> How many startups comes up with new medicines, for example?

Lots, and lots try and fail (the ones that succeed, or even make it past the first few big hurdles, often get acquired by the big players, because buying work that's done is less risky than early research...)