(FORTUNE Magazine) – For the past eight months, ever since Bill Gates named him Microsoft's first president in seven years, Steve Ballmer has been the software giant's invisible man. He hasn't been part of the antitrust trial in Washington, D.C. Nor has he spent much time making public appearances. Instead Ballmer has been on a quest. He's been walking the halls at Microsoft's headquarters in Redmond, Wash., a sprawling complex that now comprises two campuses on 295 acres, with 41 buildings housing the offices, cubicles, and labs of some 15,000 employees. A brilliant, wildly energetic, table-pounding kind of boss who for the previous six years had run Microsoft's sales and marketing, Ballmer has spent his time as president listening. He held one-on-one meetings with more than 100 software engineers, product managers, vice presidents, and other employees, most for an hour or more. His goal: to assess what the $14.5-billion-a-year giant is doing right and, more important, what it is doing wrong.

The patterns that emerged alarmed him and his boss. While Microsoft is growing like a strapping adolescent (1998 sales rose $3 billion, nearly 30%), it is showing the high-tech equivalent of premature gray: slow response to changes in the market, stubborn resistance to new ways of thinking, deafness to customer needs. For Ballmer and Gates, that added up to an imperative for change. Ballmer says, "It became clear to both Bill and me that if we're going to have our company succeed and do the right thing for customers, we're going to need to, in some senses, reinvent ourselves."

So two months ago he and Gates sat down to hammer out a reorganization. Announced in March, it aims to rejuvenate Microsoft by changing its focus from its products to the customers who use them. Microsoft will split into five quasi-autonomous groups, for corporate systems customers; programmers; knowledge workers; ordinary Windows customers; and consumers interested in digitized content, entertainment, and shopping. At fast-growing companies, reorganization is often a fact of life, but Microsoft watchers say the Ballmer plan is anything but routine. Says Rob Enderle of the Giga Information Group in Norwell, Mass.: "It's a wake-up call from their new president that it's not business as usual."

Why, you might wonder, should Microsoft need a wake-up call? Isn't that like bringing an alarm clock to Starbucks? There's nothing sleepy about the stock, up 111% in the past year to a recent $94-a-share high. In fact, Microsoft has emerged as the world's most valuable company; to judge by its $466 billion market cap, investors treasure it more than they do, say, Wal-Mart or GE (see table). The company's dominance in PC software is overwhelming; this year and next, Microsoft stands to reap millennium-worthy profits selling new versions of its Windows operating system and its Office desktop-applications software. Those products provide nearly three-quarters of Microsoft's revenues and even more of its profits.

But Ballmer is focusing on cracks in the foundation--some obvious, some subtle. On the obvious side is the threat of punishment or dismemberment at the hands of Uncle Sam (for more on the antitrust suit, see Techno File). On the subtle side are difficulties Microsoft has had trying to reach beyond the desktop and keep pace with the Internet age. Some digerati and journalists see such problems as portents of doom. Among the recent headlines: BILL GATES' ONLY OPTIONS: SPLIT OR QUIT; MICROSOFT IN PERILOUS E-COMMERCE WATERS; and 83 REASONS WHY BILL GATES' REIGN IS OVER.

To paraphrase Mark Twain, the reports of Microsoft's death are wa-a-ay overstated. Microsoft is plenty healthy; Ballmer's wake-up call is plain corporate common sense. He is preparing the company for the lessening of its monopoly power and pushing it to get ready for increased competition in everything from hand-held organizers and smart appliances to online services to software that lives on network servers and doesn't really need PCs. Meeting such challenges involves changing ways of doing business that have long fueled Microsoft's success:

Worlds beyond windows. Microsoft's strategy in recent years has been to lay claim to new markets, from palm-sized devices to giant enterprise networks, with new versions of its Windows operating system. Its biggest bet is Windows NT, soon to be renamed Windows 2000, which Microsoft wants to make the standard for nearly every size computer from desktops on up.

The quest to build software that will run such a broad range of machines is the largest project Microsoft has ever undertaken. The NT project is so big, in fact, that it is virtually governmental in scope. The product comprises, by some estimates, 75 million lines of code (Microsoft does not disclose the figure). And like many government projects, it is late and over budget and lacks features some customers want.

That's not to say Windows NT is in danger of meltdown. Early versions of the program have already begun to sell strongly in the so-called server market, for computers that power intranets and handle other business tasks, and in the corporate desktop market. Those markets alone will be worth billions in sales and profits. Where NT is falling short of Microsoft's ambitions is in so-called enterprise computing: the high-end systems that corporations depend on to manage vast databases and handle vital tasks. Microsoft has made few inroads against entrenched players like IBM, Sun Microsystems, and Oracle, which are improving their products fast. Microsoft also has headaches in the arcane but influential world of universities and other organizations in which programmers seek high-power computing at bargain-basement rates. There, it is finding unexpected competition from the Linux operating system, which is available on the Internet for free.

Meanwhile, in the burgeoning markets for hand-held organizers, cellular phones, cable set-top boxes, and other consumer-oriented gear, Microsoft has elbowed its way in with Windows CE, a stripped-down version of its PC operating system. But CE has not gained share as fast as Microsoft would like. The problem: The consumer devices for which CE is designed bear little resemblance to multipurpose PCs. Competing products, like 3Com's Palm computer, use software expressly designed for specific tasks, and so have an edge. "No question they have the lead," admits Bob Muglia, senior vice president of Microsoft's business-productivity group. "Palm has done a great job."

Isolation from customers. In the desktop market, Microsoft has long been able to boost sales, hold down prices, and reap profits by delivering software through computer makers like Compaq and Dell and retailers like CompUSA rather than by selling to customers directly. Since PC software is inexpensive and increasingly easy to use, customers have been satisfied.

The middleman approach doesn't work as well in selling costly, complex software to corporations. Such software typically requires customizing and monitoring by skilled technicians: IT managers who buy it for running accounting, inventory, and other core operations want handholding. But whereas IBM employs tens of thousands of personnel to help customers with such tasks (see "From Big Blue Dinosaur to E-Business Animal"), Microsoft relies almost entirely on independent, mostly small service businesses that it anoints as "partners." Says Howard Anderson, head of the Yankee Group research firm in Boston: "Microsoft doesn't want to be a systems integrator. It does not like businesses that are labor-intensive." Only in special circumstances will Microsoft dispatch an employee to assist full-time at a customer site, and for that the customer typically must pay extra--all of which can hurt customer satisfaction.

In addition Microsoft has little control over the quality of service its partners provide, and its isolation makes it somewhat customer-deaf. Up to now, analyst Enderle says, "Microsoft people were talking to customers, but the messages coming in from customers were being ignored." The reorganization will likely increase the amount of handholding. It will also, Microsoft hopes, start a virtuous cycle in which listening more closely to customers will help simplify its products, ultimately making handholding less necessary.

The winner-take-all pitfall. On the Internet, Microsoft has made nice gains. With more than 20 million visitors last month, according to Media Metrix in New York, Microsoft Network is the Web's third-most-popular portal, trailing only Yahoo and AOL. The company's travel Website, Expedia, and its auto-sales site, CarPoint, are highly ranked in their niches; Hotmail, which Microsoft acquired last year, is No. 1 in free e-mail. Microsoft triumphed in the browser wars as well, clawing its way to 44% market share with Internet Explorer and driving archrival Netscape into the arms of AOL.

While the Internet so far accounts for only a small portion of Microsoft's sales--$548 million in 1998, estimates Goldman Sachs--Microsoft aims to grow rapidly in e-commerce by positioning itself as a kind of half-Yahoo, half-IBM. It wants not only to be a portal for consumers but also to provide software to companies that want to do business on the Web.

Yet Microsoft's business style may lessen its chances of becoming a major force on the Internet, a realm of complex, shifting alliances and mutual dependency. Witness Amazon.com, which helped gain dominance and bolster its brand name by signing up 200,000 "associate" Websites that refer customers to Amazon in return for a share of the revenues. Microsoft, by contrast, has succeeded thus far by building (Expedia), buying (Hotmail), or crushing (Netscape)--the moves of a PC-industry survivor that prefers to go it alone.

Consider its recently launched MoneyCentral site, which aims in part to help consumers and financial services companies conduct transactions. To succeed, Microsoft must convince the companies that it will share customers, not swipe them. Most still remember the flap a few years ago when Gates labeled banks dinosaurs and implied that Microsoft would use the Internet to go after their business. Gates and Ballmer make no bones about the need for Microsoft to grow up and temper its PC zealotry. They've even changed its official mission. For as long as anyone at Microsoft can remember, the company's grail has been: "A computer on every desk and in every home." Now the objective is much broader. In a statement accompanying the reorganization announcement, Gates declared: "We see a world where people can use any computing device to do whatever they want to do anytime, anywhere. The PC will continue to have a central role...but it will be joined by an incredibly rich variety of digital devices accessing the power of the Internet."

The changes afoot in one of its five new units, the business and enterprise division, suggest how the process could unfold. The group is responsible for most of Microsoft's Windows systems, including 2000. Its boss, senior vice president James Allchin, is breaking it into subgroups, each focused on one type of customer. One subgroup will target mobile knowledge workers who use laptops. Another will target IT professionals running corporate servers. Each subgroup must adhere to a few ground rules. (Example: All versions of Windows must be able to run basic applications like Excel, Microsoft's spreadsheet.) Beyond that, the subgroups are free to develop and market versions of Windows that differ in functionality, look, and feel. Allchin calls his approach "the United States of Windows." He says, "We'll do a better job if we talk to customers more."

Making such changes throughout the company will be tough. Even if Ballmer succeeds, Microsoft is unlikely ever to replicate the lock it has on the desktop. No company is likely to. In the Internet world, the rules of the game have changed to favor a diversity of products, services, and even standards. Thus 3Com, with its Palm machines, may dominate the hand-held organizer niche without ever competing in, say, digital music players or TVs. Sony's PlayStation can be a top game machine without Sony's being competitive in PCs. Even the enterprise software market will become more diverse, with openings for many systems from many companies running many applications. In fact the primacy of the general-purpose computer itself is waning: The world is beginning to see specialized computers for specific tasks on the network, like database machines recently announced by Oracle.

You could say that, anticipating all this, Ballmer and Gates are moving fast to get Microsoft fixed before Wall Street notices that its monopoly is ending. But maybe that's underestimating the stock market as an information machine. Maybe in driving the stock to an all-time high, the market is betting on the ability of Microsoft to adapt.