Most large AI teams that we talked to are well on the way to optimizing their consumption of compute resources for ML. For smaller organizations...