Microsoft After Gates. (And Bill After Microsoft.)

The question is whether Microsoft can compete as effectively without its founder. Ballmer, who took over as CEO from Gates in 2000 after 20 years as his key exec, is confident it can. That transition had been rough for both parties; according to Ballmer, it took a while to determine who was the junior partner and who was senior. But after the bumpy transition, Ballmer is comfortable flying solo. Anyway, it won't be like Gates is falling off the face of the earth; one day a week, he'll be there (though, in a symbolic act, Ballmer is moving into Gates's office, forcing Bill to establish his part-time digs down the hall). But if Gates got so totally involved in working out Third World microfinance deals that he stopped taking Ballmer's calls, "we'd be fine," says Ballmer. "If he really said, 'You can't have a minute of my time in the next year,' I'd feel bad. But the place would be fine."

While Ballmer runs the business, the burden of filling the Gates gap in geek expertise falls to a pair of technical leaders who will split duties. Craig Mundie, a 16-year Microsoft veteran, will make more public appearances, and grapple with and lead longer-term projects involving education, health care and software that exploits the superpowerful computers that will appear in the future. He'll also concentrate on more basic technologies like chips and infrastructure. Ray Ozzie—the software pioneer (he invented Lotus Notes) who arrived in 2005 when Microsoft bought his company, Groove—will be more of an internal presence, crafting the Web-based strategy that will roll out in the next couple of years.

As for Vista, Microsoft is well aware of its failings and Gates wants to make sure the next release, Windows 7, will not suffer a similar fate. Several months ago, he outlined his product vision to the team. "He brilliantly brought together the history of Windows, the mission that was ahead of us and the importance of Windows, but also challenged us on a personal level to think about what we could do better," says Steven Sinofsky, who heads the Windows effort. To Sinofsky, it was a bittersweet moment. "It was hard to escape the reality that his time allocation to Microsoft will be shifting while we're building this product," he says. Sinofsky's worries are mitigated by the fact that Windows 7 is one of the projects Gates will continue to follow closely. (Others include search, the next version of Office and the user interface.) In fact, the last of the notorious "Bill Reviews," a sort of geek kabuki where Gates unstintingly assesses a product, was devoted to the latest iteration of Windows 7.

Anyway, that one day a week on Microsoft business isn't a firm limit. "Measuring time is always tricky when you're someone who is on e-mail night and day," says Gates.

While Gates has a clear picture of his continuing role at Microsoft, the details of his new job at the foundation are somewhat fuzzier. "He's still trying to figure out what mix of things he'll be doing," says foundation CEO Patty Stonesifer. Some things, though, are determined. "He'll be out there working with people in Germany on foreign aid, and working with Bono on the advocacy end. He'll use the same approach here that he uses as chief software architect at Microsoft."

Gates does have some specific ideas, big and small. At the suggestion of Warren Buffett—who will donate billions from his fortune to the foundation over the next few years—Gates intends to work on an annual letter, in the same spirit as Buffett's yearly missive to Berkshire Hathaway shareholders. To learn more about areas the foundation is working on, he is doing intensive reading on education and science, and has monitored online college courses in geology, history and microparticle physics. He is fully engaged on several problems already. "People know I have a particular fascination with AIDS and malaria," he says. One obsession is an AIDS vaccine, and Gates was disappointed when a trial indicated that a promising candidate for a solution, made by Merck, was not effective. Discussing the vaccine, an intense Gates cites research that implies that a variation might be more effective. Clearly, he's viewing the process the same way he views software development—maybe version 3.0 will do the trick.