CA World 99

Craig Barrett
New Orleans, Louisiana, USA
July 29, 1999

CHARLES WANG: Good evening, everyone, and I hope you are enjoying CA World.

You know, tonight I have the pleasure of introducing Craig Barrett, the president and CEO of Intel Corporation.

Now, Craig joined Intel in 1974 as a technology manager, and was named the company's chief operating officer in 1993. He became Intel's board president in May 1997, and CEO in 1998.

Craig divides his time between his residences in Arizona and California. When he is not burning the midnight oil, though, at Intel, he enjoys vacationing with family and friends at his ranch in Montana. In fact, I am told he is quite a fly fisherman.

Also, he is very athletic and enjoys riding motorcycles off road. Craig actually wanted to borrow my bike to make his entrance, but I'm not going to divulge any of my secret acrobatic maneuvers to him.

But more than a successful businessman and outdoor enthusiast, Craig is a humanitarian. He is committed to improving public education. Through his dedication and perseverance, Intel has contributed more than $100 million to education.

Intel also sponsors the Intel Science Talent Search, often considered the, quote, "Nobel Prize competition" for U.S. high schools.

(Loud noise.)

CHARLES WANG: I told Craig you cannot borrow the motorcycle. (Laughter.)

CHARLES WANG: Craig was a Fulbright fellow at Danish Technical University in Denmark in 1972, and a NATO postdoctoral fellow at the National Physical Laboratory in England from 1964 through 1965.

In 1969, Craig was the recipient of the Hardy Gold Medal from the American Institute of Mining and Metallurgical Engineers.

He's also the author of more than 40 technical papers, and a textbook on materials science called "Principles of Engineering Materials." That sounds so fascinating, I think I can't wait to and get online and buy it right away. No. (Laughter.)

CHARLES WANG: This book today, though, remains in use today at universities throughout the country. Now, strategic partners, Intel and CA, have worked closely in many areas, and most recently we participated in the launch of the Intel Pentium® III processor. And certainly CA has demonstrated a good relation like this can be beneficial to both sides.

Ladies and gentlemen, please extend a warm welcome to Dr. Craig Barrett. (Applause.)

CRAIG BARRETT: Good evening.

The subtitle of my presentation tonight is "Software May Be Fun, But You Can Make Money Selling Hardware As Well." (Laughter.)

CRAIG BARRETT: I really want to talk about the Internet and the Internet economy, and I want to talk about it in the context of a vision we have of what's taking place around the world and where we will be within the next several years. And by "next several years," I mean within five or six years from today.

We're rapidly moving towards a world of a billion or so connected computers or connected devices that you can access the Internet and communicate with others. And looking towards an Internet economy which is measured in the trillions of dollars. In fact, within a few years, we should be doing a trillion dollars worth of e-commerce in the United States alone.

And the basis for the conversation tonight is how this impacts all of us: hardware, software providers, or perhaps, more importantly, you as information technology providers. What the impact will be, what the imperatives are for you to do your jobs and to support the businesses that you're employed by.

The driving force for all of this is obviously the Internet transition. Internet is clearly a strategic inflection point in technology and economic sense as we move forward. And today, we can address each other with our Internet addresses more so than our standard telephone or fax numbers.

The key here is how you win or lose in this Internet transition, and what the requirements are to play successfully. And if you look at what has happened with the global Internet, it has really eliminated boundaries between us. It automatically links the decision makers within companies 24 hours a day. It changes the way people communicate with one another. It creates new expectations from vendors and customers alike. But perhaps more importantly for all of us here tonight, it really changes the way that our business systems and processes have to interact. It really forces a dynamic linkage between the business unit manager and the information technology manager.

And if you look at the simplest picture of the Internet and the Internet economy, it really allows big companies to act like small companies and small companies to act like big companies. Over the Internet, you really don't know anything other than the company and their presence on the Internet.

One of my favorite cartoons from the New Yorker years ago showed two dogs at a keyboard in front of a monitor and one of them saying to the other, "You know, what I really like about the Internet is nobody knows you're a dog." And in a sense, that's really true from the business perspective. You can be what you are on the Internet with your presence on the Internet, and that's why your presence is so important.

If you look at the forecasts for revenue and business on the Internet, I've shown two separate bits of data here. One is for the U.S., and one is for the rest of the world. And the current forecast in blue on your left for the United States shows by the year 2002, approximately a trillion dollars worth of e-commerce done in the U.S. That happens to be approximately 10 percent of the U.S. Gross Domestic Product.

So going from nowhere to 10 percent of the GDP I think is a monumental change for this thing called e-commerce.

The rest of the world lags somewhat behind the U.S. In that same time frame, it's estimated approximately 300 or $350 billion of e-commerce will be done outside of the U.S.

I think one of the key issues here, though, is each year that these data are compiled, we find that the previous forecast underestimated by a factor of 2 or so where we think we're going to be in a few years. And I think the current forecasts are probably still an underestimate of where we'll be.

Give you a simple example. Intel took our first order over the Internet in July of last year, 1998. By the end of 1998 we were running at an e-commerce run rate, with our customers, of approximately a billion dollars a month. This year, we'll do $15 billion of our business over the Internet.

This is the rate at which things can change. You can go from nothing to $15 billion almost overnight.

Now, this sort of rapid growth, accelerated development of e-business, e-commerce, has led some people to conclude that within a few years there won't be any Internet companies as we think of them today because all companies will be Internet companies or they won't be here.

And today we think of the Internet companies as the Amazon.com's or the EBays or E-Toys, companies of that sort who are business-to-consumer Internet companies. In reality, the bulk of the business in the future, e-commerce business, will be business to business, not business to consumer. And all companies will have to learn how to interface with both their vendors and their customers using this new technology.

And I think perhaps the best way to show how this transition is going to take place, how we're going to use this sort of technology, is to do a simple demo up here of how you can take the existing type of business that you have today and to transform it into a Internet-style business with Internet-style interfaces, rich visualization, the sort of things you heard about this morning.

So what I'd like to do is have one of my assistants, Vivek, come up here, and we're going to do a very simple demonstration for you which basically allows you to show how an existing business can transform itself.

So Vivek - why don't you tell the audience what we have here.

VIVEK: All right, Craig. What I'd like to do is use some applications to illustrate the point you were just making.

What I have here is a Pentium III processor-based system running at 550 MHz, and imagine, if you will, that we are members of the insurance industry and we are dealing with a typical application that's deployed in that environment.

What you end up dealing with is a lot of character-based or, you know, barely Windows-like interfaces. For instance, you could be looking at insurance information, life insurance information, and it will throw you off to a different interface, and you may then come back and start looking at renter's insurance and it will throw you to yet another interface.

And so it's very non-intuitive the way the information is structured.

CRAIG BARRETT: Well, if you had this type of system, and suppose you were about to deal with a client and you wanted to find out if that client had any outstanding claims on your insurance company, how would you do that?

VIVEK: Well, it's obviously doable. That's why these applications are deployed. But the process of doing it is very tedious, so I would have to plug in some kind of a code, and then that would go to the specific account that I was looking at. And as you can see, the third transaction here is pending.

So, you know, we could then go on to see that particular transaction, and I know that this is a claim for an uninsured motorist situation.

CRAIG BARRETT: How do you know it's for an uninsured motorist situation?

VIVEK: Well, actually, in our trade, code 83 stands for that kind of insurance.

CRAIG BARRETT: That's obvious. You all have code 83 in the back of your mind, I can assure.

VIVEK: But you see, that's the problem I'm talking about; that the information here is not at all intuitive.

And so what I'd like to do is switch gears and take you into an environment which makes this same kind of information browsing a much more intuitive and easy-to-understand experience.

So let's go to the environment built in an environment called Jasmine from CA, and here I have a much more graphically rich interface.

So, for instance, if I was looking at this particular customer's record, I could go in and right there I have a very rich image which gives me a lot of intuitive information. So, for instance, I know that this is a family of three. They're all insured. I also know that some of their assets, like jewelry, diamonds, and even their home is insured with me.

CRAIG BARRETT: So you've just taken all of their previous text-based data and put it into a graphical user interface.

VIVEK: Exactly.

CRAIG BARRETT: At a glance, you can tell everything.

I'm kind of interested about that lamp shade in the center of the image.

VIVEK: No, that is not a piece of decoration. What it is is an indication for me that the customer's last interaction with our company was not a happy one. And in fact, this red flag here will further give me information about the nature of the interaction.

And it appears that they have an open auto claim, and so I know the problem that they're facing and, you know, the end at which I can help them.

Now, all that we've looked at so far, Craig, is what we can do today with the, you know, technology such as Jasmine. What I'd like to do is take this the next step and see where we can take such an application in the future. And in fact, we are working with companies like CA to build those capabilities.

So imagine, if you will, that I wanted to now look at a 3-D model of this car that was involved in this auto accident, or whatever it was, and to not only see the car, which is, of course, a beautiful car, but to be able to figure out what the nature of the damage was and what the status of the claim was.

CRAIG BARRETT: Judging from the fact that it's got rear-end damage and driver-door damage, it looks like Charles has been on his motorcycle practicing next to this Corvette.

VIVEK: Okay. Let's see what kind of damage we had. So I can go in, for instance, and all this information, by the way, is coming off my intranet. So the first part of the application was perhaps on my notebook or on my desktop, but this is sitting on my corporate database.

And so I can go in and I can immediately decipher the fact that the check was paid out on the 15th of July. And so now when I call this customer, I can be rest assured that, you know, I can give them some good news before selling them on some additional, you know, products.

And using, in fact, the same Internet/intranet connectivity, I can also receive rich information like training on new products, on customer support, and so on and so forth. So this becomes a very integrated environment for me to deal with.

CRAIG BARRETT: Super. Thanks, Vivek.

VIVEK: Thank you.

CRAIG BARRETT: A simple model where you're just taking legacy databases in the insurance industry and putting a rich graphical user interface on it, allowing people to interface with customers with relatively simple training, but to know the mood of the customer, all the details of the customer, and then be able to drill down below that for details on unpaid claims. And the details of that in a graphical way I think is one of the competitive advantages people will have as they move forward, as they adopt this new technology.

Now, to pursue on this, what are the real key issues here for companies who want to be Internet companies and want to use this new technology?

The first issue is one of mass customization or personalization. The major issue is if you're going to do Internet business with people or companies, you can tailor the information just to that specific individual. And this is what big databases and mining those databases is all about. And we'll come back to this in a bit more detail in a moment.

Second issue is we're really talking about this cliche of operating in Internet time. And this really means agility or the ability to introduce rapid upgrades and changes to your applications, your business applications, is important. And for years we've hammered on IT organizations to provide the same service at half the cost. What we're really now going to do is put tension in the system to say, "Fine. I want cost containment, but I also want you to be agile and to be able to give new capability, business capability, almost instantaneously."

The third issue is there will be new usage models for the Internet. Today, the personal computer is the primary client for accessing Internet information.

I think it will still be the dominant client as we move forward in the future, but there will be other vehicles to access the Internet, whether they're Web phones or handheld devices cell phones or dedicated e-mail devices. For different societal, cultural, and business needs, we'll see a variety of appliances and clients associated with the Internet.

And lastly, there's going to be an ever-increasing demand for compute power, and this has really balanced compute power in the server space and in the client space. And we'll come back and talk a bit about this, but think of it that we have bigger databases, more information that we'd like to access, richer visualization of the data, rich user input into the system. And this will require a strong balance between both the client and the server capability with a dose of management thrown in on top.

But let's look at each one of these in order as we go through. Let's look at customization and what the standard way of engaging in Internet business or e-business is today. And it's really the vendor-centric model, which is people say, "I want to be on the Internet. I put a home page up. People will come to my home page. If they like what they see there, they can buy it from me."

And this is the way that most people have gotten started. And this is most of the way that the current e-business situation occurs today, to a sense. If you go to Amazon.com, you go to their home page. You peruse it, you find the book that you want to order and order it.

But there is a change that's occurring here due to the personalization and people involved in e-business using their databases to single out individual customers and providing them with information that is unique to them. And this is the one-to-many type of marketing. And I really choose to refer to this not as vendor-centric e-business but customer-centric; where, in fact, you as the customer have vendors coming to you over the Internet to provide you with information useful for you to carry out your business or useful for you to buy some goods.

And perhaps the simplest way to do this is to show you a simple animation. And this animation will be, imagine I'm a Floral wholesaler located in North America, and I have suppliers located around the world in Holland, Singapore, et cetera.

And today, the way I do e-commerce with these people is they have home pages. I go to their home page, I access the information they have on their home page, and then I bring that back to my company. I add it into my spreadsheets, into my way of doing business. And what I can do then is just repeat this to all of my wholesalers, going from one to the other. And as I do this, creating a home page, a spreadsheet of all the information that these people have.

And once I've done this for, say, the vendor in Singapore, I can go and do this with my vendor in Holland, access their home page, bring that information back in, update my spreadsheet. Go on to Madagascar, go on to Australia, go on to all my different vendors. Each time, my job is to access their home page, bring that information back and update my spreadsheet to do my business. That's the vendor-centric approach.

The customer-centric approach is quite different from that. What it really looks like is, in fact, automatically, all of my vendors are uploading their information to me in the background without my having to go out to them to request that information. As that information comes in in the background, with intelligent agents, I'm screening that information, I'm automatically updating availability, inventory, demand and supply requirements, and using that to run my business.

And I can use more than just the information that those people are providing. I could really combine this with other information. If I'm in the plural {?} wholesaling business, for example, weather patterns around the world might be important. And I might be getting weather advisories as a routine update and using an intelligent agent to take that weather advisory and to look at the continuity of supply of a given product; in this case, something called bear grass {?} which is supplied from someone in Australia to me.

I notice that there's a weather warning that came up here, and if I just click on that in the background, I find out, well, there's an anti-cyclone off the west coast of Australia. That's where my bear grass supplier probably is going to be impacted in terms of their availability and continuity of supply to me and I ought to look for another supplier from them.

But this is the concept where information comes to me, as the customer, from vendors and from other sources. Intelligent agents and screening programs decipher through that information, allow me to do my business in the best possible way.

This is the sort of electronic commerce and electronic business we're going to in the future. And in fact, if you look at this, it really personifies the issue of competition is only a mouse click away. If you don't have this type of systematic data capability in your competitive arsenal, someone else will have it, and you'll be at a competitive disadvantage.

I think it's a very exciting transition that we're going to see in electronic business over the next several years, moving from this vendor-centric approach, which is very exciting in its own right, but into a customer-centric approach which carries with it a whole new set of compute requirements and demands on the system.

If we look at the next issue, which is this comment I made earlier about agility and efficiency and what the requirements are on the information technology professional in this space, I think you can summarize this with a couple of basic truths on the Internet economy. And one is that agility and speed in deploying your applications and/or services are important here.

And to make the real point here, I don't think you can do this without having the information technology organization in a strong partnership with the business units involved with.

There is a race to implement this technology. The person who is there first has a competitive advantage. And so this issue of agility and speed is even more paramount than cost for the first implementation.

You take that to the next step. The way that you do this, and the way that you give yourself upside capacity and headroom is you build your information technology infrastructure on open standards, open interfaces. Open interfaces promote a number of people writing software applications for this. It provides competition and the best price/performance capability.

So we want a horizontal or open model, and we want to be able to achieve this capability as fast as possible.

Now, graphically, what does this look like? Graphically, this looks like where we were in the past, where we've been asking our IT professionals for twice the capability at half the price, and we've been worried about low cost or cost of ownership, security of our networks, stability. We want to be up 24 hours a day, 7 days a week with no disruption due to people hacking into our system. So we want a certain degree of security and firewall protection.

That's where we've been in the last three years. But with this advent of the e-business and e-economy coming forward, effectiveness is going to be just as important, and it's going to create an immense amount of tension between business units and the IT organizations as we strive for the best possible world, low cost but agile, upside capacity headroom, and improving employee productivity in the marketplace.

So I think the wave of the future for the next several years is going to be in this direction; that is, decision support systems, big databases, rapidly moving into new business areas, and using this as a business automation tool, as a competitive advantage.

And this balance or this tension between efficiency or cost on the one hand and agility and speed and employee productivity on the other hand is a balancing act that we're all going to have to face with.

And the real issue is how do you do that, and what are some of the ramifications involved? I think it's kind of exciting here.

If you look at what's happened in the computer industry in the last decade, we've seen that it has gone, over the last 10 or 15 years, from a very vertical organization where we used to have different compute systems and applications in our business area, our finance area, our manufacturing area, to building more and more on a common infrastructure. First of all, a hardware infrastructure, then a network infrastructure, and then a middleware infrastructure, if you will, where we can interface or bring our messages and directory capability to play. And then something that interfaces with our applications, front-end, if you will, to legacy applications, databases of the type we showed in the simple example on the insurance company, or a back-end to Internet applications which are written specifically for the Internet but need to interface back into the legacy systems that we have.

Going from this proprietary vertical approach to the horizontal open industry really allows you to do one very significant thing today, and that is you can have standard solutions which can be implemented at different rates for different parts of your application space.

For example, you may want to update your electronic business system every three months. You may want to update your resource planning methodology and software once a year.

Open interfaces, common architecture, allows you to have decoupled rates of changes between these various applications that you interface either internally or with the outside world.

This is the way the computer industry has been going, and I think it will continue to move in this space as we go forward. We'll make a few more comments about that in a moment. But I think equally interesting is we can no longer just think about the computer industry by itself.

The communications industry and the communications equipment industry is starting to undergo the same sort of transition that we watched in the computer industry for the last decade. And if you look at what's happening there, where we've had separate voice networks and data networks, and the data networks and data traffic has obviously been growing many, many times out of the voice traffic, people are now looking to see why do I need separate voice and data networks? I can digitize voice and I can carry it over the same network that I carry over my data. My data traffic requirements are growing exponentially compared to voice. Why don't I just coalesce these into the same environment and then adopt the same structure that the computer industry has so efficiently adopted, which is a horizontal structure with common building blocks and open interfaces.

So we're now seeing this great transition just starting, mind you, in the communications equipment industry, going from vertical application spaces, equipment and software solutions for voice, to using standard building blocks, standard OSs, and then applications built upon those with open interfaces, so the whole industry can play in this space. The industry can move faster at a lower cost.

And I think that this is one of the exciting opportunities for both the hardware and software industries, and that all of us who are involved in the IT world are going to benefit from the cost effectiveness of merging these two networks, these two infrastructures, into a set of common building blocks.

Now, the third issue that I had mentioned which I think is important is this concept of usage models, and we're seeing this to some degree already, where many of us are using high -- relatively high performance personal computers to access the Internet. There's this phenomenon in the industry today where if you sign up with an Internet service provider for three years, you can get a free PC, a relatively lower performance PC, but a lower -- different appliance, really, if you want to access the Internet.

We're going to see other models occur here, as I mentioned earlier. Cell phones, handheld appliances, Web phones, dedicated e-mail machines. And the things that are really driving this capability are twofold.

There are two laws in our industry which cannot be denied. One is Moore's law that says we'll double the processing power or the functionality in your integrated circuits or electronic components every 18 months, and Metcalf's law which says the value of the network doubles with the number of nodes in the network, the square of the number of nodes in the network.

And those two driving forces mean that networks will get bigger and will have functionality in smaller and smaller devices to access the information via those nodes in the network on the Internet. And what I think you're going to see as we move forward is, in fact, an ever-increasing variety of Internet access appliances. The PC is still -- for the next several years will be the dominant appliance here, but we'll see many other appliances as well, whether they're set-top boxes, handheld devices, phones, what have you.

Very exciting, because this will expand the use of the Internet, expand the concept of the Internet economy, and then put all of these requirements back on businesses who want to pursue this type of approach.

Now, the last issue that I wanted to talk about here was this issue of ever-increasing compute demands, and not just on the server space, but also on the client space. So we have both client and server requirements here.

And I think the fundamental reasons here are obvious. The databases are getting bigger. Number of users are getting bigger. The quality of the information that comes over the Internet is richer, and we want this balanced computing system to balance some of the tasks with an intelligent client; some of the tasks with an intelligent server backbone.

If you look at another truth, if you will, of this Internet economy, it's really that balanced, scalable computing that delivers flexibility and the headroom will be required to be an e-business. And one of the reasons for this is you just never know what the demands are going to be on your infrastructure.

If you are into e-commerce or e-business, all of a sudden you've opened up your internal infrastructure to the outside world, and if there's an excessive demand from the outside world, your customers coming at you, they can overwhelm your system. But more importantly, you've also done the flip side. You've opened up to the outside world your internal customers. They can go out to the Internet, drag back as much information as they want, and can overload your network and your compute capability just by bringing new information in.

So it's no longer a closed-loop system where you can look at your internal applications and say, "Ah, I can deem how much bandwidth I need, how much compute capacity, how much server capacity, how much storage capacity." Now open the system to the outsiders coming in or our inside people going out, pulling information back.

And perhaps one of the simplest ways to show that this need for scalability is to just let you hear it from someone who's involved in this business, an Internet company called Interview who's involved in streaming audio and streaming video and streaming information out to a large number of clients.

So let me just have you watch this video for a moment or two.

(Video playing.)

"Interview is an audio and video service provider for the streaming media industry. We enable customer Web sites for audio and video streaming so they can basically expand their audience to millions of users.

"We do events that, you know, can be anywhere from a hundred to a hundred thousand simultaneous users, and ultimately it will be a million simultaneous users, and we have the network and the architecture utilizing the Intel platform to scale up dynamically without impacting the customer's event.

"It's really seamless for us from a technology upgrade perspective to grow our business and be more profitable and really reduce cost for customers in general.

"Interview really considers the Intel architecture as being one of the most scalable and really economical platforms on the market today.

"Our plans are, in the next few months, to increase our network capacity to close to a million simultaneous unicast streams, and we'll probably be deploying about 5- to 600 Xeon as well as P-III CPUs within our network.

"Utilizing eight-way and four-way technology, it really helps us process more customers in one's box, let's say. And the box is really going to help us scale our network significantly, ultimately offering a lower unit cost for our customer base and higher quality video and audio content.

"Our main mission with our distributed network is to be able to grow to the scale that these content providers demand, and ensure that they have a high-quality experience each and every time on the Interview network. And really, the way that we're deploying the Intel architecture is in that manner, the way we have the capacity and the processing power for the future, not just for the near term."

CRAIG BARRETT: You could really look at this scalability requirement at the front end where you're talking about Web servers and e-mail and caching and security and firewalls, all that compute capacity you have to put in to interface to the outside world. You can also look at it in a slightly different perspective and just at the number of customers that are involved or the size of the databases involved. And I think if you look at some of the forecasts for the size of the decision support system databases and the number of users accessing these databases, you'd just be flabbergasted by what the projections are in those growth rates.

And here we see some projections both in the average DSS database size and the number of users going from the 10, 20 gigabyte range up into the terabyte range in the space of two years. Or maybe going from an average number of users or people accessing these databases from a few thousand to nearly a hundred thousand.

And those numbers may seem phenomenally large, but if you start to think about some of the applications that are involved in e-commerce and e-business, what the Amazons and the EBays and the Charles Schwabs are doing today, the number of customers that they have, the number of hits that their Web sites take on a daily basis, the amount -- incredible amount of personalized information in those databases that needs to be accessed, you can see that this sort of growth is going to occur, will require a huge investment in infrastructure and capability. And this is part of the whole flexibility, this issue of balancing client and server and manageability.

And what we find is we really need to have something like this simple teeter-totter picture. We'll have a range of flexible clients. This is the issue that I talked about where PCs are the main clients today, and whether they're fixed desktops or workstations or laptop PCs, the one type, but you're, in fact, probably going to have appliances other than PCs approaching the Internet that you have to manage.

You need scalable servers. There are some simple estimates that we only have a few percent of the server capacity we need over the next several years in place today. And you really want to be able to just bolt servers on much as the gentleman from Interview was commenting on a few moments.

And then you want to be able to manage this collection of servers and clients and be able to grow it as rapidly as possible and maintain up time.

So really, flexibility is the main key here. And again, I could talk about this whole issue of flexibility and manageability. Perhaps the best way to do this is to show you a simple example again. And we'll have Vivek {sp?} come back out. And what we want to do this time is have him pretend that he's perhaps managing the Internet capability of a company like Intel. And from a manageability and growth standpoint, he'll be in charge of making sure that nothing goes down, or also that we could add capacity, make sure that nothing goes down.

VIVEK: All right, Craig. As you mentioned, what I'd like to do is use a demonstration setup to illustrate some of the points that you talked about.

First, to give you a quick coverage of what we are dealing with here, what I have here is a rack of Intel Pentium® III Xeon™ 500 MHz based processor running in solo configuration.

We have a dual Xeon processor-based workstation running CA's Unicenter TNG product. And then I also have a client here. And to get a better look at what we are dealing with, on your screen here you will see a network, the demo network. And most prominently, there are two servers. One of them is a Linux server and the other is an NT server.

So really, what we are dealing with is a heterogenous operating system environment and two such as Unicenter allow us to readily manage, you know, disparate operating systems from a single-user interface.

Likewise, I can go and do some manageability operations on my NT server.

As you have shown in your past presentations, Unicenter allows you to actually fly inside, you know, a platform so that you can visually manage whatever the assets are.

What we are looking at here is Unispace which is a step further and it lets you deal with some system aspects and agents, some of which we'll be looking at shortly.

Now, to build an appreciation for the kind of modes that servers have to contend with in today's times, I'd like to show you as an illustration this product Web site that we've got on our Intel site. And really what it has is a lot of rich information. So, for example, we can browse the network for some 3-D information, such as a 3-D model of the product that we're dealing with. And, you know, I can manipulate it. I can go and look at an animated illustration of how to install this product.

And really, these are the kind of operations that users are getting used to more and more in terms of their Web experience.

CRAIG BARRETT: So you're just pointing out here to the audience that if you're accessing this sort of product, which happens to be a home networking product, you might want to look at it physically in three dimensions. You might want to see how to install it in a very schematic fashion.

VIVEK: Precisely.

CRAIG BARRETT: All that requires information and download to the consumer.

VIVEK: And once you scale it to thousands of users, the server load is tremendous.

And likewise, you can add e-commerce capability, such as the one we've got here through our iCAD division.

Now, let's switch gears and look at the other end of the story. Let's look at how the server contends with such loads.

What I have here on our screens, as soon as that is brought up, yes, on the bottom half of the screen is the CPU time graph for this four-way server that we've got. And as you can see, we are bounding this server with the simulated load of transactions which includes browsing, file downloads, e-commerce operations, so on and so forth. And, you know, we are averaging about 88 to 90 percent CPU time utilization.

And we are actually delivering a pretty respectable performance in this environment.

CRAIG BARRETT: 88 to 90 percent utilization sounds to me like you're perilously close to being maxed out.

VIVEK: You're absolutely right. And the key thing in such a situation is to first become aware of the fact that we are in this potentially dangerous situation. And that's where the concept of predictive manageability comes in. And CA has implemented this technology called Nugents {?} which does exactly that. It does a good job of predicting such situations.

So if you look at the screen here, what we have is a Nugent message which predicts that we could run into a performance bottleneck based on one of these potential reasons.

So really what this is is a signal for me to go and do something about my server infrastructure.

CRAIG BARRETT: All I can say, it's probably a good thing we have 20,000 people who are our captive audience here tonight and are not out perusing the Intel.com Web page or your system would probably die.

Anyway, I think I'll send a salesman to see you tomorrow --

VIVEK: Thank you.

CRAIG BARRETT: -- to get another four-way system to put in conjunction with this one.

VIVEK: Absolutely.

CRAIG BARRETT: All right.

VIVEK: Thank you, Craig.

CRAIG BARRETT: Thanks, Vivek.

But that's the simple model of how someone who is involved in e-commerce can use predictive tools to see when they're about to have a problem from a capacity standpoint. And I think you can understand the issue of if you were to put a special or some particular promotional device on to your system overnight, you could double or treble the hit rate to your Web site, which, in fact, could require a huge increase in server capacity to manage all those pages that you're trying to serve up.

One of the things, one of the products that we do deliver into this space are, in fact, the microprocessors, which are the hearts and brains of these sort of servers.

What I tried to show on this slide very simply is the performance that you can get from these four-way processor systems today. The metric here is basically TPMC or a market basket of transactional activities of the type that we were just showing in that demonstration.

The vertical bar height is really the total horsepower. Bigger is better in this case, and you can see that this Dell-based system with four Pentium III Xeon processors is running at about 23 and a half thousand TPMC. The other two architectures are shown at somewhat less.

But I think the really critical issue is the yellow dots, which is really the price-per-unit of horsepower -- in this case, price-per-unit of TPMC capability -- which, in fact, shows that that Intel architecture solution, because of its volume economics, really delivers not only the most horsepower, but the lowest relative cost of the other solutions as well.

Now, this is an issue, which it really is kind of a contest which changes on a weekly basis as people come out with different four-way systems. The only thing you can be sure of here is that each week or every two weeks or so, there will be a new leader in this space as they tune these systems in the hardware and software standpoint to get higher performance or better value to the end user.

If you look at what's coming along behind this, it's kind of interesting because we intend to continue the 32-bit processor performance capability. Following Moore's law, there are several generations beyond the Pentium III which are currently under design. And in parallel, we intend to introduce a 64-bit processor family with the goodness involved in 64-bit processing.

So looking forward in both servers and workstations, you can anticipate this volume economics will continue with both 32 bit at the entry level for clients and low-end servers, and then 64 bit for high-performance workstations and high-performance servers.

And we're going to get into a little bit of the techie part of the discussion of 64 bits today. So I thought what I would do is I would invite Intel's chief marketing engineer to come and talk to you. He's our product marketing engineer. And he doesn't have a motorcycle, but he has a bunch of electrons.

So what I thought I would do is just click here and get Andy Grove to join us.

Here's Andy, the Intel chief product marketing engineer who would like to make a few comments to you.

Hi, Andy.

ANDREW S. GROVE: Hi, Craig. How are you doing?

CRAIG BARRETT: Well, all the demos have worked so far, so don't let me down; okay?

ANDREW S. GROVE: I couldn't possibly let you down. I'm automated to a "T."

CRAIG BARRETT: Actually, I have, you know, 15,000 of our closest friends out here in the audience, and they're excited about hearing something on this IA-64 family.

ANDREW S. GROVE: Well, I'll say a few words about it, but first I want to make sure that you told them that our world is going to be consisting of like a billion connected computers one of these days?

CRAIG BARRETT: I did that.

ANDREW S. GROVE: And that this network of connected computers is going to be used to ship hundreds of billions of dollars around from customers to vendors and customers and vendors to other vendors. And I hope you also convinced them that there is not going to be any such thing as an Internet company in a few years' time because we're all be Internet companies or we'll be dead.

CRAIG BARRETT: Hey, boss, I told them all that stuff.

ANDREW S. GROVE: Good. Then I don't know what I can add to that, except I think I would like to give that a little bit of numerical meat.

This chart that you are looking at is our best estimate of the portion of the U.S. economy that is handled by electronic commerce, electronic business today.

When you look at '99, keep in mind that this is a logrithmic chart so that the vertical access represents -- each division represents a tenfold difference, a decade worth of difference.

You notice two divisions separating the actual GDP from the electronic portion, representing that we are one percent of the GDP in the U.S. is handled electronically.

In four years' time -- in three years' time, rather, the chart is, the electronic portion, is going to gain sufficiently on the GDP that it is going to go to the 10 percent level. And in my experience, anytime anything moves a tenfold distance, fur flies and a lot of things happen.

From our standpoint at Intel, the most important thing is that as this build-out of Internet commerce and the backbone of Internet commerce is going to take place, most of it is still ahead of us.

On this chart, I compare the, on the left bar, the current server infrastructure that's in place, and I make some rough estimates adjusting that server infrastructure for a growing Internet population, the increasing use of diverse services, such as the ones you demonstrated, the increasing sophistication of those applications, the increasing use of richer media types. And considering another factor, that because of our increasing reliance on this server infrastructure, people will have to guard against down time and, inevitably, they're going to put in some extra amount of capacity to.

When you multiply all these things together, you get the right bar which shows that when you compare those two that you conclude that only 4 percent of what we are going to have in place in the year 2005 is in place today. So 96 percent is still to come, which is very good news from our standpoint because we are getting ready to introduce a brand-new architecture that we hope will make up a very large portion of the total, and that is the IA-64 architecture which is perhaps best hopes are best characterized that it will be the engine for customer-centric e-business.

Let me talk a little bit about customer-centric. I think anybody with enough money -- you're getting ahead of me on the slides.

CRAIG BARRETT: I didn't do a thing.

ANDREW S. GROVE: This whole thing is robotics.

Anybody with enough money to hire a consultant can come up with buzz phrases like "Quality is our most important product," and "Progress is what we make," and "Customers is what we are about." These cliches have been around at Intel and everywhere else for a long time. But when it comes to the Internet, one of those cliches becomes true; real foundation statement. The asset any Internet-related business has is its customer. The customer represents an ongoing business flow, an ongoing cash flow. Maintaining that customer is essential for realizing the value of the cash flow that that customer represents to your business.

The way you go about that is by knowing as much as possible about the customer's history, preferences, desires, and relationships to your product. And that is where the characteristics of the IA-64 come in.

First, I think maybe -- Now we can go to the next slide.

Maybe I should bring you up-to-date with this. I just went to a business update meeting this afternoon where the engineering folks who I have working on the IA-64 were all gathered up, many hundreds of them, and I congratulated them on the fact that they have reached, a couple of weeks ago, a very important milestone. They taped out the product, which means that the electronic database is now being turned into a mask. In a few weeks time, we should see the first silicon; rather, real, physical work can start, physical work of debugging can start.

On the simulator representing this product, eight different OSs boot, and the most important thing is, in addition to the fact that execution is proceeding is we are on the way to turning this into a real product, is that this product was conceived to be the engine of e-business.

And just addressing the characteristics of it, first of all, the fact that it's 64 bits allows electronic merchant to hold in memory and address in memory very, very large customer databases very rapidly. This is clearly a characteristic that will appeal to e-businesses as they scale their operations from their early starts to large customer bases.

Its very architecture, the EPIC architecture and feature of it that's called predication lends itself particularly well to interactive transactions which are the element of Internet commerce. And an exceptionally powerful floating point unit in this chip lends itself particularly well to encryption, which is necessary to provide security to these transactions, and to visualization that is going to be the saving grace as we are going to turn data into information, realizing that the rapidity and capability with which we do that will differentiate this product from it.

CRAIG BARRETT: That's -- All those details --

ANDREW S. GROVE: The most important part -- Pardon?

CRAIG BARRETT: All those details are nice, Andy, but you're really getting into the technical aspects of this thing, and I thought you were assuming the role of an industry visionary in your later life. And maybe you could upscale this and talk about the global economics impact of this IA-64 family.

ANDREW S. GROVE: Well, the global economic impact is going to be hopefully something very, very similar to what our microprocessors did to the computing industry. The volume economics of the structure of the computing industry and the volume economics of our factory network producing common product for all of this, all of these servers, the 96 percent of the servers that are yet to be deployed, hopefully will provide us more choice, faster evolution, and higher performance at the lower price, just the same way as we have done it in the years past for the desktop personal computer.

What Intel 32-bit microprocessors brought to the desktop computers, IA-64 and the IA-64 family and its members will bring to the server infrastructure.

Is that better?

CRAIG BARRETT: That's better.

I had mentioned earlier to the group here the logo for CA World is that basically software can be fun. I told the audience as I started the presentation that selling hardware can be fun, too.

ANDREW S. GROVE: Particularly when you have the best hardware that anybody can build and you're selling it in volume.

CRAIG BARRETT: Thanks, boss.

Intel's chief marketing engineer. (Applause.)

CRAIG BARRETT: Just to amplify a little bit of what's going on with the IA-64 family, I just tried to list here in relatively small print, it's difficult to read, I know, but just some of the companies who are involved with the IA-64 as the platform for customer-centric e-business. And this includes a large number of computer OEMs who have committed to provide servers and high-performance workstations. As Andy mentioned, eight separate OSs putting on the Merced simulator already and waiting to go with the actual silicon as it will be out in a few weeks. A number of enterprise application software vendors and workstation software vendors who are also porting and tuning their applications to this family.

We're very bullish on this, and we think it's going to be a great success as we move forward.

Now, one of the common questions we do get asked as we move forward is why do you need more performance? I can only type so fast, and I only do word processing on my computer, and I don't need any additional capacity.

What I really want to do is just make a few comments about this and then show you a demonstration about performance headroom. And the comments are shown in this slide, and what I've tried to do is pick a four-year time period, which is the approximate average for, in fact, CPU or computer lifetime for desktop machines in the industry. And what sort of additional applications will be put on these systems over that four-year time period.

And if you look at where we are today and the sort of demonstrations that we can do today, which are vendor-centric, e-business, and HTML-based Web viewing, and you look at what's going to happen in the future with intelligent agents, customer-centric e-commerce, natural data type input into systems, whether it's voice recognition or other natural data type input, and then going to more sophisticated agents and more sophisticated natural data user interfaces, you can see that the computing requirements of a client are going to increase dramatically over that four-year time period as shown schematically on this slide.

And I can show you slides of this type and talk about it, but perhaps the best way to approach this topic is to ask ourselves how would we be using this compute power in the future? And let's fast forward ahead to, say, three or four years from today and look at a typical business environment and what sort of computing would the typical business person do in a net-centric environment with rich data input.

And we're going to do another simple demonstration for you over at the far end. Chris Wrenn is going to join me up here and going to play the role of someone who is involved in e-business and e-commerce and is really going to show us what the system looks like three or four years from today.

And before Chris gets started, I actually want to show you, this is what the computer is going to look like. I particularly don't like the lime green, but the shape is anything but these rectangular parallel pipe {?} and these boxes that we have grown familiar with.

We're going to have futuristic style in the computers. This one happens to be a full operating model with all the standard computer stuff in it, and just a couple of USB ports in the back side.

But Chris, all yours. Show the audience what we're going to do.

CHRIS: Okay, Craig. Well, in order to help your audience get a better understanding of the future of e-business, what I've got today is a demonstration of a business desktop in the year 2003.

Let's imagine that I'm going to take the role as a marketing manager for a company called Ace Cameras. Okay. Let's go ahead and get started.

In order for me to gain access to my computer here, I'm going to have to first pass a couple of security checks. And the first one is going to be through a finger print identification.

I place my finger here, and hopefully it will recognize me. Okay. It looks like it did recognize the finger print.

After that's been done, I'm going to go ahead and now reconfirm with the voice recognition. Chris Wrenn.

COMPUTER VOICE: Good morning, Chris.

CHRIS: Okay. So it looks like we're into the system and it did recognize us in both of those security checks.

Why don't we go ahead and take a look at what my schedule is going to be looking like today.

Computer, please show me new messages. Computer, show messages from Greg. Oh, I've got two, but I've got an urgent one from Greg Holland which I'm actually going to be meeting this afternoon. Let's go ahead and find out what he wants.

Computer, play message from Greg Holland.

"Hi, Chris. It's Greg Holland. Hey, listen, our meeting out at Smith's has been changed to 1:00. Hope this still works for you. To get there, head north on Highway 50 and turn off at the main street exit, and Smith Cameras is on your right."

Okay. I'm going to take some of those directions with me before I get out there in case I get lost.

Computer, please convert voice to text and update my calendar.

In addition, also, computer, please contact Greg and let him know I'm going to see him at 1:00 p.m. at Ace Cameras -- or Smith Cameras, excuse me. Confirm by voice.

CRAIG BARRETT: I kind of like that. I mean, you've got a neat user interface which is different from anything I've seen. You've got biometrics, which are the security system with voice recognition and fingerprint recognition. You've got voice to text, text to voice. You can do all sorts of neat things here on natural beta types.

I'm going to like this system when it's available.

CHRIS: Okay. Great.

Let's go ahead and see what's next on my calendar here. Computer, show me my next appointment. Oh, that's right, I've got an upcoming conference call with Lei Wong of Camera Mart. She's the customer service representative there, and she was interested in some information.

But before that call, I actually want to do some research on some of their sales history so I can better prepare for the call and find out what kind of questions she's going to be asking me.

Computer, create a new sphere called comparative retail analysis for Camera Mart. Go.

At this time, my application actually is going out on the Internet and through our corporate intranet to find relevant information, performing some data mining, finding some key relationships such as -- and getting data analysis reports, sales database, and company financials.

CRAIG BARRETT: So this is kind of neat user interface and your kind of, I call it a soccer ball approach here. The most pertinent or primary information is in the center, and then secondary and tertiary information is in the little hexagons around the center. It gives you a great visual quickness in terms of seeing what you have and where you want to go.

CHRIS: Exactly, exactly.

Let's see. Oh, looks like their sales are actually doing quite well, actually. It's about 15 percent growth in the most recent quarter. So it's doing quite well. So I think I'll have some good information to actually relay back to Lei Wong at Camera Mart.

COMPUTER VOICE: It's time for your call with Lei Wong.

CHRIS: Okay. Why don't we go ahead and call Lei Wong.

LEI: Hello, this is Lei Wong.

CHRIS: Hi, Lei. It's Chris Wrenn from Ace Cameras. How are you doing?

LEI: I'm good. Thanks for agreeing to this meeting today. I wanted to talk to you about my margins with Ace. They are really low, especially compared to some of the other vendors we use.

Let me show you this margin chart from our mobile pad. See, here's Ace. Is there anything you can do to help?

CHRIS: I think so. Currently, you're placing one large order for a lot of our products which we ship directly to your central warehouse. Did you know that you can instead place orders directly from each store which would actually cut a lot of transportation costs?

LEI: Wow. That would be great. How do we set it up?

CHRIS: Well, I'm going to need access to retail store codes, shipping addresses, locations, inventory information.

LEI: I'm giving you access to that information now.

CHRIS: I see it coming through. I've got it.

LEI: I'm going to try to convert one of these large orders now.

Thanks, Chris. I need to go. I've got an appointment in a few minutes.

CHRIS: That's okay. I think we're done right now. Talk to you later, Lei. Bye-bye.

LEI: Bye-bye.

CHRIS: Okay. So what you're seeing right now is actually my application, my 3D inventory application, going back out on that Internet, linking up with Camera Mart's ordering system and actually updating in real time orders and inventory information from Camera Mart in my database.

CRAIG BARRETT: Great, Chris. Thanks.

CHRIS: Okay.

(Applause.)

CRAIG BARRETT: But essentially every bit of that technology is effectively available today, and it's really a matter of piecing that together, whether it's voice recognition, the biometrics, the rich visualization. This is what the industry faces for the next several years. And frankly, that sort of competitive advantage you'll get when you employ that technology in your firm I think will put you leaps and bounds ahead of the competition. That's what we're all working towards.

Now, let me very briefly summarize what we've tried to say today. First of all, the scale and pace of electronic business is accelerating. Each time we do a forecast, it turns out that forecast is only half of what the next forecast looks like.

You saw the data that Andy presented. Roughly 10 percent of the U.S. GDP by the year 2002. Major strategic inflection point, major change in the way we all do business.

And this carries with it the fact that usage models associated with the Internet which allow for precision marketing or directed individual marketing are going to be very important, very important competitive advantage.

Intel, in its doing of electronic business with our major customers, tailors a home page to each one of our top 2- or 300 major customers with only the information they want to see without burdening them with the totality of our information. It's a half-step, really, between vendor and customer-centric e-business.

With this precision marketing, whether it's what a company like Intel does, whether Amazon notices your buying habits and that you buy Robert Ludlum novels and when Ludlum has a new novel come out, they immediately send you an e-mail saying, "Did you know there's a new novel out? Would you like to purchase this?" This is the one-to-many direct marketing which is going to be facilitated by these huge databases, these mining techniques, this precision marketing to the individual.

Compute demands are increasing. Whether it is, again, the huge databases, but we need this balanced client/server compute capability with the management backbone thrown in.

Again, the comment that Andy made that only 3 or 4 percent of the server capacity is in place to take care of the projected level of e-business in the mid of the next decade I think is a telling issue. We have a long, steep hill to go to put that capacity in place with the huge databases and the huge amount of Internet traffic that it's going to serve.

And lastly, the issue is if you don't take advantage of this or your companies don't take advantage of this sea change, this strategic inflection point in the way business will be done in the future, certainly someone else will. And we're seeing this left and right today as people jockey for position in the Internet, especially in this business-to-consumer aspect with the very visible part of e-business. As your competition is a mouse click away, if you don't do it first, they will, and they'll have a strategic advantage.

But I think it's important to put this in the perspective of business-to-business electronic commerce, where you put your business at a disadvantage relative to its competitors dealing with your customers.

So it's something that we all have the option to do. I think the signposts are clearly there delineating the road. It's just going to be a lot of excitement as we go through this sea change into this massive movement away from one form of commerce to another.

CHARLES WONG: Thank you very much, Craig.

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