(FORTUNE Magazine) – I know something about Microsoft no one else does: It will fail, in 2020 or 2021, although that might be a little early.

You may find this a tad presumptuous, given that other pundits are busy trying to predict the autumn ship dates for the Xbox game machine or the Windows XP operating system. But they don't understand the essence of a grand pattern that I'll call The Truly Great Company Failure Theory.

My theory is that great companies fail around the time they bring in their third major CEO. Put simply, the common pattern: visionary leader; successor; failure. People who create truly great companies tend to be people whose force of personality and vision last a long time, but not forever. Their impact carries through the reign of the hand-picked successor CEO, who runs the company with an understanding of the nuance and intent of the original vision.

But once the post-successor CEO takes over, all bets are off. The third CEO generally feels that he must take things into his own hands. He's under a lot of pressure, after all; still in the shadow of a great visionary, he feels he must try to do better than the two guys who have turned in outstanding results over a 20- to 40-year period. Not surprisingly, most often he folds under the pressure, corrupting the original vision without replacing it with an equally compelling vision of his own. This kind of pressure is why I would never try to run a company, even if I thought I could--it's a lot easier to carp.

I developed this theory by watching IBM back in the late '80s. I remember, in particular, that CEO John Opel decided to predict that IBM would be a $100-billion-a-year corporation by 1990. IBM still hasn't made it to $90 billion a year. Opel's remark seemed outrageous, since at that time no company had generated $100 billion in annual sales. But the prediction also seemed possible, for one reason: IBM then had such power that if the CEO said IBM would bring in $100 billion, it had to be so.

Of course, Opel was wrong. So why had he felt compelled to make such an outlandish forecast, setting himself up for failure? My hypothesis: He was trying to make himself stand out from the shadow of the great man at IBM, Thomas J. Watson Jr.

I'm going to count Watson Jr. and his father, Thomas Sr., as the single creative entity that invented the modern IBM. While Watson was succeeded briefly by Vin Learson, his key successor was Frank Cary, who ran IBM for nine years before Opel. Cary followed the model that Watson and his father had invented and implemented. Under Cary, the company preserved its veneer of invulnerability and omnipotence. Opel, though, needed to establish his credibility, which to me is the human trait that leads to The Truly Great Company Failure Theory.

IBM suffered after Opel's prediction, until Lou Gerstner came along in the 1990s. That leads to the corollary of my theory, which is this: Not-so-great great companies don't ever recover from the failure of their post-successor CEOs; truly great ones do. In other words, truly great companies don't actually fail, at least not terminally. Instead they go through some kind of wrenching change and reemerge in a different but equally powerful form. IBM is a truly great company.

So is Microsoft. It may be a little early in the annals of business history to make that claim, but I'm going to do it anyway. Steve Ballmer is doing a fine job as the new CEO and demonstrating plenty of independence. But Bill is still there, not far at all in the background. Indeed, it made perfect sense to me that FORTUNE put his smiling face--and not Ballmer's--on a recent cover, with the headline the BEAST IS BACK.

So as we did with the Watsons, let's count Gates and Ballmer as one founding entity, and surmise about the company's future. Gates and Ballmer are both in their late 40s, and so can last about the same amount of time running the company. The true successor CEO, then, is likely to be another individual from Microsoft's current ranks, appointed around 2012, when Gates and Ballmer finally do tire of the fray. So I figure that it will take until 2020 at least for Gates' force and vision to give way to something new. By then the company will be in the hands of the post-successor CEO, who will believe it is important to establish a new kind of Microsoft, responsive to the issues of its day.

Then--and only then, my friends--will Microsoft truly be vulnerable. Its competitors will pounce, and the company will suffer. But in the end the company will bounce back and rule anew. And if you and I are still alive, I hope you'll remember that you read it here first. If you're not, I'll find a way to remind you.

STEWART ALSOP is a partner with New Enterprise Associates, a venture capital firm. Except as noted, neither he nor his partnership has a financial interest in the companies mentioned. He can be reached at alsop_infotech@fortunemail.com. His column can be bookmarked online at www.fortune.com/technology/alsop.