Intel Corp. has set the pace in the microprocessor industry, cranking out chips that operate at breakneck computing speeds. So when the company, under the direction of CEO Craig Barrett, began steering away from its speed-fiend image and toward multicore and partitioned designs, people took notice.
For some, it was inevitable, as Intel had been enduring a string of missteps including the unceremonious cancellation of its 4-GHz Pentium “Prescott” chip this month. It was as if the Santa Clara, Calif., company had overestimated its market clout and its ability to understand the needs of its IT users. With just six months left before he is scheduled to cede the CEO position to Intel president and chief operating officer Paul Otellini, Barrett said the company is heading in the right direction.
In an interview with Executive Editor Stan Gibson and Editor in Chief Eric Lundquist at Gartner Symposium/ITxpo in Orlando, Fla., last week, Barrett explained the companys recent strategy shifts and his approach to the final six months of his tenure.
The cancellation of the 4-GHz Pentium 4 is symbolic of a shift from faster clock speeds to different architectures, such as dual core. What event raised the flag that you had to move in a different direction?
Paul Otellini said a couple of years ago at the Intel Developer Forum that theres more to life than gigahertz. Prescott [4GHz] was the tail end of that architecture. It gets more and more difficult to crank that performance out, and you have to look at other enhancements.
So, it wasnt a sharp turn in the road?
Internally, we refer to it as “the right-hand turn.” The whole world was blasting along in the gigahertz war. But the industry as a whole has recognized that with Moores Law, you cant continue to run more transistors faster—the power [heat] dissipation just gets to be too great.
We can continue to run more processors, but in a different fashion. Thats the right-hand turn: using those transistors to create other technologies—whether its virtualization, security, multiple cores or multiple threads—and to use those to bring more performance to the end user.
Does the microprocessor road map embrace that?
Yes. Everything moves off dual core—or multiple core—and multiple threads in the future. In addition, the architectures are designs that are the most power-efficient. What are the most power-efficient architectures today? Pentium M architectures a hell of a lot more power-efficient than Prescott. So were going to do two things: build off the most power-efficient capability and go to multiple cores and leave the speed game on the side.
Are your target dates for the dual-core processors holding up?
We havent been very specific about them. We said dual core in all aspects of the business—desktop, laptop and server—will be ramping up by 2005-2006. But we really havent been more specific than that.
Hewlett-Packard Co. canceled the Itanium workstation. Your thoughts?
Itanium has turned out to be a very viable competitor in a section of the market: the enterprise server—big iron. It competes with SPARC, it competes with [IBMs] Power. Its replacing [HPs] Precision Architecture and Alpha. It is a big-iron machine.
Its pretty clear that 32-bit architectures have moved up, but I dont think anyone looks at the 64-bit address extensions as making those 32-bit devices as viable enterprise servers from the big-iron standpoint. But some applications—the front-end apps, Web servers, HPC [high-performance computing] stuff—those address extensions make sense when youre addressing huge memory spaces.
How fast do you see the growth of the market for the EM64T?
In a couple of years it will be a checkoff item. Whether anybody uses it for anything is an entirely different question. There are over 100 million desktops sold each year.
In a year or two, it will be a standard checkoff item in desktops—as soon as Microsoft [Corp.] has an operating system that can take advantage of it, and everybody has compatibility to take advantage of it. Will you put that much memory in your desktop to take advantage of it? Will the drivers and apps be written to take advantage of it?
What do you think?
Thats a hell of a lot of memory to take advantage of. I think it will be pervasive, but so what? There are few applications that take advantage of that much memory. Itll be used there.
Wireless technology is pretty exciting, particularly in fixed wireless broadband, as you said in your remarks here at the conference, but how big a revenue opportunity is that for Intel?
It depends on how pervasive it is. If in fact the vision that we have that all laptops have a collection of protocols integrated in them—Wi-Fi, WiMax and 3G—then [wed be] selling one of those [chip sets] per [machine].
Do you integrate them all in one device? If I have a device on the side of my house, in WiMax, but Wi-Fi in the home, then you have a wireless connectivity capability in each PC. We make those chips. Is that as big a business as the [server] processor business? Hell, no. Its much smaller, but its a growth opportunity.
Do you see that WiMax box replacing the DSL line?
As long as youre not relatively close to a central office, theres a good chance of that happening.
-Up Call”> With regard to Itanium, you mentioned that everyone but Sun [Microsystems Inc.] and Apple [Computer Inc.] is selling Itanium. But HP may be selling 90 percent of all Itaniums. Do you need to do better with other OEMs?
Sure. Im not agreeing with your numbers by the way. Theyre off. Weve got IBM, Bull, all the Japanese guys— something like 80 systems by the end of the year, from a variety of different OEMs.
Your famous memo to employees of a few months back, was it successful? Are you seeing the kind of behavior you would like to see?
I wish that I hadnt had to write it, but I think it served as a useful wake-up call. All our organizations within Intel have taken the issue of performance seriously. They all have plans in place and are all executing those plans. I think the memo served a useful purpose.
So things are where they should be at this point? Theres no need for a follow-up memo?
A follow-up memo to that one would be—unfortunate [laughter].
Are you going to become chairman when Paul Otellini takes over?
It would be presumptuous of me to announce anything before the board of directors of Intel does.
What are you going to work on in the next six months to leave behind a legacy to Paul?
Exactly what Im working on today: the growth of our communications business, the convergence of computing and wireless, expansion internationally, and process technology manufacturing leadership. The strengths of Intel are process technology, manufacturing, architecture, worldwide sales presence—brand—and Intel capital. Those are the five things Im working on today, and those are the five things Ill be working on for the next six months.
In your remarks, you mentioned that enthusiasm for IT is greater in all the countries you visit than in the United States. Theres less interest in the country that is driving [IT innovation]. Where did the interest fall off?
If anything, the interest and enthusiasm got diverted into the biomedical area. If you look at NIHs [National Institutes of Healths] budget over the last five years, it has about doubled. The real challenge there is: Lets say my child has cancer—that is powerful. Its life. Its your family. Its trivial to get funding for that, and its nearly impossible to get funding for anything else.
I anger all my buddies in the agricultural states when I point out we can put $25 billion in agricultural subsidies, just like that, without even thinking. But what is going to drive the economy? Is it wheat or corn? Its kind of depressing not to have either Bush or Kerry talk about it—neither of them. You go to their Web sites and you get pablum, like, “Im going to make the United States more competitive.” Oh really?
But will there be some sort of event to make people rearrange their priorities?
You have signposts. You have a road map already. Western Europe is your road map. You see the slow, persistent decay of competitiveness—in France and Germany in particular. It has resulted in essentially stagnant growth and double-digit unemployment. … Im not asking that the government get into the technology management area. K-12 education is a ticking time bomb. Thats the greatest danger the United States has. Second, fix the R&D investment. Get aggressive about infrastructure. It took Bush three and a half years to say “broadband.”
And then recognize that the rest of the world is treating investments in technology as investments in the future. And the United States treats investments in technology as corporate welfare.