February 23, 1998

Read the transcript of the Andy Grove interview or discuss Intel's Enterprise Push in Shop Talk

Craig Barrett Interview

INFORMATIONWEEK: Intel is putting a vast amount of effort and money into alliances, investments, and software companies all aimed at encouraging uses of more powerful processors. What evidence do we have so far that it is paying off? Is it worth the hundreds of millions that is going into it?

BARRETT: We characterize everything we do in the areas you are talking about [in terms of] how we are generating new users and new users of PCs and new users of high- performance PCs. I guess the best measure you have of success is if we are able to continue the wave of new processors in volume production and have the public buy them. If we make these transitions then it is a successful strategy. We can then argue in detail whether it was a necessary and sufficient part of the strategy to make these investments and coalitions and all the sorts of hardware, software, and other types of company investments that we made.

Here is the processor, we deliver the silicon volume production. We work with the operating systems, vendors and ISVs to make sure that they write the programs for this.

INFORMATIONWEEK: Are there actual concrete examples that you can point to where you think it has paid off already, or is it really an act of faith at the moment?

BARRETT: If you look at some of the things that are coming out in digital imaging -- videoconferencing and some of the other visual applications on PCs -- those are exciting. A lot of those things happen not by a single investment ... we can put together coalitions of people to make a more complete product offering. So, the classic MIPS- eaters today are the visual applications, 3D, games and, certainly, speech recognition. We are not doing a whole lot of work in the speech-recognition area, but that is one that I think is finally coming of age. (One of my first jobs at Intel was on digital signal processors in 1980. It was a little bit ahead of its time. But you see all these current estimates on how many people are going to be using speech recognition five years from now. Hundreds of millions of users around the world have no idea what a Western language keyboard is, and either you do speech recognition for input and a device for them for the PC. )

I think there are indirect measures where this is successful. Last year we invested some $300 million in terms of small equity investments facilitating companies and such -- over a hundred investments. We can also measure the portfolio value, whether it is a good venture capital fund. It is a good venture capital fund. I have n o problem with that. The main value return in my mind is whether this PC model continues to roll around to the next-generation high- performance processor -- if it is accepted by the public.

If you want a test case, maybe look at the workstation. There we have a combination of investments, ventures, and our own projects, and you could measure whether the Intel model is succeeding in the workstation market or not.

That's a little bit different category than what I was talking about. The workstation and server space is where we are making huge investments both on the hardware and software side. A lot of these other investments are just to generate exciting, new applications and uses for the standard desktop PC.

INFORMATIONWEEK: Where a lot of this seems to come together is looking at the [Intel] platform as the business enabler. It kind of brings up two. One is an expansion of what the Intel platform means to business. And, the other is the need for platforms beyond the p rocessor. Think about Intel investments in networking and software. The other is, in effect, from the demand side. Is the nature of the Intel platform -- from the perspective of the enterprise -- changing?

BARRETT: Let me see if I can translate the question and answer it. We have lived with PCs for the last 15 or 16 years. The PC is exactly what it stands for -- personal computer. You start to translate that into the enterprise space, and you need to run through higher performance in the workstation space or into the server and truly beyond. You [have to consider everything] from bandwidth, to manageability, to networking, to this integrated commerce model. We are clearly on that track.

Our thrust is to continue to ride the PC wave as long as we can. But also, then, to work in the whole issue of enterprise, connectivity, manageability, commerce, cost of ownership, and everything that goes with that. The architecture is really much more if you are going to be in the enterprise server space. You are going to have to get out a signal processor, a sole processor, 4-by-8-by-16 clusters and that sort of thing. So, we are clearly moving the whole company and our attitude much more in that direction. And doing things like this Pandesic business that we have set up -- which is if you want to do integrated electronic commerce, front to back, using Intel architecture.

INFORMATIONWEEK: Speaking of Pandesic, I was talking to one of the customers, and he referred to Intel as his Internet service provider. Was he mistaken, or is Intel actually providing Internet service now, as well?

BARRETT: Well, in that instance the whole issue is, write me a check for x thousand dollars, and I will provide you everything you need to be that business, including the server bar in the backroom. In that particular instance, Intel is providing a service to Pandesic. It is really Pandesic that is the front and I am selling that service. Intel is providing the backroom capa bility behind it.

INFORMATIONWEEK: I assume you can also assure them a very high level bandwidth and storage, etc.

BARRETT: They probably would demand that like our own internal guys would demand it.

INFORMATIONWEEK: About electronic business, are you really hearing a lot in the marketplace, are customers trying to find innovative solutions? What are you hearing and how are you responding?

BARRETT: We respond in two ways. We have a bunch of people who are out in the marketplace talking to Fortune 500 companies about what their requirements are. Not so much selling specific solutions, because we are not an original equipment manufacturer that goes in and sells the computers and servers and that sort of stuff. But, we deal with our major customers, the OEMs, and interact with them and their customers. If we are going to bring the technology into the marketplace rapidly, we do that through our OEMs. So, we do a lot of joint calls and joint interact ion in the marketplace with the computer OEMs.

INFORMATIONWEEK: We want to talk a little about thin-client platforms and where X86 is going. Federal Express is in the process of procuring 75,000 [thin clients] now. There are others that are probably close to that same size order. Can you give me your thoughts of X86 serving that market?

BARRETT: This whole thing popped up with Mr. [Larry] Ellison, [CEO of Oracle], and [the] network computer days. Our industry is probably just as good as any for internal hype and throwing out ideas and seeing where they settle down and how they filter out, that sort of thing. We went from $299 network computers and replaced them with managed PCs and new clients. And I think that what is really coming out of that is the fact that you have got some 30 million dumb terminals out there that probably are going to be upgraded and replaced. The issue is what are they going to be replaced with? Are they going to be replaced with just the ultim ate dumb terminal which is kind of the network computer, or are they going to be replaced with a lean client, or a network PC, or a managed basic PC? I think you are going to see some combination of all of the above.

[Some] people are going to put in just enough capability and manage the cost without putting in any extra capability. You are going to see X86 solutions in those, as well as in the standalone, higher performance PCs. You will see probably a more dedicated processor to that environment than you would see in a high performance PC.

INFORMATIONWEEK: Do you expect to reach the same market-share level in that area as you enjoy in the PC market?

BARRETT: It sure would be nice, but I have no idea. That is another area with a lot of proprietary solutions with enterprise guys who have been in that space. But I think you ultimately start to see the same trends there that you see in the PC space -- the same trend you are seeing in the workstation and server spac e. There is an ultimate economy of scale in this business and economy of having architectural similarity for low cost of ownership overall, and whether Intel ends up with the same market segment share in dumb terminals or lean PCs or network PCs that we have in the regular desktop PC.

INFORMATIONWEEK: Are you now thinking of this cache- less Pentium II that people have been talking about for that lean-client market?

BARRETT: If you want to get into the relatively low cost of either standalone or basic PC environments, whatever you want to call it, sub-$1,000 PCs -- versions of either cacheless or integrated cache -- the lower-cost integrated versions of the P6 architecture will find a home in that space. It is referred to as the Pentium II today; that family will continue on and be upgraded.

[There's] the basic PC space, the network computer space, and then higher-performance versions ... up into the workstation server. You can see the whole market going into the th ree different segments with processors and chipsets and different levels of integration designed for each one of those categories. If you throw in mobile, you get a fourth category.

INFORMATIONWEEK: Speaking of mobile, what is your take on the whole CE market? Does Intel have interest there?

Continue...


Return to "Intel's Enterprise Push"

Send Us Your Feedback

Top of the Page

---