(FORTUNE Magazine) – Sam Palmisano has reason to feel good. Two years ago he took over one of the biggest jobs in American business and, along with it, the mantle that had been worn by now-legendary CEO Lou Gerstner. In his nine years at the top, all Gerstner had done was save IBM from self-destruction, change its focus, rebuild it, and expand the business by 40%. That would have been a hard enough act to follow. But when Palmisano came in, outside events made it even tougher. In early 2000, as Gerstner began handing him the reins by making him president and COO, the dot-com bubble burst. By the time Palmisano became CEO, revenues were down a whopping $5 billion and still declining. He found himself steering a technology company in the worst tech downturn in memory.

Things appear to be straightening out dramatically. Last year Palmisano's company grew for the first time since 2000, posting a 10% revenue jump. Most of that came from favorable currency translations, but even adjusting for the windfall, IBM nudged up 3%. Profits from continuing operations grew 43%, to $7.6 billion. And Palmisano managed to grab market share in all of IBM's major businesses--the company now trounces HP and Sun in servers, out-databases Oracle, and is taking market share from Accenture, EDS, and HP in IT services.

Such results may not be the stuff that makes CEOs into superheroes, but Palmisano is hardly finished. Leaning forward in the leather and chrome chair in his Armonk, N.Y., conference room one chilly spring day, he lays out a bold plan for his IBM: He is committed to growing the company at least 5% a year--with double-digit profit growth. He also wants IBM's return on invested capital--a telling measure of how well a business puts shareholders' money and long-term debt to work--to significantly exceed the S&P 500 average.

The profitability targets are substantial, but let's dwell for a moment on his goal for top-line growth. Does better than 5% sound meager for a tech company? Just consider: It means Palmisano is proposing the equivalent of generating one FORTUNE 500 company every year. If he can meet the low end of his goal this year, for instance, he'll generate $4.5 billion in new revenues--a figure that, if posted on its own, would instantly rank 390th on the list. And by the end of 2005, IBM would become IT's first $100-billion-a-year company. Sure, Lou Gerstner had to make IBM grow--but he never had to do it on the scale that Palmisano is now promising.

Obstacles larger than just the law of large numbers stand in his path. For one thing, the IT industry remains sluggish; few experts expect it to grow much more than 4% this year or in the near future. Meanwhile, the competitors that Gerstner faced in the early '90s are bigger, and battle-hardened from winning in the tech boom and surviving in the bust. HP, the closest thing IBM has to an across-the-board rival, is now, at $73 billion a year, nearly two-thirds larger than it was in 2001 and appears to have successfully digested Compaq. And there's Microsoft: At $32 billion a year in sales, the software Godzilla is dealing with a slowing growth rate but has a hoard of $56 billion in cash, 28 times as much as when Gerstner arrived. Attacking from Austin, Dell keeps making it tougher to sell hardware, hawking more and more sophisticated products at margin-busting prices and partnering with IBM rivals like EMC and Oracle to move deeper into large enterprises. It grew 17% last year, faster than any other big tech company, to more than $41 billion in sales. At that growth rate, Dell would be bigger than today's IBM in a mere five years.

Palmisano has personal competition too. In a star-driven industry like tech, each of his major competitors has one-name recognition with investors: Carly, Gates, Dell. Maybe that partly explains why Wall Street so far has yawned when it comes to IBM. Since Sam took over, IBM's stock has slipped 14%, vs. the S&P's 1% decline. (In that period Dell has risen 33% and HP 5%, and Microsoft has dropped 15%.) At a recent $89, IBM stock remains 36% below its 1999 highs. So add to Sam's list the need to earn a little respect.

Think that's enough for one CEO to tackle? Well, Palmisano has heaped one more huge demand on himself--and this one's intense and personal. He wants to return IBM to the dominant role it once played in the tech industry. Palmisano speaks unashamedly of making IBM as great as it was when he joined 31 years ago as a salesman in Baltimore. In those days nobody could touch Big Blue. It had essentially invented business computing and was the kind of company that could "set huge objectives that others thought were absolutely crazy" and then meet them, says Jim Collins, the author of the management bestseller Good to Great and a longtime student of IBM. Palmisano knows that for his 93-year-old company to again become the most important company in tech, it must do more than sell ever greater quantities of servers and software--it must lead the industry to new sources of growth.

Palmisano's strategy is to expand tech's borders by pushing IT users--and entire industries--toward radically different business models. The payoff for IBM would be access to an ocean of potential revenue--Palmisano estimates it at $500 billion a year--that technology companies have never before been able to touch. Those dollars now are claimed by internal human resource departments, hoarded by marketing and product design folks, and spent trying to keep track of customers. Palmisano thinks such operations should be outsourced to IBM, which has the brawn to handle them more efficiently and, more important, the brains to remake them altogether.

The results of Palmisano's efforts have just begun to show. "Give us your toughest problem," he told one of America's largest retailers last year. (Palmisano won't name the company, but a good guess would be Wal-Mart, a big IBM customer. Wal-Mart declined to comment.) The CEO and CFO came back, he says, with "trying to get more revenue per shopping cart." IBM assembled a team of management and operations consultants, along with software experts from its research labs, and dug into the problem. Among the team's insights: Customers in superstores are getting lost in the aisles, wasting time navigating instead of spending. Now IBM is testing smart shopping carts with built-in scanners and LCD screens. The carts display maps, keep running tabs, and offer specials on products related to those placed in the cart--both guiding shoppers and egging them on to buy more as they move through the store.

With other customers, Palmisano's IBM has moved beyond just taking over data centers to occupying turf previously controlled by management consultants and internal staff. It persuaded British Petroleum, for instance, to hand over its entire finance and accounting function worldwide. The data from the $236-billion-a-year oil giant now reside in IBM computers at seven processing centers around the world, including Tulsa, Lisbon, and Bangalore, where IBM employees handle BP's accounts payable and receivable, and its financial reporting. In Rochester, Minn., IBM has formed an even closer partnership with the Mayo Clinic in a project the clinic describes as a "long-held dream" to remake the practice of medicine.

Such moves are a good start. To make them the rule rather than the exception in the IT world, Palmisano must get IBM to act with the coherence and speed he says his customers now demand. That's dauntingly difficult. Former CEO Gerstner likened his efforts at turning around IBM to teaching an elephant to dance--and he pulled it off. Now it's Palmisano's turn. And he needs to teach IBM a whole new set of moves.

The first thing most people notice about Sam Palmisano is his glasses. Big and owly, they look as though he hasn't changed them since the 1980s. And he hasn't. This is a man so comfortable with himself that he doesn't see the need to get rid of something that's worked for him for decades--he laughs off those who suggest he should. His aw-shucks nature, coupled with Palmisano's ability to chat up just about anyone he meets, makes him approachable for customers and employees. People inside and outside the company call him Sam. Countless people have stories of Palmisano pulling them aside--regardless of their rank--and asking, wide-eyed through those Harry Potter glasses, "How do you think we're doing?"

And as everyone inside IBM knows, he cares intensely about the answers. He's constantly on the phone, calling all over the world: "How's your quarter?" "Did we close this deal?" "How do we fix it?" IBM sells an instant-messaging product called Lotus Sametime; Palmisano pings people on it so frequently that colleagues refer to the system as "Samtime." Software chief Steve Mills calls Palmisano an "execution maniac." Others at the company use the term "executional wizard" (which could explain the glasses). This single-mindedness about results is a big reason Palmisano was selected by Gerstner to take over IBM two years ago. Says Merrill Lynch security analyst and IBM booster Steve Milunovich: "Sam is the right guy to run IBM right now. He's great externally and a hard-charging Marine internally."

Palmisano's style is a big departure from that of the gruff and intimidating Gerstner. But then Gerstner's role wasn't to be nice; it was to keep IBM from disintegrating. He took over just as it was about to split itself up into 13 distinct, loosely affiliated entities--an effort by the previous CEO to gain favor with Wall Street, which viewed IBM as a big has-been, clipping coupons from its mainframe operations and bleeding business to more focused rivals. Gerstner, a former IBM user from American Express and RJR Nabisco, didn't believe a breakup would benefit customers and instead set about getting IBM's divisions to work as one.

His most important insight: that services could more than make up for inevitable declines in mainframe computer sales. He and Palmisano, who ran the services operation in the late 1990s, built up a force of 180,000 services people and marketed them like mad. The plan worked. While hardware accounted for 49% of revenue in 1993, the year Gerstner arrived, ten years later it represented only 32%. Far from shrinking, IBM grew: Today it is 40% larger than when Gerstner took over. The services unit alone rakes in $43 billion in revenues--making it bigger than any freestanding company in technology except Hewlett-Packard. Boasts Palmisano: "We have essentially replaced one set of businesses with another, without tanking the company."

Gerstner was unquestionably one of the great CEOs; no company of this scale has ever pulled back from the brink the way IBM did. But in Palmisano's eyes, Gerstner's most lasting legacy won't be the spectacular turnaround. "That is in a way temporary, because I can go screw up the company, or somebody else could," says Palmisano. "What will be most sustainable over time is his strategic approach, which got IBM focused on the client."

That last word is important. A man who obsesses over his speech--in two long interviews he painstakingly defined his terms and often stopped to correct himself--Palmisano has carefully picked the word he wants to use when talking about the source of IBM's revenues. It's "clients," never "customers." In the legal or accounting worlds, gaining a client can mean earning fees from him for life. Palmisano seeks to pull off the same gambit in computing. He wants those who hire IBM to think they're getting a company that plans not to sell to them but to transform them. And even better, to help them stomp the competition.

Palmisano's strategy is straight out of IBM's history book--it builds on IBM's decades-old ties with the world's biggest companies and cements them even further. A mere 100 customers--mostly global corporations--represent a full 20% of IBM's revenues. Several, like American Express, Procter & Gamble, and Wal-Mart, spend as much as $1 billion a year. In a marketplace cloven between high-priced customized IT and high-volume commodity computing, Palmisano is willing to let Dell take the bottom of the market and sell commodity technology; he sees IBM at the top. From there, he aims to apply IBM's tech-driven strategies and to couple them with answers from its research labs to get deeper into the workings of companies than ever before. The results? Well, as the effusive Irving Wladawsky-Berger, IBM's vice president of technology and strategy, would have it, "It's nothing short of the bloody transition from an industrial society to an information-based one."

A clue to how Sam means to foment this revolution can be found in his elegant and spacious conference room at IBM headquarters. A picture window shows panoramic views of forests and hills. Perched prominently on a marble-and-oak credenza next to a bowl of candy for Palmisano's sweet tooth is the only photograph in the room: a shot of the CEO standing next to a smiling Linus Torvalds. He is the antiestablishment Finnish engineer who originated the Linux open-source operating system and who is most responsible for driving down the price of software by offering it free. So what is a guy searching for growth and greatness doing posed next to someone committed to free and easy?

The answer is that for IBM, Linux offered a way to go from being the bad guy in computing to being the company that just wants to help everyone get along. On the day the photo was taken in January 2001, Palmisano, then president, announced that IBM would invest $1 billion in Linux. Suddenly his salesmen had a disarming rejoinder for customers who were sure IBM was just trying to lock them into buying IBM forever. With Linux, Palmisano could tout IBM's commitment to openness--and have people believe it. There were other benefits, of course: Linux is a capable operating system that allows similar applications to run across the range of IBM's hardware lines. And as a Windows alternative, it threatens archrival Microsoft's most important source of income--not such a bad thing.

Palmisano felt he could use IBM's brains and its vast line of products and services to help transform businesses, but he was still missing one ingredient: the people to bring the pieces together. So in mid-2002 he announced the purchase of PricewaterhouseCoopers Consulting for $3.9 billion--the biggest acquisition in IBM's history. (It was a dot-com-crash bargain; HP had held talks to buy PwCC less than two years earlier for a reported $18 billion.) Now Palmisano could sell to customers the idea that IBM has what it takes to help remake their organizations, not just run their IT. He combined the 25,000 more IT-focused consultants in IBM's Global Services arm with the roughly equal number from PwCC--there were 3,500 layoffs--and created what's called IBM Business Consulting Services, a $13-billion-a-year business that makes up almost a third of the overall services group. PwCC was organized around the industries it served; increasingly now, so is IBM.

Marrying the consultants with the sales guys, eggheads, and engineers has already started to pay off. To get a peek at how this choreography works, it helps to have an investment portfolio that is out of whack and an account at Charles Schwab. The brokerage house offers free assessments of customers' retirement portfolios; its stockbrokers use Forecaster, powerful software that combs through investment choices and suggests personalized ways to rebalance assets. Forecaster recommendations were supposed to inspire increased trading and commissions, but instead the software was turning customers off. So complex is portfolio assessment that the number crunching could take hours; often results weren't available for the customer until the following day.

Schwab needed to speed up Forecaster. It called Steve Grove, the IBM sales exec assigned full-time to the company. Grove put together a team of experts from the software and services groups, who hit on the idea of organizing all of Schwab's computers into a so-called grid that could be managed as if it were one giant unit. The benefit: Applications like Forecaster would no longer depend on specially dedicated servers, which would bog down when things got busy. Instead, programs could automatically grab the computing power they needed from anywhere in Schwab.

Since Schwab was to be one of the first commercial customers to install a grid, Grove brought in IBM's R&D eggheads, including Willy Chiu, who heads a product-development lab in the software group. He collaborated with Maggie Archibald, a top IBM software engineer who works full-time at Schwab as part of IBM's sales organization. She in turn consulted extensively with engineers from IBM Research. And they worked with Anthony Karimi, a business-strategy expert from IBM consulting.

As a result of all this effort, Forecaster now runs on the same machines that manage normal stock trading, borrowing capacity when the machines aren't using their full processing power. A Schwab broker can get recommendations from Forecaster in 20 seconds, while the investor is still on the phone. That generates significantly more trading revenues. Says Schwab chief information officer Geoffrey Penney about IBM: "They're trying to look into their bag of tricks, understand us and our priorities, and bring all that together."

Aggressive, on-the-fly team building merits a special label in Palmisano's word-obsessed IBM--it is a "solution." IBMers fight to become part of solutions because that proves they are in sync with corporate strategy. Solution status isn't easy to come by: To qualify, a project must involve IBM's Business Consulting Services, software, hardware, and a software application from one of IBM's partners, like enterprise software giant SAP or sales software leader Siebel Systems. So far about one-third of IBM's sales to the financial services industry, which totaled $22 billion last year, are officially "solutions."

Palmisano needs that proportion to keep growing--these ad hoc groups are key generators of new sources of income. Four years ago IBM had only a small presence in biotech. Even so, at the company's R&D labs some 150 scientists were focusing exclusively on computational biology. A researcher, Carol Kovac, got the job of creating a business unit targeting biotech and medical-research customers; within three years the new unit generated over $1 billion in annual sales and included employees from all across IBM. One hot new product: a $100,000-and-up piece of hardware for storing medical images. Last year Kovac was given responsibility for all of IBM's $4.8 billion in health-care-industry work. She has hired more than 1,200 health-care experts to help IBM come up with new ways to use IT in medicine.

The result is a furious push into services for the health-care industry. The Mayo Clinic, which for decades has maintained one of the world's largest and most complete sets of data about current and former patients, hired IBM to help assemble all that information into a single, easily searchable database. Kovac's group did that, and also came up with other ideas. IBM and the Mayo doctors are now taking the database a step further, dreaming up ways to cross-reference the information with genetic profiles of patients, using IBM's Blue Gene supercomputer to do the heavy lifting. The Mayo doctors see a chance to propel medicine from a world of hunches to one of data-driven decision-making, allowing for better and more individualized treatments. IBM sees that too--as well as a chance to resell this particular service to others in the health-care industry, worth $1.6 trillion a year in the U.S. alone.

All those services, of course, make it harder for IBM's customers, er, clients, to wander off and buy from competitors. Yet there is a downside to the service-heavy approach: It can be a drag on the bottom line. Armies of consultants inherently cost much more than, say, supplying hardware or lines of code. The services group's gross profit margin is 25%, vs. 29% for the hardware group and--hold on to your hat--86% for software. Now that services is his biggest and fastest-growing division, Palmisano must find ways to make it more efficient if he is to meet his goal of 10%-plus earnings-per-share growth. One way is for IBM to make an even stronger push at taking solutions it custom-designs for one customer and reselling versions of them to other customers in the same industry.

Making services pay off better was an impetus behind a major management shuffle this spring. Palmisano assigned chief financial officer John Joyce to take over that part of IBM's business, and charged him with coming up with new ways to boost margins. His predecessor, Doug Elix, a Palmisano protege and master salesman who increased the size of services by 33% since 1999, moved over to run all of IBM's sales. (Not that Palmisano is leaving all the income-boosting to the services group. Over the past few years he has shed a handful of big businesses, including IBM's disk-drive operation, that together produced $5 billion a year in revenues but were insufficiently profitable.)

In his quest to boost profitability in the services world, Palmisano is stepping further into the minefield of outsourcing--the trend that has everyone from unions to politicians riled up. Palmisano is talking about more than just taking over data centers, which IBM has been doing for more than a decade; for instance, IBM now operates most of Procter & Gamble's human resources department and also handles after-sales product support for everyone from Cisco to Philips Consumer Electronics. Such projects have been known inside IBM as "business transformation outsourcing," or BTO (the company is rebranding the effort, dropping the word "outsourcing" for obvious reasons); they involve hiring the customer's employees as IBMers and then working to improve their efficiency by bringing in new technology. Once a BTO unit is set up, Palmisano can go after other clients, letting the same IBM employees handle the work.

IBM is also eating its own cooking: It has shifted much of its own HR, finance, procurement, and customer-service operations to the same centers it markets to customers. The greater the volume in the centers, the more efficiently they operate, which both increases IBM's profit from customers and reduces its internal costs. IBM's $150 million purchase in April of Daksh, India's third-largest call-center operator, will also help cut costs in outsourcing deals. IBM contract signings for BTO deals exploded from $200 million in 2003's first quarter to nearly $3 billion in the fourth quarter.

Such outsourcing may offer better margins than other services, but its profitability is still nowhere near what Palmisano is hoping for. So he has redeployed IBM's software and research divisions to concentrate on programs and processes to make services less labor-intensive. Fully 20% of IBM's $5.1 billion R&D budget is now devoted to services-related research. New software automates the management of customer data centers. Voice-recognition software developed by IBM Research reduces the labor required in call centers and product support. A powerful search tool called Webfountain automates corporate research by scanning vast databases and the web to help customers answer critical business questions. More than a search engine, it can read and comprehend text on giant swaths of the web and recognize patterns of meaning. One huge bank uses it to create composite profiles of applicants for credit; an energy company recently used Webfountain to try to predict issues that might come up at a shareholders' meeting.

The way Palmisano looks at it, he's hatching his own growth businesses--he needs a lot of them to make that 5%-plus per year growth goal. Already IBM's annual Linux-related revenues have reached $2 billion. A variety of software businesses are growing rapidly, led by WebSphere--IBM's software suite that includes everything from tools for building business applications inside companies to the infrastructure for a secure e-commerce portal. The health sciences group continues to grow and now even handles jobs like running the supercomputers for Celera Genomics. Sales to small and medium-sized business, an area IBM has struggled for decades to figure out, last year grew 14%, to $19.8 billion. And the developing world is a huge growth opportunity for this most global of companies. Last year sales in China exceeded $2 billion, growing at a 15% rate.

One piece of Palmisano's growth machine still puzzles analysts: IBM's multibillion-dollar chipmaking arm. The unit is a spectacular innovator, yielding numerous industry-leading chip-technology breakthroughs. IBM is already building its third generation of microprocessors that incorporate two computing engines rather than one; Intel just announced it will launch its first such chips. And lately the unit has had dramatic wins in the marketplace. The big videogame companies--Microsoft, Nintendo, and Sony--all chose IBM's so-called Power chips for their next-generation machines, which will offer near-movie-quality realism. Still, IBM's giant chip fabrication facility in Fishkill, N.Y., has been a drain on capital and a big money loser in recent years. Merrill analyst Milunovich, among others, regularly argues that the company should exit the business. Palmisano counters that the unit will be profitable in 2004 and that for IBM's hardware division, chipmaking is a priceless and essential weapon. Executives argue that when all of IBM's units are involved in designing the chips, it can build even more powerful computers. Power chips now underpin a range of IBM machines all the way up to supercomputers.

For all the impressive changes palmisano is bringing, IBM faces fearsome competition. In the courts it must deal with a multibillion-dollar lawsuit from software firm SCO, which alleges that IBM stole from it to help build up Linux (for more, see "Gunning for Linux" on fortune.com). Out in the marketplace, things are even tougher, especially the competition against Bill Gates. Microsoft was only about a third IBM's size last year, but its $10 billion in profits exceeded IBM's by one-third. The fabulously wealthy software company is mentioned repeatedly by IBM executives as a rival that could potentially match IBM's capabilities. Microsoft promises customers that if they use its software platforms exclusively, they will gain efficiencies similar to what IBM promises, and more quickly. That pitch has helped Microsoft make headway selling software to small business.

What shakes IBM to its boots is imagining what might happen if Microsoft somehow seriously hooks up with Accenture, the other powerhouse of IT-strategy consulting--a combination that could have a reach similar to IBM's. Microsoft CEO Steve Ballmer sits on Accenture's board, and the two companies already have a small joint venture. (Both Microsoft and Accenture declined to comment.)

Then there's HP, which has dreams as grand as IBM's and which operates on almost as vast a scale. CEO Carly Fiorina says her company's advantage is in offering products that are as capable as IBM's, only cheaper. She calls HP the company of "high-tech, low-cost" computing. In an interview in January, she said that by contrast, "I would describe IBM as high-tech, high-cost." (She calls Dell "low-tech, low-cost.") HP failed in its effort to buy a major management consulting company during the boom; now chief technology officer Shane Robison claims that that is an advantage after all. HP can build IT systems just as well as IBM, he argues, yet at the same time, "there's no reason we can't partner with Accenture or Capgemini or Bearing Point or Deloitte." Palmisano's challenge is to prove that hardware people and software people and services people really can cross-pollinate more effectively within IBM than anywhere else in the world, and that the customer clearly benefits.

Finally, there's the biggest question: Is the world ready for yet another IT upheaval? The Internet revolution tripled the size of computerdom to roughly $1 trillion a year. If Sam is right about that $500 billion of untapped growth, IT will grow half again as large, and, as it mushrooms, will usher in another monumental shift in the ways companies operate and in their interactions with consumers.

Palmisano knows he has set high goals--higher than many think are attainable. His response is to keep pushing IBM's parts to work together ever more closely. He concedes that no more than 20% of IBM's big customers today are true "clients"--trusting IBM to help on make-or-break business issues like gaining more revenue per customer, say, as opposed to getting sold databases or Linux servers. He says IBM has these deep relationships with about 60 major customers now. He'd like the number to climb to at least 100 in two years. But he admits that people from various parts of IBM are not yet converging often enough on such client-transformation efforts--not creating enough solutions. "It's not systemic today," he says. "It happens today because it's a critical project, a very visible client, or a senior IBM person's engaged. When I talk about execution, that's what I'm talking about--to make it systemic."

The criteria for genuine greatness are daunting. "What is the definition of a great company--generically?" asks management writer Collins. "No. 1 is performance. Not relative to your peers, not just relative to your past, but such that an investment in your company is substantially superior to an investment in the general market. And no discounts for being in a tough industry. Second, you need a unique impact. Are you doing something of such excellence that if your company went away it would leave an unfillable hole?" Collins says that any company that can stage a rebound like IBM's in the past decade is surely capable of meeting his tests. "But will they?" he asks. "That's up to them."

Among IBM's top management there is extraordinary emotional resolve. Palmisano and most of his deputies joined the company more than 25 years ago, when IBM was the king of computing. They suffered through the dark years, a bitter experience that molded them. They bring up the subject of IBM's stumble without being asked. Says software boss Mills: "We missed market turns. IBM tried to protect its own spaces. It was misguided." Longtime services chief Elix says the early '90s were "humbling," and adds, "You know you can never feel safe." And here's senior vice president Nick Donofrio: "We couldn't get out of our own way, and it almost cost us our lives. In some sense it was our fault, and we are driven with a passion that says, 'Never again.'" As for the boss himself, he actually chokes up a little when he talks about it: "We remember when IBM was viewed as the greatest company in the world; then we went into the crapper. Today, before they retire, people want to remember the old feeling." Hearing conviction like that, you can't help believing these guys will do whatever it takes to get there.