Intel Developer Forum, Spring 2000

Paul Otellini
Palm Springs, Calif., USA
February 18, 2000

PAUL OTELLINI: Good morning. Thank you. And welcome to day two of IDF. This morning, I want to talk about e-business and where e-business is going and a little bit about the infrastructure required to build the next-generation of e-business.

I think today all of us would agree that the business environment has changed dramatically. Every industry that we see is using and applying the Internet as a driving force of change inside its business and as a means of competing with its competitors in the world.

There's a shift in focus occurring. And the shift in focus is coming from a shift from applications focused on personal productivity to those focused enterprise-wide or business-wide on corporate or company productivity.

In the context of doing this, a new class of killer apps is emerging. Very simply, individuals, knowledge workers, are going to move from their office suites in terms of the center of their computer day, to environments where entire e-business applications become where they reside in all of their transactions and in all of their thinking.

This change is coming at a time when the global environment for it is really ripe.

If you look around the world today, Japan and Asia-Pacific have really rebounded quite nicely from the economic problems they had of a year and a half, two years ago. And, in fact, the Internet deployment is leading the economic charge in Japan for recovery, and in many parts of Asia-Pacific.

In the United States, what we've seen is that e-business has already grown to be one percent of gross domestic product and it's forecasted to be ten percent of gross domestic product within a few short years.

And Europe, in an interesting twist from the last round of IT deployment, has decided that the United States productivity gains in the last ten years have really been based on information technology deployment, and as the U.S. moves towards Internet-centric e-business, Europe has decided to move in an accelerated fashion, company by company, country by country, towards the competitive infrastructure required to compete on a global scale.

So to borrow the cliche for a second that all business is becoming e-business, that's the simple part. The hard part is how you build an e-business for your company.

You can point on one hand, simplistically perhaps, to deployment of personal computers. Everyone from Alan Greenspan to the Wall Street Journal is attributing the bulk of the growth in the U.S. economy on the productivity from IT deployment in the last few years.

And even in this environment, you can look at PCs grew 20 percent last year, which is a very, very healthy growth rate on a large base.

But I think what we're seeing is the move beyond PCs in terms of the drive towards e-business. And, in fact, what the drive towards e-business is all about is data. It's about information. Because in business, information is power. Knowledge is power.

If you look at the amount of data that businesses are collecting worldwide, the total data of business is doubling every year and a half. And, more importantly, as that database continues to grow and grow, what's happening is that the people employed to use that database, to analyze it, to find the trends, to be able to find the new competitive advantages, is growing by a factor of 25 every year.

Now, in this context, we've noticed some growth as well. Our Intel architecture-based server businesses grew in units by 33 percent last year. And the workstation business, which creates a lot of the analytical tools to analyze those databases, grew by 26 percent, both businesses outgrowing the overall server and workstation markets in total.

Now, we, as a company, can speak with some degree of knowledge and perhaps authority on e-business, because we've spent the last two years becoming the world's largest e-Commerce company. This is data from Q3 of last year. It's actually a little bit low. We did about $13 billion of our revenue last year online. Up from zero in 1997. And we have constructed a very large e-business inside of Intel. And as we continue to expand that construction to encompass more and more parts of our business, we've learned some things. And we want to share them with you this morning.

Where we are going is very straightforward. We have about half of our revenue today online. We want to drive to nothing less than 100 percent connection with all of our customers and all of our suppliers on e-business in the next couple of years.

We already do business 100 percent electronically in a number of countries around the world. For example, all of our business in Taiwan today is online, 100 percent of our business.

The next thing we need to do, though, is turn around. Instead of having just the relationships to our customers, we need to go back and construct those same relationships to all of our hundreds and thousands of suppliers worldwide that build and feed our factory network.

The implication of doing this, once we've connected our customers and our suppliers and all of our internal systems, is we can get a great amount of efficiency. We can essentially change our build plan, the way we load our factories, the way we load the hundreds of products in the dozens of factories of Intel around the world, and be able to change those build plans inside three days, based upon a much more real-time information feed connected on both ends.

And as we do this, we believe we will see significant hard cost savings in terms of real labor savings, in terms of reducing some inventory in the pipeline for us and for our customers, and being able to shrink that through better forecasting and connected means, but also an increasing amount of value accrued to the fact that we can do business with our customers when they want it.

In every geography in the world, 20 percent of our business in each of those geographies is done outside of normal business hours, doing business when the customers want it to be done. There's a distinct value associated with that capability. It's hard to quantify, but you know that the customers have a value associated with it.

So as we look at our experiences, what we've done and what we want to do in the next couple of years, we've noticed that there's a continuum of activities in terms of the deployment of e-business at our company, and I would theorize, for many companies as they go about doing this.

And simplistically, I would divide that into three generations. A few short years back, we had generation one. Every company put ".com" after its name. Every company put its brochureware up on the Web to have a presence. It was mass marketing done on the Web.

Generation two was a little more sophisticated. Generation two took transactions onto the Web. Companies started doing e-Commerce. And this is where most companies are now, doing one-on-one transactions with hundreds of thousands of customers worldwide on the Web today.

But there's a problem with this model. The problem with the model is it will not scale out as Internet business becomes increasingly more pervasive. The model today is what I would call vendor-centric, that is, that anytime anyone is doing e-Commerce, it is vendor-centric in the sense that if you are a consumer using someone's site, you're buying something from a catalogue at Land's End our using Schwab to do your personal finances or trades, you are dealing with that company one-on-one. And that company is dealing with many, many customers one-on-one. But it's all vendor-centric. While you're dealing with that company, you have a singular experience in that connection.

The same is true in business-to-business. In all of the secure private Internet sites we have set up for all of our customers around the world, those relationships, when they're purchasing agents or when the engineering teams get on to share information, are one-on-one. At the time they're doing it, they're dealing with Intel Corporation, and we're dealing with them. But we're also dealing with many other companies. It's a vendor-centric model.

That model cannot scale up as the business needs continue to grow.

What we see coming on is a fundamental shift here, a fundamental shift from vendor-centric to customer centric (or user-centric) computing in terms of the Internet model. And that's a model where you, as the user of the information, whether as a consumer or in a business model, have access to many vendors' solutions. Being able as a consumer to be on Schwab and also on Intuit and also Wells Fargo Bank's site to be able to integrate all that data and understand the implications on your personal financial structure.

In a purchasing department, being able to look at not just the Intel supply line, but perhaps the supply line for other components necessary to build the computers and link up with your MRP systems real-time, online, dynamically.

What this is doing is it's reflecting the fact that every business is both a vendor and a customer.

And as the applications evolve to take advantage of this and turn around and connect the supply lines, e-business application types have to evolve. And they evolve, I think, very simply to what we would call third-generation e-business.

And the characteristics of the third-generation e-business are establishing this one-on-one relationship with the people accessing your sites. It is much more dynamic than the transaction loads that were commonplace in the commerce models today.

And, of course, it's customer-centric by definition.

Now, this sounds great in theory. But I'd like to give you a very practical example of some application work going on in this area from a company called TheBEAST.com. Let's roll the video.

(Video playing.)

VIDEO: Financial marketplace has no time and tolerance for downtime. And that's something that we absolutely take as mission critical, and we plan to deliver on that by having a hosting mechanism, TheBEAST framework, that is absolutely -- has all the attributes that anyone would hope for in technology.

We bring business transformation to e-Commerce. The brick and mortar opportunities, they're fairly standard. And that's not what we're about. We're about changing the entire business model, leveraging technology to do that.

We offer them multiple sources at our level and then let them have their own sources and integrate into a single environment to be able to make sure that they are not susceptible to a single data vendor's or they are not susceptible even from our failure onto their business leads.

A lot of other people are also seeing real-time in that it gives you a price that is current. But we did a lot more than that by actually giving you what's called unsolicited updates, without you asking for it, whenever there's new information, we give it to you.

We're coming at it from a completely different angle. We're coming at it from the angle of trying to be interactive with our customers. Our framework allows all our applications to be totally interactive.

Really, we're leveraging a lot of those capabilities in our system and leveraging it where it's inheritable from the technologies that we use, like Intel and Microsoft.

We believe that Intel is going with third-generation e-business as well. Where we need to be able to give customers an environment where they're not susceptible to the failure of a single vendor. Also, you need an environment where you can aggregate disparate systems and bring them together into single desktop.

PAUL OTELLINI: Now, interesting, again, in theory, but I'd like to give you a closer look at TheBEAST.com. And to look that, I've got Katherine from Intel demos. Good morning. What are we going to look at here?

KATHRYN: This is actually TheBEAST.com here. And TheBEAST.com as discussed in the video is a Web site that delivers live financial data to your desktop.

PAUL OTELLINI: This is an awful lot of data here.

KATHRYN: This is a lot of data. This is just the raw data that you would actually have streamed to your desktop.

Now, this is very difficult to use. It's hard to visualize. So what TheBEAST.com has done is actually enabled the consumer to optimize and design the interface how they'd like it to be.

You can actually port the data into any personal productivity program such as Excel.

PAUL OTELLINI: This looks hardly like a consumer set of data types, though. This looks much more professional.

KATHRYN: This is very much for the professionals, corporate treasurers, hedge fund analysts, along that type of line.

PAUL OTELLINI: Even corporate treasurers need some simplification. What does it do when you want a simpler environment?

KATHRYN: This is the exact same data we saw on the previous page, except it's in a much more user-friendly format. The advantage of being in Excel is that I can now graph the data. I can do data analysis and a lot of different things. And if you see the flashing, that's the actual live updates.

PAUL OTELLINI: This is getting the same real-time feeds in a format where you've customized exactly what you want to see and being able to manipulate the data to show you various views on it.

KATHRYN: Correct.

PAUL OTELLINI: But not everyone lives in Excel all day long, either.

KATHRYN: Right. I wouldn't want to keep it open on my desktop all day. I've taken it one step farther. I've used the Microsoft Digital Dashboard software that's available on their Web site and customized my Outlook. You can see here on the left-hand side is I have my typical E-mail, inbox, below that my calendar. And on the right-hand side, what I've done is embedded all that TheBEAST.com data into my dashboard.

PAUL OTELLINI: Plus you have all the other investor information.

KATHRYN: Stock quotes, everything.

PAUL OTELLINI: That certainly makes life easier at work. But financial markets are open 24 hours a day. How would I be able to find out something important if I'm not at work?

KATHRYN: There's a level of customization that's difficult to see here, but let me show you. If I'm interested specifically on, say, the Japanese yen and how that exchange rate is changing. What I can do in Excel is set a BEAST alert. And what that does is when it reaches a trigger point, say it goes to 110, what it'll do is E-mail me and page me with that information. So regardless of where I am or what time it is, I get that live information right to me.

PAUL OTELLINI: Still on a customized basis, when you want it, what you want.

KATHRYN: Anytime, anywhere.

PAUL OTELLINI: Very good. Thank you very much.

KATHRYN: Thank you, Paul.

PAUL OTELLINI: I think that's an interesting example of the kinds of application types that are merging a lot of the technologies that we see being required to run this third generation e-business, real-time data feeds from outside the company, real-time feeds from your normal office productivity kinds of requirements, E-mail, Excel, so forth, and yet the ability to have that information pushed out to you wherever you are. Because the recognition is you're not always going to be in front of your screen.

In the context of doing this, the application types that we're all used to are really going to go through a very fundamental changes.

And from one perspective, the business models have to evolve to meet the needs of the individual, like you saw here, tailoring for the specific needs of that treasurer or purchasing agent, whoever is using the data.

At the same time, that person's data and information needs to also be integrated into all the business systems that exist inside your company in a real-time, dynamic fashion. And at the same time it does that, it needs to be integrated into the business systems of all the people that you interact with upstream and downstream of you outside of your company.

Now, the traditional computing environment is not really set up to handle this in terms of infrastructure. If you look at how modern computing is deployed today, it's sort of a two and a half-tier model. Most of the legacy systems, the big mainframes that run all of big databases in the world still reside in the back end and are still accessed principally in batch mode. But over the last few years, the industry has developed and deployed a number of infrastructure servers, SHV, standard high-volume servers, to handle the translation between the mainframe requirements and the needs of the day-to-day business in a real-time fashion.

At Intel, our SAP system is set up like this with a lot of local HP servers all running the front-end stuff, where all the order entry people can access that real-time on the LANs and on the WANs. But this model is not going to scale as we build up a stronger Web presence and start connecting to all of our supplier base as well. And the epicenter is going to move. The epicenter today is in the mainframe. It's going to shift. And it shifts very rapidly here towards a three-tier backend computing model, where you've got these front-end servers being deployed in .coms and ISPs and the front end of every .corporation to handle a lot of the transaction processing that you heard about yesterday from companies like Google and eToys and Commerce One.

And a full deployment of the mid-tier, the translation software, to be able to do all that the infrastructure requirements, where it's accessing all the databases still residing on the back-end large servers.

But the other fundamental change here is that not everyone is going to be connected on the LAN or on the WAN. And increasingly, all this data needs to be pushed out to the Internet to a multitude of devices, not just laptops or clients, but also cell phones or pagers -- the kinds of things that are going to get you the repackaged but critical and current information to do your job.

As this structure gets deployed to meet the needs of e-business, a funny thing happens along the way. What happens is the epicenter actually shifts. It shifts in terms of the power requirements, in terms of where the knowledge is, in terms of what the mission critical aspects of the infrastructure are to all aspects of that ellipse here in terms of encompassing all of your e-business deployment.

Now, Intel is not the only company who sees the need for change here at the application level and at the hardware level. A number of our other partners in the industry are also stepping up to this observation and starting to tailor increasingly their products and their services and their software tools for this environment.

Let's hear what a few of them have to say.

(Video playing.)

VIDEO: This next wave, this third wave, is the one where the real scalability of the underlying network that we're building will show off. And if there's anything we've learned from one-on-one marketing and mass customization, it's that people want to be dealt with as individuals.

As our customers embrace e-business models and look to embrace this, they should look for ecosystems of vendors that have worked together to build value-added solutions. Look for expertise of those vendors to really get them where they need to go in terms of scalability, availability, performance, and the services that they're looking for. And, finally, look for the right architectures that really build this all together.

We don't believe in creating a technology and making it closed and proprietary, because it inhibits customer choice and inhibits innovation in the marketplace. If you can create an environment where others can innovate on top of it, you can grow the market, and in so doing, meet customer needs, because customers want choice.

And the businesses that have headroom, that have system flexibility, that are able to make modification are those that are able to keep pace with the changing needs of the market.

I certainly believe that this third generation is about empowerment, allowing people to get the right information and make the right usage of that information, even if that information doesn't come solely from within their company or solely from one Web site. That's when things get really interesting and you can really start to reach new levels of productivity, new levels of engagement with the end user.

And we're collaborating with Intel to make sure that the underlying architecture works well with all of the next-generation network products. The combination of the two gives you state of the art solutions which are much cheaper and much more scalable than anything out there today.

PAUL OTELLINI: So what we're seeing here is evolving a very loud and clear and consistent vision of the application requirements for this third generation of e-business from all the major participants in the industry.

Essentially, the major theme of the developer portion of IDF this cycle is a call to action to develop these applications that integrate the e-business's services themselves in terms of the applications, but also integrate them with all the other services that represent e-business in this customer-centric fashion I described earlier.

To do this, though, we need a common language. And if you look at the evolution of the languages of the net over time, in fact, the Net has evolved to give us the capability to do this.

In the beginning, there was TCP/IP. And TCP/IP allowed us the basic building blocks to construct the communication capability that became the Worldwide Web.

As the Worldwide Web evolved, though, there was a need for a more graphical kind of presentation, the ability to put data up when you needed it in multiple fashions, accordingly, the links required to do your business. And HTML programming did a good job.

As we move into the third-generation, we need a new tool, a new lingua franca for the Net here to allow all the data that's going to be collected, all the data that needs to be analyzed to be able to be interoperable, to be able to be shared across vendors, across application solutions, and across application stacks constructed in every vertical industry.

XML and the various schemas around XML are the tools to do this. They are the ability to do the programming and do the data assimilation required to construct these application types.

XML progress is really moving quite rapidly. If you just look at the newspaper, it seems every day there's a new announcement. Recently, Biz Talk, RosettaNet Andy described yesterday in terms of the electronics industry moving to a common schema for interoperability, Ford and Oracle have announced Auto-Xchange, GM and Commerce One, a similar deal for the GM Network and so on. We're starting see the deployment of application types that encompass a given vertical industry and have the need for interoperability of data, sharing of data throughout that industry.

And around those, you'll see applications like TheBEAST or like MySAP or, in the consumer space, Quicken, to be able to take advantage of that.

But if I take you back to a model that I think Andy Grove first talked about over ten years ago, it's the horizontal model of the computer industry, something else has to occur as we deploy these applications.

The horizontal industry worked extremely well for the last 20 years because it allowed the best of breed in terms of interoperability, in terms of mix and match capability, to be deployed for any given application solution space. You could mix hardware vendors' solutions, you could mix a variety of software solutions to be able to construct exactly what you needed in PC space.

That same horizontal industry interoperability and economics of scale needs to be deployed as we deploy these kinds of application types around the world.

So as these vertical stacks emerge, the last thing we need is for each stack to be an island in and of itself. The stacks need to be able to be interoperable across multiple industries.

And the reason this has to happen is that there's a strong requirement going forward for adaptable building blocks. And this really is the focus of what Intel is building. All of those -- that array of products that Andy showed yesterday was really a construct of horizontal building blocks associated with all these open standards.

And the combination of all those from us and other vendors combines with software solutions to give IT managers and line of business managers choice and flexibility and economics of scale like they had in the PC industry.

If you look at how all of this is going to be used, companies are starting to construct 100 percent digital pipelines for all of their goods, for all their designs and all their services.

And as you construct this digital pipeline, the advantage of it being digital is you can share it. You can share it not only within your company, but you can share it with your partners, outside your industry, outside your company.

So we need to have the interoperability built in from the start.

You also get the dramatic cost savings associated with designing something once and reusing that data many, many times.

I recently talked to the people at Kenworth Trucks. Kenworth Trucks is sort of the epitome of an old line industry. They build big trucks. Up until a couple years ago, the way they built their big trucks is they designed full-scale clay models.

They've recently moved towards designing these on workstations, very, very powerful workstations, where they can simulate all the air flow dynamics and get the fuel efficiencies honed so they can build the trucks in a more efficient fashion. They discovered that the digital model of their truck was also their single biggest asset. They could share that with their subcontractors and get bids for different pieces all in a digital fashion. As they finally built the truck and took it into production, they could take the digital model and send it out to all their service technicians and dealers around the world, reusing that data, giving people a common look and feel to be able to deal with the realities of the physical truck itself. 100 percent digital pipelines require 100 percent interoperability of data going forward.

Now, yesterday, we heard a lot about scaling out. We heard about the requirement of front-end servers to be able to handle the peak loads and to be able to be deployed very rapidly in terms of peak capacity. I think Google talked about adding 30 servers a day here.

The front-end servers are the ones that represent your company to your customers worldwide. They personalize your company. If they're down, your company is down, as they said at Google.

In front-end servers, performance matters. Performance matters a lot. If you look today at the leading performance of things like dynamic pages served, the highest-performing system today is a cluster of servers from Compaq that has four application servers and one SQL server attached to it to be able to serve those pages above 3,000 pages a second. And it does this at a cost that is substantially lower than any of the alternatives out there in terms of prior or alternative architectures.

What this has done is it's essentially delivered a lot of the front-end servers to be on Intel Architecture.

If you look at front-end servers, they represent over 90 percent of all the servers installed in the world today. And 89 percent of those servers are based upon Intel Architecture. Now, we could stand here and say we've declared victory, this is good. All the work we've done the last few years is enough.

But there are two other very critical elements of the server market that are not yet really addressed strongly by Intel Architecture.

And they're the mid-tier and the back-end servers. These servers represent much smaller volumes of servers. They represent in the mid-tier, sort of five percent of the market, and in the mainframe range, one to two percent of the market in terms of volumes. But they're extremely critical in terms of the compute model and the homogeneity of data types in this model going forward.

Intel has been investing, as you know, in midrange servers and increasingly, in back-end servers to be able to bring a common architecture base towards all tiers.

Our market segment shares here are much lower than they are in the front-end servers, but the product lines are just starting. In the second half of last year, we had just begun shipping in volume the big gateway servers based upon Pentium® III Xeon™ processors.

And later this year, we'll ship, as you know, Itanium processors.

So to give you an example of what is needed in terms of performance for these mid-tier machines, I'd like to bring out Jon Haas from Intel, and talk about the evolution of mid-tier computing in terms of new I/O standards for the marketplace.

Good morning, Jon.

JON HAAS: Good morning, Paul. Really pleased to be here and show you the latest development in Infiniband technology. What we have here is an Infiniband prototype demo of a typical mid-tier configuration where we have two eight-way servers, one from Compaq, one from Dell.

PAUL OTELLINI: These are Pentium® III Xeon™ processor-based servers?

JON HAAS: Correct. And Infiniband will scale beyond the mid-tier, down to the front end and the back end. We have a storage subsystem prototype from LSI Logic, as well as a Crossroads storage router based on prototype Infiniband silicon. And it's going to be talking to both fiber channel and SCSI arrays. And also something new to show, the first Infiniband prototype switch silicon. With that switch silicon, we can actually take the Infiniband point to point link and create a fabric.

PAUL OTELLINI: Infiniband is more than a bus, though?

JON HAAS: Yes, it's actually a fabric-based bus model. So the real benefit is we'll be able to take the I/O adapters and components out of the servers, shrink the U-sized packaging of the servers of tomorrow and create higher density configurations for IT shops and the ISPs.

PAUL OTELLINI: We announced the Infiniband consortium six months ago at the IDF, and showed a rudimentary demo on it. What is this running at?

JON HAAS: A full Infiniband link speeds at two and a half gigabits per second. What we showed last time was a prototype that was running at half-speed links and had some ASIC technology linked with firmware to emulate either the host end channel adapter or the target end channel adapter. What we have today is basically all of that in one ASIC.

PAUL OTELLINI: Can we see what it does?

JON HAAS: Sure. So let me go ahead and pull up a Web site here on the screen. Let me go to the other one. We have a little problem with that one. The server is going to go through the Infiniband fabric and go to the LSA storage subsystem and start pulling AVI files.

PAUL OTELLINI: Give it a tough task. Pull up a multitude of files.

JON HAAS: Well, let me try it again.

This is brand-new, cutting-edge technology. So it may be bleeding right here.

PAUL OTELLINI: Well, it was working half an hour ago.

JON HAAS: It was working great.

PAUL OTELLINI: Okay. Well, risk-taking is a value at Intel.

(Laughter.)

PAUL OTELLINI: Maybe we can get it working for the showcase.

JON HAAS: Actually, it was in the demo showcase running in the last few days, it'll be running today.

PAUL OTELLINI: Okay. Thank you for trying. Infiniband really has come a long way in six months despite the fact this crashed. What we wanted to demonstrate is that it's now running at two and a half gigabits per second full speed. The consortium has grown to over 95 members, developing products associated with Infiniband.

We are on track for a full specification release midyear this year. And second silicon willing, we'll be able to deliver products into the marketplace in volume next year.

[Editor's note: It was subsequently determined that cables had been disconnected accidentally, preventing the demonstration from being launched during the presentation.]

At the eight-way level, the mid-tier level, the other aspect of buying criteria is performance. And in the past, performance has been the only attribute that buyers looked at when they wanted to find these solutions, because there were not that many out there.

With the emergence of the eight-way Pentium® III Xeon™ processor-based machines in the last year, we've been able to bring out a whole new class of performance in terms of meeting the performance of the other non-Intel Architecture servers that are out there, but, again, significantly reduce the cost associated with delivering these servers. And it's changing the dynamics and therefore driving the deployment of the mid-tier infrastructure on a worldwide basis.

But the real challenge for us is moving up to the back-end, these big database servers, the places where the single-image databases reside, where all the corporate legacy systems reside, that do the high transaction loads required to run an e-business. And this is where Intel has been aiming the bulk of its architectural activities in the last couple of years, trying to develop machines that continue to address the scaleup needs for the back end.

And to give you an example of that, we've got a piece of really big iron here to talk about. Let me bring Randy out to see what this machine will do. Randy, what do we have here?

RANDY: I'd like to introduce you to an impressive piece of hardware, the Unisys ES7000, which offers 32-way, Pentium® Xeon™ processor scalability in a single platform. And a real key feature here is that it offers interoperability, that you can take this and divide it or partition it into several independent computers, each doing its own thing.

So what we're going to show today is a fictitious e-business running on this platform.

RANDY: It's a shoe distributor.

PAUL OTELLINI: Pretty fictitious. You said it's partitioned. It actually runs multiple operating systems inside this environment?

RANDY: Absolutely.

PAUL OTELLINI: It seems it may be unique to our architecture.

RANDY: It is absolutely unique to our architecture. What we have is a diagram of the Unisys box itself. On the left side, you see we have Windows NT running. And on it we have the Unisys Cool ICE application integrates the databases and provides Web content to the customers, as well as the Microsoft SQL Server, which is handling order databases.

On the other side, we have SCO Unixware running Oracle and an inventory database.

PAUL OTELLINI: Both are resident and running on this.

RANDY: Simultaneous.

PAUL OTELLINI: This could satisfy all the back-end needs for a very large business.

RANDY: Absolutely.

Say we're a shoe retailer and we want to place an order. What kind of shoes do you think you want?

PAUL OTELLINI: Palm Springs. It's got to be golf.

RANDY: Got to be golf. Let's order golf. What we're going to play is a game of follow the bouncing red icon. You'll notice the red icon bouncing on the screen. Pay attention. It kind of goes quick. As we click on this, the red icon goes out, check inventory, and comes back and lets us know that the men's golf shoes are in stock. Let's go ahead and order, say, three cases, one for every day of the month, and add it to our shopping cart. So it comes straight back, telling us the price. We take a look at it, say, sure, let's go ahead and place the order.

PAUL OTELLINI: It's doing all these queries just living in this online repository at this point?

RANDY: Right. At this point, it's cached a fairly good amount of working information. It doesn't need to check the inventory database repeatedly just to verify that the same product's still there.

Finally, we can add the purchase information, shipping information, and place the order. And it goes out, updates the inventory database and the orders database and it comes back and lets us know our shoes are on the way.

PAUL OTELLINI: Great. So it's very flexible.

I understand that part of the flexibility of this design is that it's also upgradable to Itanium processors.

RANDY: You can pull out a pod of four Pentium® III Xeon™ processors and change it to Itanium processors and you're instantly running on that.

PAUL OTELLINI: Let me get Kenny to come on up. Good morning.

KENNY: Good morning, Paul.

PAUL OTELLINI: What is this loud thing you have here?

KENNY: As you know, as many users and developers of e-business solutions out there would tend to agree, implementing a highly scalable, secure, and responsive e-business solution is a big challenge.

And what I brought today with me is a prototype, four-way SMP Itanium processor-based server, the first public showing of this machine. And we're running the 64-bit SMP Windows 2000 on here, and E-components like a Patsy Web server, the SSL enabled one, and also a 64-bit Oracle 8I.

PAUL OTELLINI: We've come a long way since last IDF when you first demonstrated the Itanium processor. You've good it in a four-way clustered environment, Windows 2000, 64-bit running, and Oracle and Apache running on here, all optimized.

KENNY: Exactly.

PAUL OTELLINI: Getting close to launch?

KENNY: Pretty close.

PAUL OTELLINI: Maybe you can show people the capabilities of this system.

KENNY: Sure. And what you'll see on the screens right now is my server monitor. And what we have is actually the task manager, which is actually showing four active processors running right now. As you can see, there's not really much activity going on.

PAUL OTELLINI: Could you throw a few transactions at it.

KENNY: Sure. Let me go ahead and actually launch a benchmark called WebBench which is going to initiate a thousand secure SSL connections performing various Web transactions.

PAUL OTELLINI: I think we've all been familiar with the press issues in the last few weeks about hackers and so forth. Increasingly, things like SSL transactions are going to be required as a secure means of maintaining stability of your site, I think.

KENNY: Exactly. So, as you can see on the screen, where actually the pages are being requested and being served, and we're getting some activity on our meter here.

PAUL OTELLINI: But it's hardly taxing the computer.

KENNY: Only about barely 20 percent, which leaves a lot of freedom and room to do other things, a lot of growth right there.

PAUL OTELLINI: Maybe you could show them something else you could do with it.

KENNY: Sure. What I'd like to do now with the Itanium architecture, we're able to actually address 64-bit memory spaces, which implies that we can actually load up databases that are on the order of ten to even hundreds of gigabits in size directly into memory.

And with that capability, obviously, options such as e-business data viz applications can have very fast response time. So I'm going to go ahead now and log on to that Oracle back-end.

And, of course, we're going to be going ahead and give a very simple illustration of how fast this system is accessing a fairly large database.

PAUL OTELLINI: Do some real-time queries here?

KENNY: Exactly.

PAUL OTELLINI: Okay.

KENNY: We're going to go ahead and do a data analysis.

And keep your eyes on the screens. And the -- the query is going to be very, very fast, showing a very instantaneous response. And there we have it.

PAUL OTELLINI: That was so fast, I think everybody missed it.

Maybe you could go and do another step of mining below that.

KENNY: Sure. It looks like Santa Clara, which is highlighted in green, is showing the highest profitability.

So I'm going to go ahead and perform a second query and perform a drilldown here.

And, once again, --

PAUL OTELLINI: So, essentially, near instantaneous response on the query in the large database.

KENNY: Exactly.

PAUL OTELLINI: And the usage meter didn't move. So you can imagine, these kinds of servers are going to be able to run a lot of background requirements, SSL protection, but also be able to do hundreds and hundreds of queries real-time from every user who may need access to the database.

Thanks very much, Kenny.

KENNY: Thank you.

PAUL OTELLINI: What we tried to show when we designed Itanium processor, the hallmarks of that design were focused around the needs for e-business, the needs to have flexible computing, to have adaptable computing, to be able to handle this requirement for scaling out, but also the requirement to run a multitude of operating environments in a partitioned or even an unpartitioned mode to be able to solve all of your needs. There's a lot of progress on the Itanium processor. I think if you go to the IA-64 booth outside, I think you'll see eight OEMs showing prototypes of their systems running multiple operating systems, multitude of applications.

We've got over 30 OEM designs in progress for machines that will be introduced, ranging from two processors all the way up to 512-processor systems. And we are shipping thousands of prototypes to OSVs and ISVs around the world to accelerate the development of code focused and optimized for the Itanium processor. And then Intel and its OEM partners are also focusing on developing 40 distinct server and workstation application stacks that solve an end-to-end e-business requirement, all focused on optimized 64-bit computing in a multitude of industries and a multitude of hardware solutions associated with the product launch later this year.

So what I've tried to demonstrate this morning is that we really have changed our design principles in terms of where we're focusing our server activity. We're focusing on developing the infrastructure for the emerging requirement of customer-centric e-business. But as we do that, we're also trying to bring forth all of the goodness that we've all enjoyed in terms of the PC industry: The openness, the freedom to choose among multiple vendors, the superior capabilities in terms of performance and price performance, while driving cost-effectiveness all the time.

Now, Intel's role here is not just as a silicon supplier. As you know, we do a lot of other enabling activities. And in this space, the enabling activities are fairly varied.

Over the course of the year 2000, we'll be opening a series of up to 30 Intel e-business solutions centers. Think of these centers as very large laboratories equipped with multiple pieces of hardware to be able to bring in, from an e-business solution provider or Web development agency or any OEM, application stacks so that we can tune these stacks in IA-32-bit space or IA-64-bit space, for the application environment, again, focused on all the interoperability. In addition to this, Intel capital that is invested over $150 million in over 50 companies, all focused on delivering these third-generation e-business applications that I described earlier.

And lastly, the Intel 64 Fund, which is focused on developing code for Itanium, has now made investments in over 20 companies since last IDF. So we're well along the way towards software enablement to follow the hardware deployment over time.

What I tried to show you this morning are four points. One is that Internet business is becoming a true global reality. The second point is that business, as we see it, is moving from vendor-centric to customer-centric e-business. The third is that as this happens, the industry needs to rally to be able to deploy these interoperable volume, cost-effective solutions, like we did in the PC days. And lastly, that Intel's role in this is to continue to be the building block supplier for this emerging set of needs and Internet space.

And to explain to you more about our needs and our products in networking and communication space, the other parts of our building blocks, let me introduce Mark Christensen.

(Applause.)

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