This is not possible because the underlying cost of compute time is the same. Operating slower machines for longer periods of time is the same as operating faster machines for shorter periods of time. We aim to get every job done as fast as possible and we are constantly improving on that dimension.