Every search query burns about 0.3 watt-hours of electricity. So does every ChatGPT query. Multiply that by billions of queries a day, add the servers, add the cooling systems keeping those servers alive — global data centres consumed an estimated 460 TWh in 2025, roughly 1.8% of world electricity demand. The IEA projects that crossing 800 TWh by 2028.
What It Actually Covers
Green computing (green IT) means designing, using, and disposing of computing resources to minimise environmental impact — chip fabrication, data centre cooling, software efficiency, e-waste, renewable sourcing. The term got formal recognition in 1992 when the EPA launched Energy Star, originally targeting monitors and CPUs that ran at full power even when idle.
Where the Energy Actually Goes
Cooling eats 30-40% of total data centre power — the second-biggest draw after computing itself. The efficiency metric here is PUE (Power Usage Effectiveness): industry average improved from 2.5 in 2007 to 1.55 in 2022. Amazon's data centres hit 1.14 in 2025, beating the cloud industry average of 1.25.
The Real Levers
- Renewables — tech companies signed ~40% of all corporate renewable power agreements in 2025
- Virtualisation — traditional servers ran at just 6% utilisation; running multiple workloads per machine fixed that
- Hardware efficiency — modern AI chips use nearly 99% less power per computation than 2008-era hardware
- E-waste management — one of the fastest-growing waste streams globally ## Why AI Complicates Everything AI-focused data centre power use is set to triple by 2030. Google's data centres alone consumed 6.1 billion gallons of water in 2024 — up 20% year-over-year. Water, not just electricity, is becoming the real constraint.
Read full article here: https://www.weejix.com/technology-articles/green-computing
Top comments (0)