Craig Barrett's exit interview

Craig Barrett will retire at the end of May.

More from Fortune

Will Mmmhops be a hit? NBA confirms L.A. Clippers sale to ex-Microsoft CEO Steve Ballmer FBI and SEC probe into Carl Icahn and golfer Phil Mickelson

FORTUNE 500 Current Issue Subscribe to Fortune

(Fortune Magazine) -- At the end of May, Craig Barrett, the chairman and former CEO of Intel and avid horseman, will ride off into the sunset.

The laconic, 69-year-old former professor spent much of his 35-year career masterminding the transformation of the storied semiconductor innovator into a manufacturing juggernaut that dominates its industry. Gordon Moore, the Intel (INTC, Fortune 500) co-founder, may have discovered the "law" that chip capacity and performance could double every 18 months, but it was Barrett and his factories that fulfilled that prophecy year after year.

Barrett succeeded the legendary Andy Grove as CEO in 1998 and became chairman in 2005 when Paul Otellini was named CEO. He sat down for a 90-minute conversation with Fortune, and reflected on the good, the bad, and the beautiful aspects of his time in the saddle.

FORTUNE: There are only a handful of business leaders in the world that have the chance to affect the world in such a profound and immediate way as the CEO of Intel can. Perhaps you don't think of the company in this way, but isn't it an amazing tool that you held in your hands?

BARRETT: Well, the neat thing about Intel is it's a great demonstration of what smart people and smart ideas can do. And I think it's a good blueprint for the U.S. and what the U.S. ought to be doing.

We have ongoing debates about competitiveness with China and India and the rest of the world now that we're all a bunch of capitalists. And very quickly you get down to the point that there are only three levers anybody can pull to do anything.

One of them is the education lever - that's where the smart people come from. If there aren't enough smart people, game over. The next one is smart ideas - and that's investing in R&D. If you don't do that, you don't add value. And the last one is to provide the right environment. In the company, that's getting smart people with good ideas together and just letting them do their thing.

And at the national level, the right environment involves tax rates and regulatory climate, encouraging the availability venture capital and building the right social attitudes. All that good stuff. And unless the U.S. can do that, our value-added goes down, our standard of living goes down, everything goes down.

That sounds like trying to create a world that is a little bit like Lake Wobegon, where all the kids are above average. I suppose that at Intel, everybody is above average, but in the greater U.S. or in the world that's just not the case. Can you really apply the lessons that a disciplined and rarefied company like Intel has learned to the larger, more chaotic world?

Well, I can't argue with your statistical model. Half of the people are above the median, half are below, probably. But it's pretty clear that your ability to add value is related to your education. And if you don't have a good education system, you're toast.

If you look at the U.S. for the last 20 or 30 years, we've had what is basically a bi-modal education system: We have a great university system that's capitalistic and merit-based; that competes for people, competes for ideas, competes for students and faculty members and research contracts.

But we've got a monopoly public K-12 system that doesn't compete for anything. So if there's a "Lake Wobegon syndrome" it's in that K-12 tier. You can't find a governor or state school superintendent who would say "Our kids are below average." This is despite the quantitative evidence from international tests that all say U.S. kids are below average. So we fool ourselves.

Smart people used to flock here from the rest of the world for our universities and the economic opportunities. But now, increasingly, they're either staying home or they're coming here to get their higher education and some experience and then go home. Every survey says kids from India and China see more opportunity in their home country. So we're losing that ability to be the magnet for brains. It's a tough time.

Still, I think there is a direct correlation between the national challenge and the corporate challenge. International corporations can't be successful without really smart people and smart ideas. The U.S. can't possibly be successful without smart people and smart ideas.

Let's talk about Intel's unprecedented dominance with what's now called the Pentium family of microprocessors. That chip family transformed all of computing and continues to be the underpinning of most of IT.

The original chip came out 30 years ago, about five years after you joined Intel, and you were instrumental for most of your career in making it a success. That's quite a legacy, but is it a once in a lifetime experience for either Intel or your industry?

I don't know if it's once in the lifetime of the universe, that's a pretty long time. I think if you partition this stuff down, about every 50 years a technology comes along that reshapes the earth. And whether it was the automobile or the airplane or the computer, clearly the last 25 or 30 years it's been the PC and the Internet that have reshaped everything around us.

So, we were fortunate to be associated with the thing that has swept the world during our time here. And then we just continued to invest in that. If you go back over the Intel history, you'll see how we relentlessly followed Moore's Law, and we relentlessly cannibalized our own products, and we relentlessly moved the technology ahead. All of that helped to make this snowball get bigger and bigger and bigger with more momentum and made it more difficult to shift or stop. The only other example I can think of in the high-tech industry is printer ink from Hewlett Packard.

Probably the most surprising thing in our industry - and golly, if you talk to Gordon Moore, he amplifies this. We never thought Moore's Law would last. We never even thought you should call it a "law." I wrote a paper about it in '65, and said I thought we'd be off that track by 1970, but then it just continued and continued and continued. And even today we can see it going at least another 15 years.

This is ridiculous. It's going to be 60 years - 60 years of Moore's Law. Without a hiccup. Tell me anything that matches that. And beyond that, the significance of Moore's Law is not just that it gives you more, but that it's anti-inflationary. It's more for less. More functionality, more capability, for the same or less cost. How can that be?

Intel has been trying with only marginal success to become a dominant player in wireless and mobile communications technology, compared to the amazing Pentium phenomenon. Why is that?

If you look back through the '90s, we were growing at 30 or 35 percent a year, top and bottom line. That's a pretty damn good business, and we were struggling to build factories and keep up with Moore's Law and get the next generation Pentium architecture out there. We pretty much had our hands full there, as opposed to bifurcating and to say, "Okay, half the company, you go off and chase this cell phone business." By the way, the processors in the cell phones were selling for 5 or 10 bucks. Meanwhile we were selling millions of PC microprocessors for $150 apiece. So it wasn't so difficult a decision where to focus.

Now, in the process, the volume of handsets grew fantastically as well. So a whole new ecosystem grew up in that space and our industry ended up with two big areas of integrated circuitry usage: Computing, which we were doing pretty well in, and communications and the cell phone. And then there were lesser markets like consumer electronics and industrial stuff, which weren't quite as exciting.

So we single-mindedly focused on the computer. We did try a little bit back in the early 2000 time frame, when we bought the Digital Equipment semiconductor operations. They had something called the Strong ARM processor that we thought we could use to compete back into the handset business, but we made a couple of assumptions which were not particularly good. First, we thought 3G network speeds would happen faster than they did, and so we targeted that transition. That didn't happen, so we were left out to dry. We also made a bad assumption thinking that it made sense to integrate flash memory with the controller and it turned out that that combination was not a good technical decision.

So we made a couple of bad assumptions that hurt us in that space. But we were still focused pretty much single-mindedly on, "Hey, what's the next generation of the X86 [Pentium] architecture, and how do we get a new generation out every couple of years?"

So what markets represent the next likely big opportunities for Intel?

The areas that I think are exciting going forward is, one, you do see kind of a collision between the handset and the laptop. There's a huge opportunity in that space for full Internet functionality in a decreasing form factor. You'll see super small laptops or mobile Internet devices getting smaller screens. You'll see smart phones getting even smarter and getting somewhat bigger screens. And so I think there's a huge business opportunity there.

Part of that's driven by form factor, part of that's driven by the web of connectivity that we have around us progressing to 3G and now going to WiMax or LTE, whatever protocol you want. Full broadband wireless connectivity increases the functionality of all these devices. That's one big area for us to continue to focus on.

I also believe the consumer electronic devices are finally getting to the point where you're going to really integrate the Internet into those, too. You know, there are some areas in our industry where we talk about for ten years and not a damn thing happens, and then one day, all of a sudden, the whole thing toggles. And I think you're starting to see, oh, it's about time to get all of this Internet capability built into these home entertainment systems and cars. So there's a huge place for us to move the architecture.

You've always placed big bets investing in more and better manufacturing capacity in down cycles. Perhaps they should call it Barrett's Law?

You can't save your way out of a recession, you can only invest your way out. Bob Noyce and Gordon Moore said that and Andy Grove and I have said that, and we're doing it again now. Paul [Otellini] made the announcement a few weeks ago of Intel's $7 billion investment in the next generation of manufacturing during a terrible recessionary time. We're going full guns ahead. We 're going to continue to invest, and here's our commitment to Oregon and New Mexico and Arizona and so forth. That's a feature size of 45 nanometers today, but it's 32 tomorrow, and it's going to be 22 two years from tomorrow, so we're going ahead driving Moore's Law. And to the rest of the industry, we say "Please keep up with us because this is what's coming."

Is this basically the same thing as stimulus spending by the federal government to get the economy back on track? Is there any parallel at all?

People always think you're crazy, but lately I've found a greater receptivity to that message. I was at a technology conference a month or so ago and some folks from Australia were telling us how that country is putting $20 or $30 billion into broadband infrastructure even in this bad economic time. They have a population of 25 million people, which is smaller than California, so this is a massive project, but it's an investment into the 21st Century. And I have found just about every government is pretty receptive to this. So I'm hopeful we'll see more because investments like that pay huge economic dividends.

In our industry, however, we have no choice but to invest in good times or bad. If we miss a development cycle we pay for it. Actually we screwed up back in the '04-'05 time frame. We single-mindedly focused on putting higher performance in the laptop or the desktop PC and prioritized it way ahead of power consumption. But once you do that, battery life suffers. So we left the door open to another class of architecture to come in, which is focused on low power consumption as opposed to the highest possible performance. We actually shut down work a generation of chips to refocus on power consumption.

Well, it took a couple years before we got another architectural refresh. We were on a four-year cycle for new architectures, and we had to extend it to six years. And that's when AMD (AMD, Fortune 500) got to be their most competitive with us. You know, they came out with some good products, and at the same time we kind of were trying to extend the old architecture a bit farther. And that really hurt us competitively.

It's easy to talk about it looking back, but it was not easy to live through it at the time. We got back on track, though. Our latest product introductions I think blow them away, and this is an industry where the best technology wins.

That sounds like what Andy Grove would call an "Oh s**t" moment. Every CEO faces them, of course. Any others stand out during your tenure at Intel?

You know, most of the more recent "oh s**t" moments have been probably associated with the unbounded euphoria with the technology in 2000. It was an "oh s**t" moment when we woke up one day and realized that in fact the total throughput of the Internet was not going to double every 30 days and all of this talk about unbridled needs and capability was absurd, and suddenly the PE ratios of 60 disappeared. You said to yourself, "Oh s**t, I should have realized that!"

We also did some acquisitions and investments in '98, '99, and 2000 that violated the basic precept of buying low and selling high. We bought high and sold low. Those were definitely "oh s**t" moments.