I have a very different reason. In most large companies, you pay some sort of productivity overhead - for X units of work, you have to pay say 30%X for internal marketing, coordination(meetings) and other random activities. When that ratio crosses a tipping point, and that may vary from person to person, a startup becomes way more viable than a large corporation.

The other more important reason is that startups (at least the ones I've built) are far more tolerant of "failures/mistakes" than large companies. And thats pretty much the only way to learn to build a successful enterprise. Experiment without fear of failure, and then adapt.