The Download: AI’s energy future

The Download: AI’s energy future

Summary

MIT Technology Review highlights the growing electricity demands of AI and the potential impact on power grids. The newsletter points to a major investigation (and a short video) that traced AI’s carbon footprint down to single queries, showing data-centre energy use has risen sharply from 2020–2025. The piece balances the risks—rising electricity prices and strain on grids—with industry claims that AI could make grids cleaner and more resilient.

Key Points

1. Data-centre electricity consumption rose about 80% between 2020 and 2025, driven largely by AI workloads.
2. Researchers sought per-query energy figures for leading models; firms initially withheld numbers but have begun to release some data.
3. Big Tech argues AI can improve grid operations—integrating renewables, forecasting demand, and preventing outages—but the benefits are contested.
4. Transparency gaps remain: knowing a model’s per-response energy cost is still incomplete despite recent disclosures.
5. The newsletter bundles other tech developments—DeepMind interpretability work, Meta VR controversy, and several must-read stories—giving broader context to AI’s societal impacts.

Content summary

The story accompanies a short video that distils MIT Technology Review’s in-depth investigation into AI’s energy use. Reporters traced the industry’s carbon footprint and tried to pin down how much electricity models like ChatGPT and Gemini consume per response. After persistent reporting, companies started revealing some numbers, but key uncertainties remain.

AI’s rapid adoption is reshaping electricity demand: concentrated data-centre growth is driving local price rises and could force changes in grid planning. Proponents in industry claim AI will ultimately help the grid by speeding up renewable integration, improving efficiency and predicting failures. Skeptics warn that rising loads and opaque metrics make the net effect uncertain.

Context and relevance

This piece matters if you’re tracking climate, infrastructure or the economics of AI. Energy use is a practical constraint on how and where large models can scale. The investigation’s findings feed into regulatory debates, corporate sustainability claims, and planning for grid capacity in regions with heavy data-centre concentration.

Why should I read this?

Because it cuts through the noise. If you care about whether AI will wreck your electricity bill or actually help get more clean energy online, this gives you the facts and the drama — plus a neat video. We’ve done the digging so you don’t have to.

Source

Source: https://www.technologyreview.com/2025/09/10/1123489/the-download-ais-energy-future/