disruptive technology that provides a much more intuitive and powerful way to query and analyze data than ever before. And it leads to better decisions, faster and it can be done by more people. I have to say the IA/NT architecture has been critical to getting this out and much more widespread than ever before.

PAUL OTELLINI: Is there a chance we can take a look?

DOUG COGSWELL: Let's take a look at the demo. This is Ken Fritz, our vice president of business development.

Ken, what do you have up here?

KEN FRITZ: Basically, what I've done is selected about 50,000 transactions out of our 250 gigabyte data warehouse and using our Advisor product to take a look at it.

PAUL OTELLINI: It looks like you've got data from a grocery chain there.

KEN FRITZ: Absolutely right. This is the type of data would you expect to analyze in conjunction perhaps with card information when you go to the grocery and they identify who you are.

DOUG COGSWELL: About 50,000 transactions in this database and the pie chart on the left has them aggregated into the top, middle, and bottom third customers. We've colored all the charts that way. On the left, the bar chart shows the products purchased, colored by the tier of customer. And the chart on the bottom shows the aggregate purchases across a 30-week period.

The top, the 3D view in the middle, shows the product purchases broken by customer tier.

I guess one question that would come up, Ken, is week 21, we've got the mouse pointed right there, looks like it was a bad week.

KEN FRITZ: It wasn't a bad week. It was one we anticipated. It was one that was a vacation week.

DOUG COGSWELL: Can we take a look at week 21 and 22. Those two weeks in the context of the whole, how about dropping the rest of the data so we're just seeing 21 and 22. Let's also go to the left and grab the top five or six product categories to make it simpler to see what's going on.

It's now looking at the sales of soda, milk, laundry, pizza, et cetera, across this two-week period. Let's look at just week 21, which was the problem week.

Ken, as I look at this and I go to the bar chart on the left, the product purchases, I see that pizza had a good week, and in fact was very heavy sales to the top and the middle-tier customers, the red and the green, but not the blue, the bottom tier. What happened?

KEN FRITZ: Well, anticipation of the vacation week, what we did is we did a promo to our top-tier customers.

DOUG COGSWELL: Can we now go down and look at the actual customers in the top tier who might have purchased it and see if this worked across the board.

We're doing a drill back and query against the database. And we've brought up how many customers?

KEN FRITZ: 2,000.

DOUG COGSWELL: Let's zero in on the top ten or 15 percent of customers and see what they look like and go back and select pizza only.

We see that the promotion does seem to have worked. A number of our large customers came in and bought $30, $40 worth of pizza in this week 21 period.

PAUL OTELLINI: Is there a way to do -- to go beyond looking at what happened and perhaps use this information to predict how to run a promotion?

DOUG COGSWELL: Sure. Let's take a look at our market basket analysis. In the chart on the top, let me explain for a second, the circles represent products purchased. And the bigger the circles and the hotter the color, the more the sales of that individual product. The lines between the circles represent the number of times they were bought together. The bigger the line and the hotter the color, the more times they were bought together.

For example, in this, if you, Ken, could grab private label whole milk on the bottom, pull it off to the side, you can see the affinities. That has strong connections to Oreo cookies, Chips Ahoy cookies, as well as skim milk.

Let's isolate on whole milk. This would be an example of if you want to bring people into the store and draw other products, you see this would be a good one to choose.

PAUL OTELLINI: That's great. I can see how this would be of great value to big companies and large chains and stuff. But small companies, I think, would also have the same kinds of needs for analysis and predictive behavior.

Do you have products that meet their needs?

DOUG COGSWELL: Good question. We have a product coming out in the fall called Advisor 2000 which is aimed at the Microsoft Office environment and will be an Excel plug-in.

What do you have here, Ken?

KEN FRITZ: Basically, a rather large Excel data sheet that has about 25 columns worth of data, 3,000 rows.

DOUG COGSWELL: If you were to say -- if I asked you the question of, for example, what are our three largest products and how are they doing in the three largest states, would that be easy to get?

KEN FRITZ: Ken not impossible, but not easy, either.

What I'm going to do is hit a button that we've inserted into the Office suite of products and actually visualize the data that's in that data sheet.

DOUG COGSWELL: The bar chart on the left shows the top coffees by type and the bar chart on the bottom shows the top states it sold in. Let's grab the top three coffees and the top three states and get rid of everything else.

PAUL OTELLINI: You're doing realtime analysis off that Excel spreadsheet.

DOUG COGSWELL: We're going back and querying against that and cutting it down. We'll quickly see the top three coffees are Colombian, lemon tea, and cafe mocha. The top states are California, New York, and Illinois.

Also, looking at the cafe mocha, the Illinois is that light blue, it's a heavily sold product in Illinois, but very little yellow, which is New York. So there's a different sales mix going on there.

Let's say we want to share this with the organization, so we're going to do a write-back to Excel and create a subset data sheet off just this cut of the data.

Now let's create a report.

Back into Excel. You've got your report. You can export this to Word, PowerPoint, and distribute this to the organization.

PAUL OTELLINI: That's fantastic. I think we'll be ordering about 10,000 of those.

DOUG COGSWELL: We have a limited availability beta program going on. You can go to our Web site, visualinsights.com, find the beta page, and we'll get you a copy of this.

PAUL OTELLINI: Thanks, Doug.

DOUG COGSWELL: Thank you. Appreciate the support. (Applause.)

PAUL OTELLINI: So what I tried to paint today is a picture of leadership in the workstation market segments, leadership from a hardware standpoint, leadership from a software standpoint, leadership from an operating system standpoint.

Intel is trying to do its part in leading this parade also. Over 50 percent of all of our investment inside of Intel Architecture this year is going into the high end, into workstations and servers, even though it's nowhere near 50 percent of our revenue today, we are committed to this industry, we're committed to the potential that these kinds of products have going forward, because the workstations in many ways are the precursors for the kinds of work that's going to be done on everyone's desktop or everyone's notebook in the not-so-distant future.

In the past, a lot of the innovations that are now on the desktop have shown up first in workstations. The kinds of analytical tools you're seeing today really are precursors to what all of us are going to have on our desk in a few years.

In addition to working, though, just in terms of driving the processors and the chipsets, we're also focused on driving new standards, things like WTX, which is a new chassis standard form factor to lower the cost and enable better flexibility, better thermal dissipation over time to handle the higher-performance systems. Things like AGP Pro which I described earlier.

Yesterday, Craig Barrett described the enterprise solution center, we're focused on driving application porting not just in the ISV community, but also in the proprietary application space to ensure that corporations can seamlessly move their applications over to our architecture and NT.

And then I think Craig also talked about IA-64. But I wanted to also remind you that we're taking not just IA-64 to the new level of performance, but we're going to continue to investment in the 32-bit side of the product line to ensure that that maintains the kinds of performance and capabilities you've become used to over the last few years.

All of this has generated a tremendous amount of industry momentum. Over 20 OEMs today are now shipping IA/NT workstations. And I think Craig told you yesterday, it's growing at over 20 percent a year in terms of a compound annual growth rate.

So the proposition, I think, that's changed in the last few years when looking at workstations is that when you looked at a workstation purchase, when you considered Intel Architecture and NT, it was just one of many choices that you had out there.

I think, increasingly, though, when you look at the capability of the hardware, the capability of the OS, the capability and the availability of the software that's now being written to these platforms, it's now become a no-compromise choice in your computing and workstation environment.

And if I were to summarize, coming back to Dr. Christensen's theory here, where you've got an Internet-enabled economy, an economy where competition in all fronts is moving faster and faster, what's important is that you as a corporation, you as a business, you as a department, grab your own version of disruptive technologies and be able to move faster than your competition.

From our perspective, workstations and tools based upon the technologies we've seen today really are the disruptive technologies to allow you to move faster than anyone that's in your way.

And if I give you -- if I leave you with one piece of advice from one who's been there, it's that when you're faced with the opportunity of a disruptive technology, it is far better to be the one who disrupts than to be the one who is disrupted.

Thank you, and enjoy the rest of the conference. (Applause.)

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