Over the past year Greenpixie has been working with customers to quantify AI emissions using the best-available approaches in market. Now we’re taking the next step with our Q1 Production Release.
Our 2026 Sustainable AI product is designed to move beyond “directional estimates” toward real operational accuracy. Notably our methodology:
· Pioneers an energy consumption estimate per input token (the biggest impact on cost we are seeing in enterprise)
· Uses an accurate throughput model that accounts for batching, prompt/response length, and model architecture
· Is trained on Greenpixie benchmark data,
· Is applied at the cloud SKU level – so AI emissions can be understood and managed like any other cloud cost or sustainability metric
In this session, Kevin, James, and Joshua will share how customers have used our early approach in 2025, what we learned, and how the Production Release will unlock more trustworthy AI measurement in 2026 and better decision-making around deployment.
Who is Greenpixie?
Greenpixie helps cloud teams bring ISO-verified sustainability metrics (carbon, water and energy) into everyday cloud decision-making by applying them directly to cloud billing at the SKU / line-item level. That level of granularity gives customers maximum flexibility with their data, meaning sustainability data can be embedded into FinOps, platform governance, and architecture & design choices, where complex decisions are easiest (and cheapest) to change.
It’s free to join. Email lpratico@iasaoffice.org for the meeting link.