# Thermal Impact of AI Data Centers on the Environment: Up to +9.1 °C
Data centers dedicated to AI tasks create significant thermal anomalies. Satellite data shows an average 2 °C rise in Earth's surface temperature in the first few months after these facilities come online. In extreme scenarios, the effect reaches 9.1 °C, impacting areas up to 10 km away.
A University of Cambridge study examined 8,400 AI data centers. The analysis focused on sites remote from major cities to rule out urban factors. Over 20 years of observations, researchers detected consistent heating independent of seasonal variations.
Scale of Thermal Pollution
The "heat island" effect extends far beyond the data center perimeter. At a distance of 7 km from a data center, heating intensity drops by just 30%, pointing to convective heat transfer through the air.
- Average temperature rise: 2 °C in the direct impact zone.
- Maximum recorded: 9.1 °C in peak cases.
- Impact radius: up to 10 km.
- Affected residents: around 340 million people within 10 km of data centers.
Forecasts from Jones Lang LaSalle predict data center capacity will double by 2030, with 50% of demand driven by AI infrastructure. This will intensify local climate shifts.
Research Methodology
Scientists compared satellite measurements of Earth's surface temperature with data center coordinates. Focusing on remote locations minimized the influence of human factors like traffic, industry, and urban development.
Key steps:
- Data collection over 20 years.
- Identification of 8,400 AI data centers.
- Correlation analysis of heating post-commissioning.
- Normalization by distance and time.
The results confirm a causal link: heating begins a few months after operations start, ruling out prior effects.
Outlook for the Growing Problem
AI development demands exponential increases in computing power. Server density in modern data centers is rising, proportionally boosting heat output. Without cooling optimizations, the effect will worsen.
University of Bristol researchers highlight caveats: some heating may come from solar radiation on buildings rather than computations alone. Ground-truth measurements are needed to separate sources.
Key takeaways:
- AI data centers cause local heating up to 9.1 °C within a 10 km radius.
- 340 million people live in at-risk zones.
- Infrastructure capacity will double by 2030.
- More research is needed on heat sources.
- The methodology rules out urban artifacts.
Thermal effects from data centers must be factored into climate models, especially in regions with high data center density.
— Editorial Team
No comments yet.