And that's without even thinking about vertical markets. For example, I think if you sat down in a room with an industry expert in, say, the oil & gas industry and a few smart hackers, you could come up with a way to apply computer science to improve oil and gas prospecting. There are dozens of similar vertical markets, where some smart hackers with a modicum of domain-specific knowledge could really make a pretty significant improvement on the state-of-the-art, IMHO.
The group-think among certain communities on the Internet toward consumer-oriented, Web 2.0, advertising-dependent web startups is really pretty amazing -- I just get terribly bored reading about all the different people building Web 2.0 task management web apps and similar junk. There are lots of other kinds of startups out there -- it's just that YC doesn't focus on funding them, and TechCrunch doesn't cover them, so this particular community of developers never learns about them.
I do have at least one idea which is not consumer-oriented... My problem is that since these domains are not in-your-face advertised, or talked about here, it's harder to evaluate the market and the competition.
Plus, these apps usually need promotion of some kind, word-of-mouth doesn't really spread otherwise.