Measuring the environmental impact of delivering AI at Google Scale
Summary
A study measuring energy, carbon, and water consumption for AI inference at Google scale, specifically analyzing Gemini Apps text prompts in a production environment.
Key quotes
the median Gemini Apps text prompt consumes 0.24 Wh of energy
Google's software efficiency efforts and clean energy procurement have driven a 33x reduction in energy consumption and a 44x reduction in carbon footprint for the median Gemini Apps text prompt over one year.
The research proposes a comprehensive methodology for measuring environmental metrics across the full stack of AI serving infrastructure, including accelerators, host systems, and data center overhead. It provides empirical data on the resource consumption of the Gemini AI assistant.