Steve Ballmer's Big Moves

In the automotive industry, you've always got lots of brands, but the question is How many underlying platforms should there be? If you're Ford, should you have 20 platforms worldwide? Or should you have 10? Or one? Having more platforms gives you more flexibility, but it means that you don't have nearly the efficiencies of scale that you could enjoy. Having a smaller number of platforms always seems desirable, but getting there is never easy. How do you manage that stuff? These issues have a lot of the complexity that I'm thinking about.

It's nice to have simple businesses. But sometimes it's the complicated businesses that make a lot of money. IBM used to be more complicated, but it is getting to be more simple, because services is starting to be its real business. I don't want that to happen to us. I don't want to get simple in that regard.

You have a huge dot-net initiative under way, which is not going to make Microsoft any simpler. Explain why you're doing it and what you hope to accomplish.

We want to give users and developers the next-generation platform for taking advantage of Internet technologies. Right now, it's not at all easy to mix and match things from different Web sites. With our application-integration model, that should become much more practical. Also, the user interface should scale up and down with you, so that you can use the same software on a big screen or on a little screen. And finally, a new generation of software should take care of itself. You shouldn't have to install it and update it formally. The software should live in the Internet cloud as well as on your desktop or on a server -- so that it can be updated effortlessly.

Any time a big company does something this fundamental, the tricky part is to keep the train running while making this big bet for the future. In our case, Windows and our Office applications suite are two of the best businesses in the world. We could argue about whether the profit margin on any new business would ever be as good as the margins on Windows and Office. The answer would probably be no. But for us to get revenue and profit growth, our newer businesses -- servers, small business, and consumer applications -- will be very important. They'll help us grow faster than the basic rate of the PC business. We're certainly not underinvesting in these areas.

As our dot-net technologies get included in each of our businesses, they will make the non-dot-net versions of these products obsolete in four, five, or six years -- whatever the magic number is. You know, there are people in the Office group who would love to say, "Let's just do everything on the Internet right away. Let's abandon those current customers." But you can't do that.

I'm sure that any time people want to pitch you on a new initiative -- whether it's someone internal or external -- if they encounter resistance from you, they warn you that you are about to succumb to "the innovator's dilemma." You've read the book on that subject by Harvard Business School professor Clay Christensen (The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail, Harvard Business School Press, 1997). What's your thinking about whether Microsoft is vulnerable?

Let's remember precisely what the book says. Christensen contends that something crummy can get traction in the marketplace even when the established companies don't want to make it. That happens when there's something different enough and important enough in the way that the thing is made that, eventually, a better version of it will prevail. His book doesn't really focus on the question of whether big companies or small companies can innovate better. Frankly, big companies can do things that small companies can't even dream about doing. Being big or small isn't the crucial issue. If you don't move, you don't move.

Look at what's happening with the Linux operating system. By Clay Christensen's definition, it's crummy. It's not as easy to use as our Windows operating system. It's not as debugged as Windows, and it's not in as many applications. But in some markets, crummy has been okay. In multi-Web-site hosting, Linux has some traction. In that market, we recognize Linux quite clearly as an A - number one, first-class competitor. You can't just dismiss these things because they aren't as good.

Or take things like ThinkFree Office or HalfBrain. These are office applications and Internet hosts, and they aren't very good. But if they're sufficiently more convenient than what's been available before, could they make us obsolete? You have to make judgments. You can't just assume that those products will go away. There's always going to be someone out there asking, Is there something that will disrupt Office -- by providing half of the features but 20 times the simplicity -- something that could capture the world?