I wonder if:

The energy crisis will put pressure on scientists to optimize code? I've come across hideously unoptimized code in my career; where merely understanding general basics about the compiler / OS / database result in significant improvements. (And thus less energy consumption.)

There is a market for moving supercomputing to places with lower energy costs?