Why Data Centers Matter for Climate
The global information and communications technology sector accounts for roughly 2 to 4 percent of global greenhouse gas emissions, comparable to the aviation industry. Data centers specifically consume enormous amounts of electricity, and the absolute demand for computing continues to grow faster than efficiency gains. The carbon intensity of electricity varies enormously by location and time. In Norway the grid runs on almost entirely hydroelectric power. In regions of the United States with heavy coal infrastructure that same kilowatt-hour might produce ten times as much CO2.
Spatial Shifting: Running Code Where the Grid is Greenest
Spatial shifting means choosing which cloud region to run a workload in based partly on the carbon intensity of that region’s electricity grid. The electricityMap API and WattTime API provide real-time and forecast carbon intensity data for electricity grids around the world. By integrating these APIs into your infrastructure automation you can make intelligent decisions about where to route workloads. For a batch ML training job that does not need to run in a specific region, a carbon-aware scheduler might compare carbon intensity across available regions and route the job to the greenest one available.
Temporal Shifting: Running Jobs When the Grid is Cleanest
Within a single region, the carbon intensity of the electricity grid can vary by a factor of two or three throughout the day. In California solar energy is abundant during midday hours which dramatically lowers grid carbon intensity. Batch jobs like nightly database backups, model training runs, data pipeline processing, and video transcoding are perfect candidates for temporal shifting. None of these need to happen at a specific time. A carbon-aware scheduler can look ahead at forecast carbon intensity, find the lowest-carbon window in the next 8 hours, and schedule the job accordingly. The Carbon Aware SDK from the Green Software Foundation is an open-source toolkit specifically designed for building carbon-aware applications.
What Cloud Providers Are Doing
Google Cloud’s Carbon-Intelligent Computing platform automatically shifts flexible batch workloads to times and locations with lower carbon footprints. AWS has committed to powering operations with 100 percent renewable energy by 2025 and net-zero carbon by 2040. Microsoft Azure has introduced features in Azure DevOps and AKS that surface carbon metrics alongside cost metrics. The trajectory is clear: carbon impact will become a first-class metric in cloud infrastructure alongside cost and performance.
The Business Case
Beyond the environmental impact, carbon-aware computing has a strong business case. The EU Corporate Sustainability Reporting Directive now requires large companies to report Scope 3 emissions which include cloud computing. Organizations that can demonstrate verifiable emissions reductions have a real competitive and regulatory advantage. Cost and carbon often correlate since running workloads at off-peak times frequently aligns with lower-cost spot pricing windows.
