Methodology
How Greenpixie calculates the environmental impact of cloud and AI usage.
Our approach: Maximum coverage - uncompromising rigour
Greenpixie's methodologies are designed to deliver consistent, evidence-based estimates across an entire cloud and AI estate, not just a small subset of core services or frontier models. In production, we typically map over 99% of customer cloud usage while applying the same documented methodology throughout. We do not use any spend based proxies.
At-a-glance
Supports AWS, Azure and Google Cloud.
Measures cloud usage at the individual SKU / rate code level.
Covers 99%+ of typical customer cloud usage.
Covers energy, carbon and water
Uses one consistent methodology across supported services.
Avoids spend-based carbon proxies.
Covers over 500 AI models, with more added continuously.
Built on real hardware benchmarking and predictive modelling.
Maps cloud services to underlying infrastructure.
Verified against ISO 14064-3 through independent limited assurance.
Continuously updated as cloud platforms evolve.
Cloud Methodology
Read →
AI Methodology
Read →
Compute Methodology
Read →
Embodied Emissions Methodology
Read →
Grid and Regions Methodology
Read →
Region Scorecard
Read →
Try the data yourself
To allow everyone to get started with cloud and AI sustainability, we have made our compute data available for free to all. Learn more here.