Yes, you're right and good point. Many grants require a power analysis now, which helps you decide on the sample size you need if you have a hypothesis about effect size. Many studies are still underpowered, but I think that there is a growing understanding of the problem.