New research suggests that the rapid expansion of artificial intelligence infrastructure may be contributing to localized temperature increases around large-scale data centers. Scientists have found that these facilities, which require significant energy for computation and cooling, can create “heat islands” that raise surrounding land temperatures and potentially affect hundreds of millions of people.
The study, led by researchers including Andrea Marinoni of the University of Cambridge, examined temperature data collected over two decades and compared it with the locations of major data centers worldwide. It focused on more than 6,000 facilities situated outside densely populated urban areas to isolate their specific impact on surface temperatures, according to a report by CNN.
Researchers reported that surface temperatures increased by an average of 3.6 degrees Fahrenheit after data centers began operations. In some extreme cases, nearby temperatures rose by as much as 16.4 degrees Fahrenheit. These effects were observed across multiple regions, including areas in Mexico and Spain that have become hubs for hyperscale data infrastructure.
The study also found that the warming impact extended beyond immediate surroundings. Temperature increases were detected up to 6.2 miles from the facilities, potentially affecting more than 340 million people globally. Scientists attributed this to the continuous release of heat generated by high-performance computing systems and the energy-intensive cooling mechanisms required to maintain them.
While data centers have long been associated with high electricity consumption and carbon emissions, the localized heating effect has received comparatively less attention. The findings highlight an additional environmental dimension as demand for AI-driven services continues to accelerate.
Experts not involved in the research urged caution in interpreting the results. Some noted that while the data presents notable trends, the magnitude of temperature increases appears higher than expected and requires further validation. Others emphasized that emissions from energy production remain a primary concern in assessing the environmental footprint of data centers.
The study has not yet undergone peer review, and researchers acknowledge that further investigation is needed to confirm the findings and better understand underlying mechanisms. They also highlighted gaps in existing research, particularly regarding the cumulative environmental and societal impacts of rapidly expanding digital infrastructure.
As governments and companies invest heavily in AI capabilities, the research raises questions about how to balance technological growth with environmental considerations. The authors suggest that improved planning, more efficient cooling technologies, and sustainable energy use could help mitigate some of the identified impacts.
The findings come at a time when global temperatures are already rising due to climate change, increasing the urgency of understanding additional heat sources linked to industrial and technological development.
