Carbon-Aware Computing: Reduce Emissions Without Rebuilding Everything
Actualizado: 2026-05-03
Carbon-aware computing is the idea that flexible workloads can run when and where energy is cleaner. Grid carbon intensity varies by hour and region: running a batch job at night with high wind emits 3-5x less CO₂ than midday on natural gas. No hardware change, no app rewrite, just smart scheduling.
Key takeaways
- Grid carbon intensity varies up to 16x between the optimal and worst scenario depending on zone and hour.
- 20-40% of an organisation’s total compute is deferrable without user-facing impact.
- Typical reduction is 10-30% CO₂ on flexible workloads with minimal effort.
- The main tools are Electricity Maps API, WattTime, and the Carbon Aware SDK from the Green Software Foundation.
- Carbon-aware computing adds no cost in most implementations: the scheduling logic is trivial.
The base concept
Grid carbon intensity isn’t constant. A workload consuming 1 MWh: in a clean zone at optimal time ~50 kg CO₂; in a dirty zone at bad time ~800 kg. 16x difference by changing nothing but timing and location.
Flexibility types
Time-flexible (deferrable without user impact): batch processing, ML training, data warehouse refresh, backups, indexing.
Location-flexible: stateless multi-region workloads, compute-only tasks.
Not flexible: real-time user-facing requests, OLTP transactions, control loops.
Data sources
- Electricity Maps[1]: real-time and forecast data per region for 50+ countries.
- WattTime[2]: US focus with granular historical data.
- Carbon Aware SDK[3]: open-source multi-language wrapper.
Implementation patterns
Simple job scheduling: check the cleanest hour in a 24-hour window and schedule accordingly. Multi-region routing: deploy to the region with the lowest carbon intensity. Carbon-aware KEDA: scale down replicas when carbon intensity exceeds a threshold, defer work until intensity drops.
Real ROI cases
Google reports millions of kg CO₂ avoided annually through scheduling. Microsoft prioritises low-carbon hours for Xbox updates. For a mid-size company: 10-30% CO₂ reduction on flexible workloads with one implementation sprint.
Honest limitations
Forecasts have ±20% error. Regional coverage isn’t universal. Savings are 10-30% of workloads, not 100% of compute. Greenwashing risk if reported without significant real reduction.
Conclusion
Carbon-aware computing is low-hanging fruit for IT sustainability. Without hardware changes or app rewrites, you can reduce flexible-workload emissions 10-30% with intelligent scheduling. In the context of CSRD and growing European regulation, carbon-aware computing will move from option to expectation in the coming years. Better to adopt before being forced to.
