The Download: AI’s energy future
Summary
MIT Technology Review has released a short video to accompany a deep investigation into how much energy the AI industry consumes — down to estimates per query. The piece highlights that data-centre electricity use jumped about 80% between 2020 and 2025 and is set to rise further as billions of users adopt AI services. The article explores the tension between two outcomes: AI driving up electricity demand and local grid stress, versus AI tools helping to integrate clean power, optimise operations and prevent outages.
The investigation also documents a shift toward greater transparency: after months of requests, several big tech firms began publishing per-query energy figures. Despite that, key unknowns remain about model-level consumption, cumulative system impacts, and how concentrated demand will affect local grids and prices.
Key Points
- The video summarises MIT Technology Review’s investigation into AI’s energy use, including per-query estimates.
- Electricity use by data centres rose roughly 80% from 2020 to 2025; growth is expected to continue with AI adoption.
- Proponents argue AI can aid the grid (better integration of renewables, predictive maintenance, efficiency gains) while opponents warn of higher overall demand and local stress.
- Big tech initially withheld per-query energy numbers but have recently started publishing them, improving transparency.
- Three major unknowns persist: accurate per-response costs at scale, total-system impacts over time, and geographic concentration of demand.
- Rising electricity prices are already being felt in regions with high data-centre concentration.
Why should I read this?
Short version: if you care about power bills, the climate or whether AI will strain local grids, this is worth your few minutes. The video makes a knotty topic easy to grasp and the reporting pulls together the latest figures and arguments — saving you the legwork. It’s a crisp briefing on why AI’s growth matters beyond just new apps and chatbots.
Context and relevance
This story matters because AI’s rapid uptake intersects directly with energy policy, climate targets and infrastructure planning. As models scale and daily queries reach into the billions, the industry’s electricity appetite could reshape where and how grids are upgraded, influence energy prices, and alter the pace of renewable deployment. The piece is relevant to technologists, energy planners, policymakers and businesses that rely on cloud AI — all of whom need clearer metrics to weigh benefits against costs.
Source
Source: https://www.technologyreview.com/2025/09/10/1123489/the-download-ais-energy-future/