Approach that schedules computing workloads based on when and where clean energy is available to minimize carbon emissions. Like running heavy tasks when solar or wind power is abundant.
Machine learning training jobs shift to regions and times when renewable energy is plentiful, reducing carbon footprint by 30-50%.
No major cloud offers a single native 'carbon-aware scheduler' that automatically shifts arbitrary workloads across regions and time. Azure and Google Cloud provide carbon/emissions measurement tools (and guidance) you can use to build carbon-aware scheduling. On AWS and OCI, you typically rely on carbon reporting tools and custom orchestration (e.g., batch schedulers, Kubernetes, workflow engines) to shift flexible workloads based on grid carbon intensity and capacity.