Fall Internet World

Craig Barrett
New York, NY, USA
October 6, 1999

ANNOUNCER: Normally I'd have one of those microphones, wireless mics so I can wander around, but I'm stuck behind the podium, so if I seem a little awkward, bear with me.

I think the Internet has become so much more complex both from a business and technical perspective, that it will be difficult to navigate. What I'd like to do is throw out six ideas or six topics that we at the magazine believe are really hot topics today and will be compelling issues going forward. I'd like you to as you listen to the keynotes throughout the week, as you attend the conference sessions, as you meet with vendors on the trade show floor, to think about these, and ask questions about these topics.

I'll just kind of rattle them off; perhaps we can talk more about them as the week goes by.

This idea of personalization. The Web can be a dark threatening place, sterile environment if you let it, but the technology is to a point now and some of the skill sets of architecting sites, user interface design, we can create a much more personalized experience, so we can turn it into a high touch experience, so think of it in terms of personalization on the Internet. Think about customer perception.

This is more a business interest -- businesses are spending hundreds of dollars to attract customers. Part of this is the stock market valuation process, where the number you want to throw out is the number of users you have. This is something that AOL was using for a number of years. It's one thing, though, to acquire a customer, another to attain one, and it had profound impacts on your business. So think about customer retention.

Dynamic price. Traditional economics -- we talk about the demand curve, supply curve, that's where you place your product. Internet has heated up a lot of different pricing modes; we're really, in many cases, moving from a world of static pricing to dynamic prices where many businesses will want to develop a pricing strategy, so we have auctions, reverse auction, demand aggregation models, a lot of exciting plots that are being played with here in terms of price.

Globalization, the Internet which is now pretty much to every corner of the planet. Projections estimate more than 50 percent of the on line users, customer base will be outside the U.S. in the next year or two. We also have broadband -- the pipes are being laid rapidly, the pipe is being let and broadband is on its way, whether it's wired or wireless, how is broadband going to impact your business?

Lastly, the idea of ubiquitous IP, IP everywhere, where the last fifteen, twenty years when you thought about computing, you probably thought about a PC, you probably pictured a desktop PC in your mind. Well, now we've already seen, I think we're just at the beginning of the coming explosion of IP capable devices. Right now two of my favorite are my cell phone and my Palm Pilot. We'll see more on the show floor today during the fashion show in terms of wearable computing.

So if you are the head of a multi-billion dollar company that really made its name in the PC marketplace, you might be wondering or you might be very concerned about what you're going to do in this new world, post PC paradigm, if you will, of the ubiquitous IP.

So our presenter today, Mr. Craig Barrett, president and CEO of Intel Corporation, will be giving us a presentation on his vision of the future of Internet, and he'll be on stage after a brief, two minute video, thank you.

(Applause)

(Video shown)

ANNOUNCER: Ladies and gentlemen, please welcome President and CEO of Intel Corporation, Craig Barrett.

(Applause)

MR. BARRETT: Well, good morning. It's a pleasure to be here. Coming to one of these events always reminds me of one of the alternatives to Moore's Law which states the number of transistors and IPs doubles every month or so, and it's kind of interesting to watch the Internet double every six months or so throughout the world, probably best personified by attendance at this particular meeting, which is up well over 100 percent from last year.

What I want to talk about today is the Internet economy and the building blocks and how we all play in the major changes occurring in e-Commerce, e-Business around the world, and what I will talk about is three major subjects. We'll talk a little bit at a high level about the Internet economy, how fast it's happening, the impacts on all of us. We'll talk about implementing e-Business, and in that section of my presentation, I'll try to draw on some of the experiences that we've had.

We started doing e-Commerce, e-Business, business-to-business standpoint with our customers about a year ago, currently running at about, oh, roughly in the range of $35 million a day, something over a billion a month, hope to get to roughly $30 billion a year within the next twelve months.

Lastly, I'd like to come back and talk about the Internet and talk about it from the concept of the basic building blocks, scalable infrastructure that goes with this, and throughout my presentation, I'd like to focus on a couple of issues. One is basically the necessity for scalable infrastructure, putting building blocks in place to which you could easily add capacity and volume and also the price of performance of this capacity that you put in place. If we're dealing with everything that doubles every year or so, the whole issue of a scalable infrastructure and the price performance of this in the long term I think has to be of critical importance to all of us.

You start with the absolute highest level issue; we're talking about a vision of a billion or so connected computers or Internet appliances and a trillion connected dollars, and we can talk about the time frame for each of these things happening. It's clear that a trillion dollars of e-Business or e-Commerce is going to happen faster than a billion connected computers. Most of the forecasts show it will be well over a trillion dollars of e-Commerce within the next three years. It may take well into the next decade to get a billion connected appliances, computers on the Internet, but the issue is this is happening at a phenomenal rate.

My boss, Andrew Grove's, vocabulary -- this will be a major strategic connecting point in all our lives. We need to take advantage. The question to ask yourselves as you sit there is am I involved, am I involved at the right level, am I moving rapidly enough in this arena?

If you just look at the current forecast for e-Business in the United States, something in the range of 100 billion this year, and most of these are forecasts that really get updated on an annual basis, we usually double the out year's forecast. But let's say we're going to be 100 to 200 billion in '99 and going to something like a trillion dollars in 2002. This is a measure of the amount of e-Business that's done. Most of it will be business-to-business transactions and not the more popular press copy of business to consumer transactions. But even so, the magnitude of the volume of business is phenomenal. That's 10 percent of the gross domestic product of the U.S. Three years from a basically nothing a year or two ago.

So 10 percent of our total economy moving over to this new form of business, that's significant in its own right, but I think even more significant if you look underneath these numbers and look at the competitive aspects of doing business this way. The cost savings, the productivity and efficiency improvements associated with it, these are probably even more significant than the gross numbers we're talking about in terms of the volume of e-Business.

Now, clearly the United States is the center of this activity. The forecasts here for e-Business and e-Commerce are much higher than the rest of the world, but the rest of the world is also looking at an exponential increase here, and if you look at the non-U.S. numbers, it may be only a third of what we're forecasting for the U.S., but just as we're seeing a doubling of our activity every twelve months here. If you go to the emerging economies around the world, those economies which don't have strong telecommunication or computer infrastructures, you find that the level of interest and activity, whether it be number of ISPs, volume of business the ISPs are doing. or the projection for e-Commerce in Latin America, the Asian Pacific region, eastern Europe, you find they're all seeing a similar growth rate.

I just returned from a one week trip to five countries in Latin America, met with ISPs in each of those countries, and every one of them I met with, if they weren't seeing a 200 percent per year increase in their business, they thought they were failing. That's a measure of how rapidly the groundswell is building up in these parts.

All of this leads me to a conclusion that whereas we talk about Internet companies today as a unique breed, in five years we won't be singling out Internet companies, because all companies will be doing business over the Internet, and all businesses will have some component of e-Business associated with it.

The simplest catch phrase here that is driving this I think is competition is only a mouse click away in our futures, and either you get involved in the process, or you're going to get left substantially behind.

Just to give you an idea of what this looks like, something that most of you are familiar with, I'd like to show a quick video here of a born-on-the-Web company. The company happens to be 800.com; the speaker is going to be Gregg Drew who is the CEO of that company, but this gives you an idea of how fast this is happening, what sort of customer service you can provide, and some of the unique competitive advantages. So please roll that video.

(Video shown)

MR. BARRETT: I hope to buy 50 servers real quick. Actually, the quarter is over, so I have three months to worry about that.

Let's talk a little bit about implementing e-Business. I really want to follow the conversation here with four separate phases of implementation, and I'll talk about these in the context of Intel and a couple of other companies who are also considered to be leaders in this area, either in the business-to-business or business-to-consumer implementation of e-Business. We'll talk a little bit about the infrastructure, what's involved here, talk about how you have to modify the process by which you interface with your customer. We'll talk about exploiting data or feedback loops and how you get competitive advantage. Then we'll talk about the business process.

Just to put this in perspective, from Intel's example, as I said, we're doing roughly a billion dollars a month of e-Business today, with hundreds of customers. But the way we really do this is to tailor our interaction with each individual and each customer, so although there may be only 300 or 400 customers that we're dealing with, we have 6,000 customized Web pages for the manufacturing manager, the purchase manager, the logistics manager, etc., in each one of these companies. So the personalization is going to be very critical as I go through the discussion.

Let's talk first about phase 1, and really the building of the infrastructure, and most of us are not really on the Web type companies, just started business on the Internet recently, we have some business background and some legacy infrastructure that we have to accommodated in this media of doing business. If you look at our history, I think it's really no different from any other corporation. We started out in the back office, or basically a corporate database ERP type system, running SAP some six or seven years ago, actually implementing that structure well before we even had a Web presence, which didn't happen until 1994, and the initial Web presence was obviously an intel.com, informational presence, not a commerce presence.

We didn't activate a commerce presence until last year, the middle of last year, so the legacy database was there we've been building over the last six or seven years, the Web presence for the last four or five years, the commerce presence just for the last year or so. To give you an idea of the infrastructure of the commerce presence, it's a couple hundred servers just for the commerce aspect, not really the intel.com in its entirety, and a fair amount of bandwidth. As I travel around the world I realize how important bandwidth is as I address countries in Europe, Latin America, or the Asian Pacific, and I compare the international bandwidth those countries have compared with the bandwidth a company like Intel has. We have a couple hundred megabit per second bandwidth interface with our country. I think in all Latin America, Brazil is the only country which has that much bandwidth, so to understand how the billions are dollars are going into place to put bandwidth capacity, fiber capacity, not only inside the United States, but from the United States to all the other continents and then within those specific continents.

So you could put the commerce structure on top of a legacy database, and I'll show you an example of exactly how that works a little bit later on and the importance of the stable infrastructure that we have.

The other issue I think is significant is how rapidly your customers become involved in e-Business, or e-Commerce. When we started our program in the middle of last year, we dealt with a number of major customers around the world in places such as Taiwan and Brazil, Argentina, Chile, who really were not Web centric. They weren't even on the Web basically, and within six months of dealing with those customers in those diverse spots, they basically converted to 100 percent of their business with us on the 'Net and were suggesting improvements to us and how we could interface with them. So in a six month period they went from basically no presence to doing 100 percent of their business with us, and, in fact, suggested how we could improve the interaction with them. So the infrastructure builds in very rapidly. People become Web centric very rapidly, and the whole process moves forward.

If you look at page 2, it's really then the changing or automating the way that you interface with your customers. Comments here of automating the tacticals is very straightforward. If you phone up Intel today and talk to one of our customer service representatives, that customer service rep has to scroll through several screens to interface with you on the telephone and enter your order, and what you really want is a hands off transaction, and you want the customer to be fully in control, and this is where personalization of the Web pages to your customer becomes absolutely critical. So we've had to modify our internal processes and automate them such as the customer sees only the information that they need, sees it in a totally automated fashion, and we're working to get totality of our e-Commerce transactions now in a total hands on phase where Intel never touches the order, we only receive it and then automatically acknowledge it.

There's a lot more you can do than just the buying and selling of the transaction over the 'Net.

Similar example that I have again with Intel is we distribute a lot of confidential documents to our customers. These are design documents, specifications for products, architecture, etc., and the way this was done in the past was someone wanted these documents, we had to get the document to our field sales representative, field sales rep went to the customer, handed the document over, got them to sign a nondisclosure form, brought the nondisclosure form back, forwarded it to our central office where we entered it into a database.

Today, the gate, all those manual transactions, primarily authorizing the customer with the confidential document over the Internet, what happens -- they download the document, the electronic signature goes into the database. You may not think that's a great productivity savings until you realize we had 150,000 of those transactions last year. Now those 150,000 transactions are taking place without the middle man involved, direct Intel to customer transaction. Productivity savings are great to us, but they're probably even more significant to our customers who get those documents on a real-time basis and then start their design and their implementation more rapidly than they could have in the past.

And lastly here, in terms of modifying the process, the major issue from an internal management standpoint is I think you have to start combining your various functional organizations, sales organization, the logistics organization, the IT organizations, production organization, warehouse organization, such that in fact they all work together and don't exist in a stand alone, smokestack-type operation. So having them joined at the hip is absolutely critical in the automation or implementation of this process.

Now there are lots of competitive advantages you can get from doing business. I've talked a little bit about some of our productivity savings. You can look at some of the other obvious examples in the industry. Dell, who has converted a lot of their business over to Dell.com business, I think has a major strategic advantage over some of their competitors in the computer industry. Basically, they don't build a computer until it's ordered, and what they're capable of doing is working with a very short supply line, a very short inventory pipeline of only about seven days of inventory to satisfy all of their customers. This has two advantages, really. One is that you can flush through the old technology to the new technology in the space of one inventory turn, which is about one week. This allows Dell to have the newest technology in the marketplace to sell to their customers before their competitors, but the other one is kind of interesting as well, which is Dell does business with people like Intel on a standard 45-day term. They do business over the Internet, they bill their customers with credit cards, they get almost instant payment, so they get paid in seven days, and they pay me in 45 days.

I've talked to Michael about that, but he's such a good customer, I think I'm going to let him have a float on our money for that process. But again, a unique advantage he has in terms of his suppliers are in fact funding the capital of a part of his business. Not many of us make that similar claim.

Amazon I think you're all aware of in terms of their capability of their database, the buying habits of their particular customers, to have a feedback loop to turn this to their advantage. You buy two copies of Robert Ludlow from them, you automatically become a Ludlow fan. They tell you when his next book comes out. They also can market other things besides books, CDs, what have you -- they're a large operation. So once you're into this group, you can, in fact, consistently use it to your advantage.

The fourth issue here is building customer centric solutions, and by customer centric, I mean this is more than just putting up your home page and having customers come to you and then sort through your information. It's really having a worldwide presence and building customer-centric solutions rather than vendor-centric solutions. I also talk about putting a home page up and having customers come to you which is customer-centric -- which is having the customer come to the vendor and deal with you on the vendor's terms. You really want to deal on the customer's terms, personalizing situations, which we do with engineering management. We don't have a home page with all of the company's information on it for the engineering manager. We have only information the engineering manager wants, only information the purchasing manager wants, logistics manager wants, production managers wants, so personalizing the content and driving home to the customer is absolutely critical.

We can talk on and on about personalization or customer-centric solutions, but I think maybe the best way to address this issue is to give you a little demonstration, and I'm going to have Kathy Moran from our Intel Content Group come out and do a little demonstration here which has to do with the travel industry, how we can present that sort of information to the customer in an easy fashion. Kathy?

MS. MORAN: Good morning Craig.

MR. BARRETT: Tell us what you have here.

MS. MORAN: In this demonstration, we'll see a series of technologies optimized -- Intel Pentium® III processor. These enable businesses to leverage and extend their content for e-Business. While the context here is travel e-Commerce, the concepts and the technologies we'll see are applicable from business to business and other business to consumer environments. So first we'll go into Adventura. This is a site that's a fictional site we created that is similar in concept to Travelocity or Preview Travel, a travel aggregator, and going here we'll see that I've got a personalized section that I've customized for me.

In this space, the messages right from the start are tailored for me, based on past travel experiences. Places I've gone to, information I've asked for from Adventura. A trip to Venice, hiking in the Rockies -- I like to do these outdoor kinds of things. We've also got aggregation of my frequent flyer information from all the different airlines that I participate in and while I live in Virginia, I like to travel to San Francisco, so in Fare Watcher, I've listed San Francisco. Adventura, through their e-Business relationships with numerous travel vendors, is able to access, in real time, travel inventories, and they've created a travel package for me to San Francisco which I can purchase at the click of a button.

MR. BARRETT: I'm not interested in your travels. I'm interested in where I want to go.

MS. MORAN: You've been doing a lot of travel, particularly to South America to talk about e-Business, so to highlight this customer centric information, I've planned a vacation for you. Where would you like to go?

MR. BARRETT: Someplace where there's white sand and warm water.

MS. MORAN: Sounds like Hawaii is a good fit for that. See what you can find on Hawaii.

Today when you do a search, you get a list of textural responses and you can work your way through that text based list, but here we've got a different approach. We're showing the search rules in a graphic representation. If you look at each resort and attraction, we can see its location relative to each other. So a while back, I saw an article here on this particular resort, the Hamawi resort. This may be a good fit for what you're looking for.

We can see this resort, because using life picture technology, they're letting us take a virtual tour.

MR. BARRETT: Pretty cool.

MS. MORAN: Well it gets even better.

MR. BARRETT: I should have known that.

MS. MORAN: Here we've got an IV I embed entered this virtual tour. You can imagine how that kind of content can be used to promote a local restaurant. We continue our tour. This is the pool area here. What do you think?

MR. BARRETT: I think it's pretty good, but I'd like to do stuff other than just sit by the pool.

MS. MORAN: Let's see what else I've got for you in resort activity. This is using IMIX panorama technology. We can zoom in, see things, but most important, now we've got the opportunity to send e-mail, so this is a great way to capitalize on virtual marketing. So I thought maybe we should let your wife know what we're doing here, since we're planning a little vacation for you. We'll send her this content.

MR. BARRETT: As long as I can depict what we're doing to Barbara.

MS. MORAN: She'll see it and get excited, too. This will be sent as an attachment to e-mail. It's a JAVA based executable, so she won't need any special technology on her end to see it. She can manipulate it as we do. How about outdoor activities while you're there?

MR. BARRETT: Good -- I like hiking.

MS. MORAN: Malakawi Falls, attraction in Hawaii. Here, with live picture technology, we can zoom in, see up close in detail how lush the area is. It's one thing to see a waterfall in this kind of detail, but this is a technology that's being used quite a bit now by business to consumer e-Retailers. They're using it to showcase a product to consumers so they can simulate that experience of picking something up and being able to examine it. Can you imagine with apparel, detailed stitching?

MR. BARRETT: Yes, looking for some.

MS. MORAN: I think you might be safe here. Let's go back to the resort. Does this look like the place you want to see?

MR. BARRETT: I think I'd like to stay, but I'd need more detail on their activities.

MS. MORAN: Let's get some of the package. Honeymooner or golfer's dream? Do you want to play golf?

MR. BARRETT: In deference to Barbara, I think I'll play golf.

MS. MORAN: Let's see what's included. Here in the cars we can see, using Meta 3D technology, to look at the Ford Explorer. What do you think?

MR. BARRETT: I like it. In fact, I need a big car, not only for the golf clubs, but for the surf board.

MS. MORAN: I don't see information right up here, but let's look at customer service and see if they can help us find some information for you on surfing. Today sending e-mail back and forth, you cross your fingers hoping for an answer. Let's see a different approach.

Well, I think he's asking us where we want to surf. We're looking for something in the Mallawich resort.

MR. BARRETT: Lipreading helps on the Internet.

MS. MORAN: Steve here sent us some information electronically using video over the Web for customer relationship management, so now you got your package. Does this look like the trip for you?

MR. BARRETT: This looks pretty good so far.

MS. MORAN: Go ahead confirm this so far. Here's the itinerary mapped out, upgrade to the Explorer. You can confirm this. Looks like the golf resort through their relationship with minute that law which has a package for you. They've offered the process and will give you $125 off for tee time now.

MR. BARRETT: I'll commit, and I want to surf during the day, so I'll take an evening tee time.

MS. MORAN: Now you see your trip to Hawai was added, interactive itinerary and e-mail for anyone who wants to know about your travel plans.

MR. BARRETT: Great there's a lot of stuff going on here.

MS. MORAN: You're right. The need for businesses today is to create digital marketing with the Web in mind. We've seen what consumer centric Internet looks like. If you use what you learn about individual customers, you'll succeed the most in this Internet economy.

MR. BARRETT: Great. Thanks very much.

(Applause)

MR. BARRETT: I think that that's a good demonstration, because it really suggests how people are going to have to change their business to function differently in interaction with customers, and this also, I think you can turn around a little bit of it. The way you function with your customers has to change, and this old comment about form follows function, if the function is going to change, you have to make sure the form is in the proper fashion. The form in this case is really the infrastructure that you have to carry out your business, and if you look at the basic Internet infrastructure, there are several pieces to this, and we refer to these as the building blocks to the Internet. The Internet building blocks are very simple and straightforward.

First there's the appliance or the client platform that you access the Internet with. Today, that's essentially a PC, but there are other Internet appliances. We'll have a few comments about them later on.

There's the networking infrastructure, and that is what your client is attached to and goes off to the Internet backbone, and the Internet backbone is a series of servers or browsers, networking devices that direct your request and respond to your request. And then somewhat embedded in servers off the Internet space is the interaction or the transaction capability to satisfy your needs. So we have the client, the network, the server backbone and then the solutions and services that reside on all of this infrastructure.

And I think one of the interesting things is, in fact, as you build up your capability, you have to look to build up this entire backbone, this infrastructure of servers and clients and network capability.

Now, what I want to do here is not just talk about this, but to show you an animation of what happens when you're sitting at your keyboard and what goes on over the Internet when you make a request of one sort or another, and we all know how to type in, I want to go to www.intel.com, something like that, and put that on the screen. Then what we do is hit the "enter" button on the keyboard.

Of course you all know in detail what happens when you hit the "enter" button, right? I'm not hearing any response, so maybe you don't know.

What I want to do is show you a very simple animation of what happens, and that animation will go through three or four steps. The first step is we're going to hit the enter button, some information is going to go from the client here out of their telephone wire to the ISP, something is going to happen at the ISP to redirect this off to this huge backbone of the Internet which is billions of dollars worth of fiber, high-banded infrastructure. This information is going to get magically routed to the right information center that pulls that information that I'm trying to get once we get into that information data center. The request is going to be directed through some routers and will go through the Web servers, some appliance servers, the Web server is going to gather information. We compile it, then it's going to send it right back to the customer back to their screen.

So with that little bit of an introduction, let's follow this step by step. Let's imagine what I'm doing is I'm sending a request to a web site. I've got it piped in in textural form. I want to go to some web site. I punch the enter button. My digital message basically lifts off to the PC out through the telephone line to my local ISP. My local ISP is someone who is not too far away, typically, and there are thousands of ISPs in the United States, big ones and little ones, but they tailor their customer service to you.

Once at the ISP, that information now has to be translated to something that can go out over the Internet, so if you follow what happens here, in fact, over a local area network, into the ISP, over a LAN, basically into a Web server, over to a domain naming server, where we're going to translate that textural form of the Web address into a digital format, so we'll see that, here the xyz.com gets translated into the Internet address. Then we're going to go through a traffic management server, then to a router, then we're going to go out over the Internet.

Where we're going to go to next is from the ISP to a network access point, a point where we can get on to this high-banded infrastructure over the Internet. And in this little segment, we're going to go to the NAP and then on to the data center where the information actually resides.

So, we're going to leave the router here, on to a NAP. Does go a little faster in real life. Once there, information gets sorted, routed with thousands or hundreds of thousands of other e-pulses, gets sent on to the data center where the data resides.

Once in the data center, it goes through a router, a management server which finds a Web server that's unoccupied. That Web server then sends out requests from application servers and the application might be commerce, encryption, streaming, audio or video, whatever you have. Pulls that information back to the Web server, builds the Web page that we want, and then magically, I want to get the information back to my screen, response from the web site goes back exactly through the traffic that we followed, and we're looking at a total elapsed time here from when you hit the enter key to when the message gets back to your screen of 480 milliseconds.

Not all of you may get the information you want back in 480 milliseconds. This was for demonstration purposes only. Your actual mileage may vary.

But, in theory, if you want a small message and you got reasonable bandwidth, that's what happens. The reason for showing you that little animation is, in fact, to show you all of the network infrastructure that is required when you hit that enter key for something to happen, and those are the building blocks that I want to talk about now; what's going on in those building blocks, what some of the futures are, what some of the trends are, and what some of the factors are in going forward.

We'll go from the left side, the client side, and go through applications and solutions over to the right side.

This is the Web page we got back. Let's talk a little bit about what's happening at the client, and why I think performance headroom is important in the client as we move forward.

As you know, we're able to put more and more processing power in the client each year, and, fortunately, the applications that are demanded off the client increase on an annual basis. What I tried to show in this simple schematic is what's going to happen over a four year period. There are certain applications required by the client today where you have a certain degree of visualization, certain degree of input, a certain degree of complexity in terms of the messages that are going out over the Internet. They're primarily HTML based today, but there's a certain amount of horsepower needed to do that. If you look forward over the next three years, what's going to change when we go forward, you'll see a number of things change.

One is we're going to go from HTML base to JAVA apps and XML base, which is going to pull out the user and give them a richer experience. We're also going to make increased use of knowledge management.

Another way of thinking of knowledge management is intelligent agents, contour processing. While you're doing one task, you want your computer to be working in the background about some other task -- gathering information about a customer, a client, without having to have you intervene. So you really set the system up as an intelligent agent to go out, pull that information off the Internet while you're working on a secondary task. If you look at the knowledge work of the future, this is, in fact, what they're going to do. They're going to have background tasks going on while they're doing something in the foreground.

Lastly, what we're going to have is better user interface. By better user interface, I don't mean just a better keyboard. I mean, we're going to be able to speak to our computers. If any of you are starting to use that technology today, more importantly, many users around the world don't particularly find western language style keyboards particularly attractive to interacting with the Internet or with computers. If your native language is an eastern language, in fact, speech input is a much more valuable commodity than keyboard. As you all know, speech recognition algorithms get better and better, get more accurate with more processing power. So if you take this computation from HTML to XML, JAVA apps, and the legacy database, better use of interface, better visualization, going from rudimentary knowledge management to sophisticated agents in the background, we're going to need more processing power as we go forward with the client.

So client headroom is a significant situation and, fortunately, the industry is working in that direction. It's not just PCs as clients here. We're really talking about other clients as well. Whether these are cell phones, in fact, the shipment of cell phones around the world is larger than the PC shipment today. Both of those are larger than the number of TVs that are shipped today, but cell phones, especially as we get to the next generation of communication which gives you greater digital bandwidth on the cell phone, are going to be access devices to the Internet. Things like handheld devices, Palm Pilot are access devices to the Internet, dedicated IP phones or Web phones are going to be accessed. All sorts of network appliances are going to exist, as well as the PCs, and this forecast, we can see PCs continue to grow, but other appliances also start to make a presence on the Web.

I basically think this is wonderful we have other appliances, because it will get more people involved in our business, and as there are more people involved on the Internet, that has to be good for computing, Internet commerce, Internet building blocks as a whole. So very exciting issues here. I think it's going to be very competitive to see which of these appliances are successful and what format. As we all know, Metcalf's Law, the value of the network doubles with the number of attachments and number of nodes, so as we add more nodes, PCs or appliances, that makes the network more valuable from a commerce standpoint.

So basically good stuff is going on here.

If you look at what's happening in terms of the backbone of your infrastructure within your company, I think there are a couple of critical issues here. One is it absolutely has to be scalable, and secondly, it absolutely has to be the best price performance you can get. As business grows, what may seem inexpensive today from an overhead standpoint will be very expensive as you start to put in hundreds or thousands of servers.

A couple of issues could make this a little more complicated than you might expect. One is, if you're an IT manager in a company today, you pretty much know how to size your infrastructure. You know the internal applications you have, you know the size of your network, you know the number of clients on your network. But the two things that are happening to make your life more challenging -- one thing is that your employees are starting to go out and access information on the Internet and pull that information back into your home network, and you have very little control over that because your employees are doing things to make the job easier, make their life more productive. So your employees have open access to the Internet, are able to bring information back, and you have to scale to accommodate that. But the reverse is also true. People from the outside are starting to have access to your network as they come and do business with you, and you also have very little predictability in terms of how many people are going to come at a given time and do business with you. If you project yourself to be the IT manager at Charles Schwab, and you're doing online trading, or e-trade, and the market dynamics probably impact how many customers come to you on a daily basis. So maybe it's a thousand customers now or today, but the market crashes or goes dramatically upwards tomorrow, maybe 10,000 customers will come to you. So you have to scale your network and your computing capability to be flexible in terms of how many customers come from the outside and how much your internal people use the outside capability.

Key issues here are flexibility, scalable server structure, scalable clients to be able to do what your employees need to do on a product basis.

Now, fortunately, it's a little bit easy to do this because of the way the computer infrastructure has been modified over the last decade or so. We used to have computer infrastructures which were vertical silos, vertical application stacks. If you wanted to do a manufacturing application, you went to Digital and bought their manufacturing package. A business package might come from IBM. Financial or banking package might have come from MCLI. So there are different solutions for each one of the applications. If you wanted to try to superimpose e-Commerce on top of that, it would be extremely difficult.

Now, what's happening here in our structure is that, in fact, we've gone to a horizontal building block structure where we can put unique applications on the very top of that structure. We have a standard hardware infrastructure, standard OS, standard duplicate infrastructure, something called middleware, the messaging and directory services which allow things like e-mail for us to communicate with each other, and that's application services, which, in fact, that layer of the structure which allows you to interface with the legacy database.

If you look at Intel, typical company here, their e-Commerce applications, to an applications services layer on an SAP infrastructure, we've interfaced that legacy infrastructure. SAP had no idea of the Internet when they created the ERP database, but we can interface with that database and modify or change e-Commerce application at the top layer as much as we need to. We can do it independently from underlying databases or independent of any applications. I think if all of you are aware, if you can modify your e-Business applications, you probably modify that much more often than your ERP applications or HR applications, so these vertical application stacks on top of this applications services layer on top of legacy databases allows the implementation very rapidly.

This horizontalization of the computer industry is exactly what's happening in the Internet. The Internet today runs over a couple of networks. Most of us have digital and data networks in our companies, but when we go outside, we have telecommunications or voice network structure. Increasingly, as data becomes more important, and as you can see in the very schematic graph on the left, today data and voice are about equal in magnitude or volume running over these networks, but data is broken into a couple hundred percent a year, voice is only a few percent a year, so data networks have to win in this competition -- they have to become the dominant transportation mode, and there's no need to have two networks, because you can turn voice into a digital pattern and take it from a digital switch to a package switch capacity.

So what's going to happen here is the voice communications industry, which looked like the computer industry a couple years ago, vertical application stacks, you buy something from Lucent or Ericson or Alphatel or Nortel, they're all proprietary hardware software solutions, not very interactive. They're all going to go to the standard building blocks just like the computer industry has gone to the standard building blocks with standard infrastructural layers and then open interfaces on which you can put these new application patterns.

This is what really Mother Nature likes. Mother Nature doesn't like vertical structure; she likes horizontal structures. It allows the whole industry to write applications and add value to these open interfaces that allows technology to move in the marketplace much faster.

Now, just to make a couple of comments about servers. In the animation I showed you, the server infrastructure was very important. There's a ton of servers in the background of those clients, and a simple estimate is that for every ten clients you have, you probably need one server in the background to handle them, and if you look at the demand on servers infrastructure that's going to take place over the next several years, what you find is we have only a small fraction of that server infrastructure put in place.

The best estimate is about 5 percent of the server infrastructure we need over the next five years for current Internet growth is in place, and you can rationalize this in a simple way. First of all, you could take the current number of Internet users, and it's going to increase by fivefold going from a couple of hundred million to a billion over the next five years or so. That gives you a fivefold increase in the server infrastructure you need. Users will become more Internet savvy, take advantage of more differentiated services over the Internet so that they're using the Internet more. The service features, capabilities of each application on the Internet will increase, will have rich streaming video and audio capability, similar to the demonstration I showed you before, and this will all add up to about a twentyfold increase in the number of servers.

Again, one of the critical things that's going to be important here is the fact, price performance and scalability of the server infrastructure. We're all going to want to instantaneously add capacity in a price competitive fashion. What you see coming down the stream are new and exciting advances in the server space. We just introduced a name for our next generation server a couple of days ago. Many of you may recall it was Merced in the past, our first 64-bit server, we named it Itanium™. I kind of like the name. I was educated as a metallurgical engineer many years ago. Titanium was our favorite metal. We took the first T off of that; it's a strong, cost-effective solution.

But this architecture is designed to be the backbone for what's happening in the Internet; 64-bit gives you better memory addressability, parallel execution gives you higher performance, allows you to take care of more requests, more user applications in any one period of time. Strong floating point performance gives you good encryption technology, gives you rich visualization capability, good animation or rendering capability to do rich imaging over the Web. Because people are putting more and more mission critical applications on the Web, what you want to have is a 24 by 7 capability with some error detection and correction, self correction capability on the Internet.

These are the technical specifications of the product. I think the most interesting thing to do is perhaps give you a demonstration of what the IA 64 family Itanium™ architecture looks like. What I'm going to do is have Steve Hunt come out and do a demonstration for us on the IA 64 family.

What I think is important here is to look at an industry momentum that is behind this capability and what it's going to do, or cost-effective implementation over the Internet, and what we're going to do is try and demonstrate a couple of aspects of the Internet. One is content based which is also very important, but also Web usage or business usage over the Internet. Steve?

MR. HUNT: Good morning, Craig.

MR. BARRETT: Welcome.

MR. HUNT: We have actual machines we brought from the lab. We have two of them. Right over here, you can see the actual Itanium™ processor. It's the black cartridge there. We have a couple of memory board and graphics and network cards.

MR. BARRETT: I should have told the audience that the products came out of the production lab in about thirty days, but I think you might be aware of the advances we made in 30 days of debugging.

MR. HUNT: Yes. We've actually got here three machines. Let me describe the demonstration. On the left, we have a Windows 2000 server up and running, all 64-bit code running on the Itanium™ processor base workstation right here. That's the latest, 2128 called Techies; we're running the Internet information server on top of that operating system.

MR. BARRETT: Great. Let's show them what this thing can do.

MR. HUNT: Okay, give it a test, right. Let's point and fire up the windows box. Here I've got a little visualization we can run. This would be something that we might use to trade an image on the Web site if you wanted to market a product on the Web. Here's a robot, so if we know in the future we might be buying a robot.

MR. BARRETT: Kind of like this?

MR. HUNT: So this is actually an application that we've worked with ISP imported over, from using the IA 32 development machines, cross compiled the code and then transferred the binaries on to the IA 64 box.

MR. BARRETT: Everything you're going to show us is basically 64-bit code?

MR. HUNT: That's right, all 64-bit code. This is from mental images, called mental Ray. It's using the Itanium™ processors, advanced floating point architecture to do floating raise.

MR. BARRETT: Why don't you show how we can use this in business?

MR. HUNT: Sure. Over here we'll pull up the Internet Explorer and go ahead and run a Web site here. Now, this would be more like a business-to-consumer type of situation. I like high tech gadgets, so I thought the sharperimage.com Web site we could look at. We're actually serving content here. See what some of the hot things are out there. There's a CD radio alarm clock. I've got an actual 3D plug in here; we can actually serve that content and actually manipulate it and look at it. You can see it's online as well.

MR. BARRETT: Tell the audience, business-to-business and business-to-consumer can afford it as well.

MR. HUNT: That's right. So in this scenario, let's assume we're Sharper Image. What we'll do is go on ahead and call up another Web site from Vistel, and this is the latest pager, it's a voice pager and it allows us to take a look at it, pull up, serving from the Itanium™ processor--oh, I think we've got a page. Let's see--.

VOICE: Don't forget the ice cream on the way home.

MR. BARRETT: That wasn't Barbara's voice. That must have been your wife.

MR. HUNT: I think so. Better get some ice cream.

So, as you can see, we can also manipulate this thing, close it up, open it; it's a very interesting kind of way to deliver the Web content to the client.

Now, one thing I should point out -- we're running on a 733 megahertz Intel Pentium® III processor based client. So with that floating power there, we can actually do this rendering there.

MR. BARRETT: Okay, we're playing business-to-business transactions, so let's carry that transaction out.

MR. HUNT: We actually want to buy this thing now. We want to stock our Sharper Image store. What we're going to do now is actually take the link -- we're actually going from the Windows IS based Web site over to the other Itanium™ processor based system that we brought over on the right, which is running Linux with an Apachi Web server and SSL, secured socket layers; so we'll go ahead and enter the secure web site.

As you see, we're getting detail of the access lag from Apachi. You can see it generating a key, so we're using the advanced floating point architecture to generate the RSA purchase.

Let's go ahead and enter our customer number and password. Submit the order, and let's see. They want us to buy 50. That's probably a good number for a trial to stock the store up with. Let's get a delivery date here; okay, go ahead and submit the order. Okay, we got a confirmation already, and you can see they're going to send us a confirmation e-mail within one hour. I bet they're faster.

MR. BARRETT: It only took 480 milliseconds in my animation, so they'll be faster than an hour.

MR. HUNT: Let's go and pull up Outlook and check for mail. We'll connect over to the mail. Oh, there it is. Let's pull it up. Got our mail confirmation right there.

MR. BARRETT: That wasn't canned, right, Steve?

MR. HUNT: This is all completely live, running all 64-bit code on Windows, on launch in this action, two interlinked Itanium™ based workstations with the IS 32 client all in the server.

MR. BARRETT: Cool. Thanks.

(Applause)

MR. BARRETT: You won't be able to buy those systems for about nine months or so, but what I wanted to demonstrate to you is how far we are in the engineering applications standpoint. We've got four major OS's running; there will be several more, dozens of people doing the applications on all the OEMs, targeting applications and servers, everything from two processor systems working here, all the way to 500 plus processor systems with huge database applications. Very excited about that technology moving forward.

The other area we think is very interesting is the service side which, in fact, is what's happening relative to services on the Internet, and we're really talking about going to the next generation of service application space when people have done Web hosting in the past at location mode, kind of ad hoc combinations of hardware or software to the next generation. We have a highly reliable management structure with qualified hardware software network combinations, absolutely necessary you have 24 by 7 critical mission applications. These consumer forms or Web hosting capabilities will look a little bit like the photographs here, space age type control centers, because reliability and management is so important, as well as the access servers in the background. I think this will also be a major change in the way this operates.

Let me conclude by saying first of all, I think all business is moving to e-Business. Competition is only a mouse click away, and all of us are required to get involved in this business as we go forward.

The Internet forces all of our business to change in terms of being more customer centered and tailoring personalizing the information we give to each customer, so your business will have to function differently and it will require something of a different infrastructure, require your internal organizations to be more cooperative and work closer together to achieve solutions.

And the last message I have for you is really the architecture of your solution which really needs to be a price performance, competitive, scalable architecture to grow. All of us are looking for phenomenal growth aspects as we go forward, and I think it's incumbent on us to put that infrastructure in place now as we start to build these infrastructures.

Thank you for your time and attention this morning. It's been a pleasure being here. I think this is perhaps one of the most exciting changes in our industry that we have seen in the last decade and is certainly going to prove to be very significant and challenging the next five or ten years as we go through this transformation. Thank you.

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