AI Data Centres Create Heat Islands Warming Surrounding Areas by Up to 9.1°C, Study Finds
Key Takeaways
- ▸AI data centres raise land surface temperatures by an average of 2°C, with extremes reaching 9.1°C in surrounding areas
- ▸The thermal effect extends up to 10 kilometres away from data centre facilities, affecting over 340 million people globally
- ▸Data centre capacity is expected to double by 2030, with AI accounting for half of new demand, intensifying future heat island effects
Summary
A landmark study by researchers at the University of Cambridge has revealed that AI data centres generate substantial localized warming, raising surface temperatures by an average of 2°C in surrounding areas, with extreme cases reaching 9.1°C. Using satellite measurements of land surface temperatures over 20 years and cross-referencing data from over 8,400 AI data centres globally, the researchers discovered that the heating effect extends up to 10 kilometres away from facilities, with 30% of the warming intensity still detectable at 7 kilometres distance. The findings have significant implications as data centre capacity is projected to double between 2025 and 2030, with AI expected to account for half of that growth.
The research estimates that over 340 million people currently live within 10 kilometres of existing data centres and are experiencing measurable temperature increases attributable to these facilities. Specific regions such as the Bajío region in Mexico and Aragon province in Spain have experienced unexplained 2°C temperature increases over the past 20 years that correlate with data centre development. While some researchers suggest follow-up studies are needed to distinguish between heat from computation versus building infrastructure, the University of Cambridge team emphasizes that data centres are definitively raising ground temperatures and cautions policymakers to carefully consider site selection and design when developing new facilities.
- The study analysed 20 years of satellite data across 8,400+ AI data centres, providing robust evidence of environmental impact
Editorial Opinion
This research reveals a critical but underappreciated environmental cost of the AI boom. As data centre capacity explodes to meet AI's voracious computational demands, the localized warming effects documented here could have cascading impacts on agriculture, ecosystems, and human health in affected regions. The scale of impact—340 million people already experiencing measurable warming—demands urgent attention from policymakers and AI companies to develop cooling solutions and implement more geographically distributed infrastructure strategies.
