The rapid global expansion of data centers used to power artificial intelligence is creating “data heat islands” that significantly warm the surrounding environment, according to new research.
The study, led by researchers at the University of Cambridge, suggests that these vast AI data centers can increase local land surface temperatures by an average of 3.6 degrees Fahrenheit (2C), with some extreme cases recording rises of up to 16.4 degrees Fahrenheit (9.1C).
This localized warming effect is estimated to affect more than 340 million people worldwide.
As the tech industry races to build “hyperscale” facilities — some spanning over a million square feet — to meet the computing demands of AI, researchers are warning of a lack of oversight regarding their environmental footprint.
There are still big gaps in our understanding of the impacts of data centers, even as they boom in number, Andrea Marinoni, associate professor at the University of Cambridge and an author of the study, told CNN.
Unlike previous research focused on carbon emissions or water usage, the study, which has not yet been peer-reviewed, examined the physical heat released by server cooling systems and computation.
Researchers analyzed 20 years of satellite data from NASA sensors, mapping it against more than 6,000 data centers located away from dense urban areas to isolate the facilities’ effect from other factors such as residential heating or heavy manufacturing.
They found that the warming effect is not confined to the immediate vicinity of the buildings. Significant temperature increases were detected up to 6.2 miles away from the sites. The scale of this warming is similar to the “urban heat island” effect seen in large cities.
The study identified consistent warming trends across the globe, including in Spain’s Aragón province, where a surge of 3.6 degrees Fahrenheit (2C) stood out as an anomaly compared to neighboring regions.
A similar effect was seen in Mexico’s Bajío region, which has experienced unexplained temperature increases of approximately 3.6 degrees Fahrenheit (2C) over the last two decades as data center construction intensified.
In Brazil, researchers recorded even higher surface temperature rises of 5 degrees Fahrenheit (2.8C) across the states of Ceará and Piauí. The warming, centered around dedicated AI service centers in Teresina, was noted as particularly unusual for the region's climate.
The findings come at a time when data centers are projected to become one of the most power-hungry sectors of the global economy. Within five years, the study warns, the electricity needed for data processing will likely “exceed the amount budgeted for manufacturing” worldwide.
Deborah Andrews, emeritus professor of design for sustainability at London South Bank University, told CNN that while concerns over data centers are growing, this research is the first to focus specifically on produced heat.
“The ‘rush for AI-gold’ appears to be overriding good practice and systemic thinking,” she said, “and is developing far more rapidly than any broader, more sustainable systems.”
Other experts have urged caution.
Ralph Hintemann, a senior researcher at the Borderstep Institute, told CNN that while the figures are “interesting,” the reported effects “seem very high.” He argued that carbon emissions from power generation remain a more pressing concern.
To combat this warming, the researchers point to several solutions. These include “carbon-aware” software designed to make AI training less energy-intensive, and new hardware capable of recycling electricity rather than releasing it as heat.
The researchers also highlighted passive cooling technologies, such as radiative cooling coatings for buildings, which can reduce cooling energy consumption by between 8 percent and 20 percent.
Marinoni told CNN he hopes the research will spark a broader discussion on how to decouple technological progress from environmental degradation.
“There still might be time to consider the possibility of a different path ... without affecting the demand of AI,” Marinoni said.
