(FORTUNE Magazine) – There are executives who set lofty goals for themselves: turn their company into an industry beater, develop a world-changing technology, maybe even have a business dictum named after them. Then there is Craig Barrett. "When I get through with my tenure here," says Barrett, speaking in his soft but quick monotone, "I want people to say, 'By golly, Craig was able to keep Intel in the center of the action in computing: Intel is still a high-tech growth company; Intel is still one of the most valued companies in the world; Intel does great work; Intel is a great place to work.' That's the legacy I want."

Barrett is the fourth CEO at Intel. The first, Robert Noyce, were he still alive, almost certainly would have received the Nobel Prize this year for inventing the integrated circuit. The second, Gordon Moore, noted that the power of a computer chip would double every 18 months, and so created Moore's law. The third, Andy Grove, turned Intel into one of the most powerful companies in the world, made it a must-have stock for countless investors, and transformed microprocessors from intricate PC innards into brand items. For most of Grove's reign as CEO, you didn't buy a computer unless it had Intel Inside.

So by comparison Barrett's desired legacy sounds a bit, well, underwhelming. But look closely at Intel today, and you realize that his goals are actually incredibly high. The company that Robert, Gordon, and Andy built has taken a beating lately. First there's the threat from outside, where PC growth has slowed and new competitors like Transmeta have come on to challenge the chip giant. Then, and even more stunning, there have been the problems inside. In the past six months alone, Intel has recalled all of its swiftest Pentium III chips, canceled a planned low-end chip called Timna, shuttered an applications service provider it co-owned with SAP, and warned that it wouldn't meet Wall Street's third-quarter earnings expectations. And in a supreme ego blow, Intel lost the bragging rights for fastest processor to onetime has-been AMD in February--before regaining the lead with the release of Pentium 4 in November. Revenues, which spent the '90s growing at a compound annual rate of 25%, are likely to increase only 17% this year, to $34 billion; operating profits, which grew 33% a year, will likely increase only 11%, to $11 billion. And at least a few analysts think net profits may actually drop in 2001. Not surprisingly, Wall Street has pulverized the stock. Since hitting a 52-week high of $75.81 in August, Intel's stock is down roughly 45%. The S&P 500, by comparison, is down just 9% in the same period.

By anybody's definition, that is not the "center of the action." Which is why Barrett has been working so hard to redefine his company's own center. In his 32 months as CEO, he has spent $8.5 billion buying his way into new businesses in the fast-growing communications and networking markets--from making chips for routers and hubs to building the innards of cell phones to selling Intel-branded servers. He has moved the company into services, setting up seven Web-hosting farms--locations that store and deliver Website content--around the world. And he has begun dismantling the once rigid Intel hierarchy in which all decisions cleared Grove or didn't get cleared. None of this has been easy, as investors have noted. Indeed, if Barrett is going to accomplish his goal--and he's giving himself just four more years to do so--the ride may only get rockier.

On a beautiful fall morning in Santa Clara, Barrett is holed up in a small fifth-floor conference room in the building Intel calls its "head shed." He's a big man, 6-foot-2, with the build of the Stanford track star he once was. He walks fast and bowlegged. But there is little swagger in his talk. A conversation with Barrett can feel like having tea with a patrician senator. A recent interview started with a five-minute discourse on various types of huntable and nonhuntable animals on his 10,000-acre property in Montana. "I took up elk hunting a few years ago. I shot my big elk," he says, leaning back in a conference room chair. He's dressed in gray slacks and an open-necked shirt; a suede khaki jacket from Macy's hangs outside his cubicle. "I'm not sure that I'm going to shoot another one, but running over the hills and stalking elk and whitetail deer is just great fun. The property is full of mule deer, and you can't get a mule deer tag. So our mule deer are tame. You can drive up within 30 and 40 feet of them before they get off the ground." Then, after a few more minutes on the problems of the Forest Service, Barrett turns his attention to business.

This is the executive who's changing Intel?

Just down the hall is a man who exudes leadership. On the same day, Andy Grove is seated in his cubicle, dressed in a black mock turtleneck. In 1998 he moved from CEO to chairman, but he's hardly retired. He works 40-hour weeks, weighs in on important business discussions, scouts out new technologies, and acts as an in-house advisor. His secretary informs him that someone named Dan is on the line. Over the cube wall Grove screams out, "I called him two fucking weeks ago!" He then suggests she tell him that. (She doesn't.)

But ask around Intel, and you get the same message about the two: It's Barrett who's the more demanding, the bigger risk taker. " 'Push' is an easy word to describe him," says Mark Christensen, the head of Intel's Network Communications Group, deliberately using the noun instead of the adjective. "Craig is really pushing us for growth. He's pushing us to be aggressive, to take some risks."

Have years of being yelled at by Grove made guys like Christensen delusional? No. In his own way Barrett really is more pushy than the famously forceful Grove. But where Barrett is quiet, Grove is loud. Where Grove craves confrontation, Barrett cuts to the chase. And where Grove's temperament was right for the days when Intel had a single-minded focus on the PC market, Barrett's may be the thing Intel needs in its effort to remain not just relevant but as dominant as ever. "Craig is a more collaborative thinker," says one former executive. "He'll challenge your thesis and clearly force you to articulate your concerns, but he won't wave a wand and say, 'That's the way it is.'"

Ever since he took Intel out of the memory chip business in 1985, Grove has deployed his wand for one purpose: making Intel the No. 1 purveyor of microprocessors. And he got his wish. Intel's processors power more than 80% of the world's laptop and desktop computers. But in the past few years that goal has become less attractive. Consumers and companies have started to doubt the usefulness of upgrading their PCs every year. From 1994 to 1999, PC sales in the third quarter grew an average of 20% a year; sales slipped to around 16% this year. And analysts don't expect the corporate PC market to grow more than 10% in each of the next two years. One former Intel executive characterized the company's continued quest for dominance in microprocessors as "fighting over a graveyard" (albeit one with incredible 70%-plus margins).

The happening place right now, as anyone who has been alive in the past five years knows, is the Internet. Consumers and companies are trying to figure out better, faster ways to get access to the Net, to their data, to one another. Moore's law has been eclipsed by Metcalfe's law, named after Bob Metcalfe, the founder of 3Com. It loosely states that a network grows more powerful as more people connect to it. And as networks have moved front and center, chip sales have followed. Consider this: Chips for PCs accounted for 40% of all chip sales in 1994. Now they make up 25%. In that same period communications chips that can parse and direct data grew from 15% of sales to more than 21%. Under Metcalfe's law, it's how well the network connects people that matters, not how fast the PC works.

Barrett wants in. But tackling the Internet as a business proposition is much like navigating it as a user: There's no clear place to enter, no clear place to go. So Barrett has remade Intel to be more like the Net--less focused, yes, but much more adaptable.

Barrett has been Grove's designated successor since 1993. All operations ran through him, then to Grove. When he became CEO in 1998, he flattened the reporting structure and made it more competitive. He tapped five senior managers to handle various areas--networking, wireless, new business (think Web hosting), communications, and, of course, microprocessors--and told them to make their divisions No. 1 or No. 2 in each of their areas.

"None of us saw the Internet five years ago," says Barrett. "People ask me to forecast five or ten years out. I say, 'Jeez, the Internet economy is wonderful--it's going to continue to grow PCs and servers; vocal communications is wonderful--it's going to continue to grow flash memory [chips] and processors for cell phones; networking is wonderful because that's the backbone of the Internet.' So all of them are going to grow. I just want our new business to grow at 50% a year. I'll let the rest take care of itself."

It's a classic Barrett move: to push without specifying what to push hardest on. A former professor, Barrett prefers framework and goals to line items. In 1999 he assigned Intel CFO Andy Bryant the task of running all e-business operations but refused to define precisely what he thought e-business was. Bryant studied the issue, talked to people in and out of Intel, and told Barrett the job was much bigger than Barrett thought--that all of Intel would be e-business within two years. Barrett told him, "You have the mandate. Pull it all together." So now, in addition to being CFO, Bryant oversees everything from the system that handles Intel's $2 billion of monthly Internet sales to an online system that will allow workers to easily cash in their options. "When Andy was CEO, he had Craig as COO," says Bryant. "Craig has chosen to basically have a distributed COO. He wanted to develop a lot of different people on his staff."

Indeed Barrett says that while he will retire when he's 65, he has made no move to appoint a successor. Instead he appears to be setting up a GE-style horse race among his business heads. He says there are five in the running: Bryant; New Business Group boss Gerry Parker; Intel Architecture Business Group head Paul Otellini; Sales and Marketing manager Sean Maloney; and Technology and Manufacturing Group chief Michael Splinter. That's Craig Barrett's Intel: tireless, dogged, and, yes, very competitive.

It is, in fact, exactly like him. When Barrett travels he never stops moving. He lives in Arizona, where he works on Fridays and Mondays; he spends the middle of the week in Santa Clara, and at least a weekend a month on his Triple Creek Ranch in Montana. And that's only when he's in America. Since April, Barrett has been to 30 countries visiting heads of state, checking out technology, and meeting with overseas customers. He flies commercial, spends no more than 24 hours in a country, forces his staff to pack only a carryon, and expects everyone to have absorbed detailed packets of information on each country before entering. Inside Intel, the trips are called "death marches."

What's most surprising about Barrett's rise to the top of Intel is that his start at the company was such a failure. It was 1973, and Barrett was a professor of material engineering at Stanford. At 34 he had already written a textbook on the subject (called The Principles of Engineering Materials, it's still in use today) and was considered an ingenious problem solver. Morning coffees with colleagues invariably attracted graduate and undergraduate engineering students eager to hear him hold court. But Barrett wanted to move faster than university life allowed. So when he got a call that year from a friend at Intel looking to hire a consultant, he jumped.

Intel had a problem: A small ceramic package necessary for its memory chips was malfunctioning. When heated, the package produced moisture; when cooled, the interior turned into a tiny rain forest. Water was eating through the package's aluminum lines and wrecking the 1103 memory chip, which had accounted for 90% of Intel's $23 million in revenues the year before. Intel had already made hundreds of thousands of the packets. Barrett had to figure out a way to save the inventory.

Every time Grove, then vice president of operations, saw Barrett in the hallway he'd ask him how the work was progressing. When Barrett replied that it was coming along, Grove would point out that the problem was costing Intel $24,000 an hour, big money for a company then grossing about $66 million a year. "Grove took great joy--no, joy is the wrong word--pleasure, in saying it," recalls Barrett. His eventual fix--laser-drilling a hole in the packets, then baking the water out--ended up destroying most of the chips.

Barrett left Intel and returned to teaching. But the Intel folks were impressed enough to ask him to come back. ("It was actually a clever idea," recalls New Business Group general manager Parker. "Something you'd expect out of a Stanford professor, not out of a manufacturing guy.") Six months later Barrett returned. For good. He was 35.

He rose through the manufacturing ranks, but it took a decade for him to make his mark. In the mid-'80s the Japanese were clobbering Intel's main memory-chip business by producing more chips at lower cost. Intel couldn't keep up and got out, focusing instead on microprocessors. Now it was beginning to lose ground in that sector too. Barrett, who was overseeing the fabs, grabbed his factory managers, put them on a plane to Japan, and told them to figure out how the Japanese were doing what they were doing. He took field trips to places where manufacturing and process were important: Xerox, Epson, even some fish markets. He forced Intel to realize that there was a way to beat the Japanese, and that was by being just as rigorous as they were. "It was really refreshing to have somebody not talking about getting out of manufacturing," recalls Parker. "He did the unfreezing that got our manufacturing group to stand up and say we know how to make a factory instead of just lying down in defeat to the Japanese. He got them going. Craig brought a competitive can-do attitude instead of woe is us, let's outsource." Out of this came what's known inside Intel as "copy exactly": the idea that every Intel manufacturing plant should be an exact copy--from the paint on the wall to the temperature on the floor to the diameter of the pipes that bring water into the fabs. Intel was already a tightly controlled ship, with orders coming down from the top. Barrett set those orders in stone.

Grove took notice. In 1993, on the strength of Barrett's success in revamping manufacturing, he promoted Barrett to COO, signaling to the company that the CEO job was Barrett's to lose. "Craig had performed at everything that he had been asked to do, and on a very big scale," says Grove. "And as head of technology and manufacturing, he ran half the company anyway. So he had business experience with it. He is an immensely intelligent guy--kind of a strong, attractive personality. People want to follow him. He has big characteristics that I call leadership, which you can't really describe. Sometimes you can sense it in people. Gordon and I sensed it in Craig."

When he was tapped to fill Grove's position in March 1998, Barrett suddenly realized that he was no longer just working with a business genius; he was expected to become one. "I was nervous as all hell," he says. "Look at who you get to follow: Bob Noyce, Gordon Moore, and Andy Grove. So what do you do for an encore?"

The Intel board had similar concerns. When Grove started lobbying for Barrett, the outside directors called a meeting. Everyone agreed that Barrett had done well for the company. But Grove was a visionary who could outthink competitors; Barrett was a manufacturing guy. Barrett was copy exactly; Grove was crush entirely. "There were questions as to whether Craig had the ability to do the strategizing," recalls Charles E. Young, the president of the University of Florida and an Intel board member since 1974. "Was he strong enough and broad enough to be the CEO?"

Barrett got the job--and quickly dispelled any questions the board had about him. Almost immediately he started doing something un-Intel-like: buying companies. "It was an exciting time," recalls one former executive who spoke on condition of anonymity. "The thesis was that we were a $200 billion to $250 billion company [market cap]; how do we become worth $500 billion? Christ, if we had to go into the potato chip business, so be it." Wisely, Barrett decided to go where the action was: networking and communications. He knew it was doable. After all, Intel already had the manufacturing efficiencies and knew how to design silicon. It also had $10.6 billion in cash.

Barrett had a strong sense of where the industry was going. What he didn't have was any framework for buying companies. So he created an in-house team of key people and sent them to learn from companies that had mastered the art of acquisition, like GE, Microsoft, and Cisco. He also brought in consultants to work out a system for assimilating new companies and--in a typical Intel move--built a matrix for evaluating why the deal needed to be done and for measuring whether it had been done well.

In October, Barrett struck his first deal. He bought a networking company called Shiva. Then the acquisition team went on a rampage: In 1999, Intel purchased 12 companies for $6 billion, more than a third of which went to buy Level One Communications, a major supplier of networking chips. All told, Barrett has spent $8.5 billion buying 25 companies. "I'm sure that Andy looks at this and says, 'Wow, we didn't do this before. Are we getting our money's worth?'" says Barrett. "And quite frankly, we've been pretty aggressive, so I'm giving him stuff to worry about."

Indeed it's not entirely clear what Barrett's acquisitions binge will accomplish. The company doesn't break out individual financials for its new business. But last year, as a whole, those businesses lost about $1.6 billion on $4 billion in revenues. And this year the new businesses have contributed about $4.6 billion to Intel's $25 billion in the first nine months of revenues. By comparison, Intel's investment portfolio alone (which was started by Grove but expanded under Barrett) netted the company $3.3 billion in the first nine months. Nonetheless, Dan Niles, an analyst at Lehman Brothers in San Francisco, thinks the new businesses could account for 40% of Intel's revenues five years from now. That would translate into an estimated $32 billion.

But none of these businesses has shown any sign of becoming the next Pentium: a must-have, market-defining, dominating product. And there are questions about whether Barrett's doing enough to bring in new blood for new projects. After all, each of the business units is run by an Intel veteran: The newest one (communications products chief John Miner) joined the company in 1983; the most tenured (new- business chief Parker) in 1969. They may have trouble applying anything that veers from the old Intel way.

Take a look at two of Barrett's biggest bets: Web hosting and networking chips. Both show promise but have had rocky starts. When Intel first launched its networking chip in 1999, it arrogantly refused to join an industry group called CPIX, which was defining the interface for software on network chips. During Intel's rise to dominance in the PC arena, it made the microprocessors that defined the standard. Why let other, smaller companies tell them what to do? "When I was talking to people at Intel, their attitude was, Well, why should we participate in these organizations?" says Linley Gwennap, the principal analyst of Linley Group, a tech analysis firm in Mountain View, Calif. "But in the networking processor space, they're the new kid on the block." No one had to follow their lead, and no one did. This summer Intel, now the fourth-biggest network chip maker, relented and joined. (Christensen, the group's general manager, denies this account, saying Intel initially misunderstood CPIX's goals.)

In Web hosting, where Intel competes with such new-economy upstarts as Exodus and Digital Island, Parker has been sticking to Intel's old-economy playbook too. "We built this business on the Intel brand, with Intel values and Intel assets, and I think we can be successful with them," he says. "We had one manager who was quoted as saying, 'Bullshit, we're going to have our own culture here.' He's no longer here. We took care of that." But such heavy-handed behavior has taken its toll: In the first six months the turnover rate at Intel's data centers ran at about 25%.

"It's baffling to them. It's like, I'm bulletproof--how can I be bleeding?" says Howard Anderson, the founder of research firm Yankee Group in Boston. "I give Intel a C-minus on what they've done so far outside of chips."

Even with chips, Barrett hasn't been pulling straight A's. In February, Intel lost its claim to producing the fastest chip on earth to AMD. This damaged more than just Intel's psyche. In June 1999, AMD's chips sold for an average of $72, while Intel's commanded $198 on average. By last September, AMD was charging $91 and Intel $182, says Jonathan Joseph, an analyst at Salomon Smith Barney. AMD's newfound competitiveness is putting pressure on Intel's ability to price its chips.

Barrett can ill afford losing ground in PCs: The Intel Architecture Group, which makes computer components, still accounts for about 80% of the company's $34 billion in revenues and virtually all of its operating profits. Otellini, who heads that division, blames his group's recent problems on poor memory-technology choices and the fact that his group had become "kind of overstretched." By the time Intel announced that it would miss third-quarter estimates, employees in Otellini's group were wandering around, in his words, like "deer in the headlights." "I think there were a lot of employees just kind of stunned for a couple of days," he says. "And the reaction, as I walked around and talked to the employees in my part of the company, is that people have gone from kind of this stun mode to getting angry and getting energized. They're wanting really to ensure that we bounce back. And I think that's a very powerful motivator."

Through all of this--the mistakes, the setbacks, the headlong leaps--Barrett says he's relearned the importance of setting goals. This fall he has held one-on-one meetings with the heads of all of his divisions, telling them to check through the ranks of their employees to make sure everyone undergoes Intel's rigid training schedule. He has told them to make sure they are following the fundamentals of decision-making: plan, do, check, act. And he has tied bonuses to performance in each group. "My job is to refresh in everyone's mind: This is how we do projects. These are the exact steps you go through. Don't take shortcuts," he says. "We are refocusing internally on our best practices." One thing he hasn't done is abandon his policy of delegation. It will still be up to the managers, not to him, to see that they are performing.

These are big changes at Intel. But will they be enough to allow Barrett to claim his place beside Intel's original three? Barrett reiterates that that is not his goal. "I know what my job is, and it's not to be Andy Grove or Bob Noyce or Gordon Moore," he says. "It's to make sure Intel is going on the successful track that it's been on."

That goal seems to suit Grove just fine. In Barrett's cubicle, just 20 feet down the fifth- floor hallway from Grove's, sits a framed picture of Grove and Barrett dressed in the famous Intel bunny suits. The inscription reads, CRAIG, CONGRATULATIONS ON YOUR PROMOTION TO CEO. I'M GLAD TO SEE THAT YOU'RE FINALLY MAKING SOMETHING OF YOUR LIFE. It's signed ANDY.

FEEDBACK: droth@fortunemail.com