Logo

AI data centers as a source of local warming: major implications

In extreme cases, data centers heat the land surface by as much as 9.1 degrees Celsius.

Published on March 31, 2026

data center

Pixabay

Masterstudente journalistiek aan de RUG, stagiair bij IO+, schrijft graag over de integratie van AI in het dagelijks leven

In a new study by researchers at the University of Cambridge in the United Kingdom, satellite temperature measurements were combined with the geographic coordinates of 8,400 AI data centers. The results show that land surface temperatures increase by an average of 2 degrees Celsius in the months following the commissioning of an AI data center. In extreme cases, the increase reached up to 9.1 degrees. The Netherlands has recently announced that it will proceed with the installation of seven data centers despite criticism. How the effects will manifest locally remains to be seen.

For the study, historical satellite temperature data were linked to the exact GPS locations of AI data centers. Locations in or near densely populated areas were excluded to minimize interference from other heat sources.

The temperature increase is not limited to the immediate surroundings of the data centers. Elevated temperatures are measured up to a distance of approximately ten kilometers. Only after about seven kilometers does the intensity of the warming clearly decrease. The researchers describe this phenomenon as “data heat islands,” local microclimates that are warmer than their surroundings.

Data heat island effect may intensify further

The analysis indicates that the findings could have significant implications. Lead researcher Andrea Marinoni of the Cambridge study expects the number and capacity of data centers to increase substantially between 2025 and 2030, with AI playing a major role in that growth. As a result, the effect of these “data heat islands” may also intensify.

“Our results show that the data heat island effect could have a significant impact on communities and the well-being of regions in the future,” Marinoni writes in the introduction to the research report. According to the researcher, the issue should therefore play a role in the broader debate on the environmental effects of AI.

A global pattern, not isolated cases

The warming effect was observed in multiple regions. In the Mexican Bajío region, a growing data center hub, temperatures increased by approximately 3.6 degrees Fahrenheit over a period of twenty years. A similar pattern was observed in Aragón, Spain, where the increases were not reflected in nearby provinces.

The study also found that the impact extends beyond the immediate vicinity, with temperature increases recorded up to 6.2 miles away. In total, more than 340 million people could be affected by this localized warming.

The Netherlands has recently announced plans to build seven new data centers. Four will be located in the Schiphol area, two in Amsterdam, and one in Lelystad. The precise impact of these facilities on temperature increases will only become clear after they are operational.

A data center is like a city

According to Professor of Energy Technology David Smeulders of Eindhoven University of Technology in an interview, the study appears scientifically robust. However, he raises several caveats. It remains unclear whether the measured warming is actually caused by data processing itself, or whether it is more closely related to the physical scale of large data centers.

Such complexes often cover vast areas, where vegetation is replaced by built infrastructure. A city, for example, has less vegetation than the countryside, which also leads to measurable temperature differences.

Additionally, the cooling methods used may play a role. Data centers typically rely on air cooling, in which warm air is expelled outside, or water cooling, in which heated water is discharged or evaporated. In both cases, it is important to better understand how these processes contribute to the observed temperature increases.

Further research is needed to more precisely understand these underlying mechanisms and to better map their environmental impact.