I recently found a bug on the Yahoo website (a graph was being drawn incorrectly) which could be fixed in a startup in about 15 minutes; but because Yahoo is such a large company and has so many users, individual developers can't go in and make even a trivial, obviously-correct fix to the live website -- instead, it takes three weeks for the fix to go through QA (along with all the other changes which have been prepared since the last QA cycle).
A founder who has spent three years building a startup -- and being able to fix bugs and see the results 30 seconds later -- isn't likely to be happy working with a 3 week QA cycle. There's nothing wrong with either approach; they're just suitable for different people.
Do what you're good at doing, and let someone else do what you're not good at doing. If you're a startup hacker, then go do a startup; but once your company is successful and not a startup any more, bow out and let someone else take over. Staying around to do a job which you're neither interested in or good at doesn't do any good for either you or your company.