My two previous employers were both started in 2001, in terms of the idea and not the corporate entity. Last I heard, they are both still around, though in one case a founder let the company fold and then ran off to China with the IP to start again with cheap labor. But the ideas are increasingly irrelevant now.
In one case, the product was software to tunnel through firewalls. Most recent products just build that into the product itself now (witness: FogCreek CoPilot, AIM, anonymous proxies), using essentially the same technology we were working on 8-9 years ago (FogCreek calls it the Reflector) so there is less need for a general solution.
In the other case, the product started out as a desktop quant analytics interface. But then around 2002-2004, everything switched to the web. They did too, but they fell off the leading edge of technology: other firms now have much more web expertise than they do. And now with the financial crisis, I'd guess that people are increasingly suspicious of quant analytics.
I keep thinking about whether it's better to go for breadth or depth (in anything, not just startups). The answer always seems "Go for depth, as long as you're going deep into the right thing." But it's often not possible to tell what the right thing is until you've gone deep into it. By then, of course, you've already sunk a lot of effort into exploring the territory. Maybe that's why people ascribe so much of success to pure luck.