It's not feasible because the fundamental cost of compute time remains constant. Running slower machines for extended periods equates to running faster machines for shorter durations. Our goal is to complete every task as quickly as possible, and we are continuously enhancing our performance in this regard.