AI models vary widely in energy consumption per query, study finds

Research examining the electricity demands of different artificial intelligence systems shows significant efficiency gaps that could shape data center power demand and fossil fuel reliance as AI workloads surge.

First covered Sep 29, 2026 · 6:00 AM PT · 1 article · no longer in the live window

AI models vary widely in energy consumption per query, study finds — Image via heatmap.news

Coverage

45d Heatmap News Which AI Models Use the Most Energy? — The rise of artificial intelligence is driving an historic surge in electricity demand that’s boosting fossil fuel use and threatening climate progress.

Primary sources

epa.gov Greenhouse Gas Equivalencies Calculator - Calculations and References | US EPA

More coverage from around the web

microsoft.com Scaling AI with 8 to 20x energy efficiency | The Microsoft Cloud Blog

arxiv.org Energy Use of AI Inference, Efficiency Pathways, and Test-Time Scaling

aimultiple.com AI Energy Consumption Statistics

digitalapplied.com AI Model Sustainability 2026: Energy Use Data - Digital Applied

sciencedirect.com Energy use of AI inference, efficiency pathways, and test-time scaling - ScienceDirect

Discussion

Bluesky: @kbrigham.bsky.social Katie Brigham · 44d ago · 6 ♥ great @emilypont.bsky.social story about a new initiative to quantify the energy intensity and emissions of proprietary AI models from Google, Anthropic, and OpenAI

Appeared in: 2026-09-29 (#3) · 2026-09-30 (#5)

← Back to the brief