AI Sustainability — Shubham's Tracker
Environmental impact of AI · 2025
Live
Sustainable AI Initiative

The environmental cost
of artificial intelligence

As AI scales globally, so does the infrastructure behind it. This tracker visualises the energy consumed, carbon emitted, and water used by AI data centers — and why measuring these matters for responsible innovation.

Period2019 – 2030 (projected)
SourcesIEA · MIT · Nature Sustainability
Updated
Energy
415
TWh / year
Global data center electricity consumption
↑ 40% since 2022
Carbon
40
Mt CO₂ / year
AI-related carbon emissions (2025 est.)
↑ 35% projected by 2030
Water
620
bn litres / year
Data center water usage for cooling
↑ 22% vs pre-generative AI era
Global data center electricity demand (TWh) — 2019 to 2030
AI share growing significantly · values from 2026 onward are projected
Non-AI workloadsAI workloads
Energy per AI action type
Estimated Wh per task
Estimates based on published model efficiency benchmarks
Energy cost by model size
Wh per 1,000 queries
Smaller optimised models can reduce energy use by 10–50×