It's the risk that kills it. If your compound fails at late Stage 2 or early Stage 3 you just did 7 years for nothing. Also: the penguin effect. A first-in-class compound requires target validation, which adds another few years plus you can't patent that effort, and then the stuff arrives at market and physicians don't know why they should prescribe it. Just compare profit from the first statin to current 5th-generation profits from that same class. Silicon Valley is so much more predictable and offers returns in a shorter timeframe.