That's a good point, I was implicitly assuming that the lost power due to efficiency losses would be greater than the gained power due to cheaper costs.