Finding the Lowest-CO₂ Regions in Real Time

CarbonRunner intelligently routes your jobs to the lowest-CO₂ regions across AWS, Azure, and GCP — in real time. By keeping every workload under 100g CO₂e/kWh, we cut emissions by up to 90% compared to running in default regions.

The result: your GitHub Actions and AI training run faster, cheaper, and dramatically greener, with just a single line of code change.

By partnering with the best in live grid energy data — Electricity Maps — CarbonRunner can shift workloads to regions with the cleanest energy in real time.

And where more granular data is available, we take that in too. All of it feeds into our dynamic weighting system, which continuously balances carbon intensity, cost, and performance to place your jobs in the lowest-CO₂ regions possible.

How Do You Find Data Centers in Water-Stressed Regions?

We overlay global data center maps with water scarcity datasets — including the World Water Map and investigative reporting such asThe Guardian’s coverage of drought-prone areas. Where available, we also incorporate operator-reported Water Usage Effectiveness (WUE) to refine the picture.

How do you know what GitHub's emissions are?

CarbonRunner estimates GitHub’s emissions by running small, lightweight jobs directly on GitHub-hosted runners. From inside the job, we query the instance metadata service, which reveals the cloud provider region the job is running in.

Once we know the region, we cross-reference it with live electricity grid data (via Electricity Maps and other sources) to calculate the carbon intensity at the exact time the job ran. By combining runtime location with real-time grid data, we can build a transparent picture of the emissions from GitHub’s default runners — and show how CarbonRunner improves on that baseline.