Findings by academic researchers suggest that hyperscalers' AI data centers contribute to local warming, but not everyone agrees.
Findings of a new study conducted by a group of academics from around the globe have revealed that land surface temperature (LST) increases by 2°C (3.6°F) on average after the start of operations of an AI data center, an effect detectable up to an estimated 10 kilometers (6.2 miles) away.
The study, The data heat island effect: quantifying the impact of AI data centers in a warming world, was conducted by a dozen experts from leading universities in the UK, Singapore, France, Italy and Hong Kong who specialize in either computer science or energy and environmental issues.
In it, they wrote, “taking advantage of land surface temperature measurements acquired by remote sensing platforms over the last decades, we are able to obtain a robust assessment of the temperature increase recorded in the areas surrounding AI data centres globally.”
Through the use of what they described as a “plethora of remotely sensed temperature measurements,” report authors estimated that the land surface temperature increases by 2°C on average after an AI data centre commences operation, inducing “local microclimate zones, which we call the data heat island effect.”
They noted, “with global data volumes growing rapidly, data centers are expected to be one of the most power-hungry [venues] in the next decade. In fact, it has been estimated that in three to five years, the power consumption for data processing will exceed the amount budgeted for manufacturing. As such, it is possible to expect that the impact of data centers and AI hyperscalers’ activities on climate might not be negligible, indeed being further exacerbated by the use of AI in the next decades.”
The findings, they stated, are based on a multiscale multimodal analysis of data collected from various sources of information and sensors, with the integration of LST data from remote sensing platforms, including those around the locations of AI hyperscalers that have been established in the last 20 years ”in order to assess the change in atmospheric heat induced by the data centers.”
The researchers also made use of a database provided by the International Energy Agency (IEA) that the authors pointed out contains more than 11,000 locations worldwide, of which 8,472 have been detected to dwell outside of highly dense urban areas. The latter locations were then used to “quantify the effect of data centers on the environment in terms of the LST gradient that could be measured on the areas surrounding each data center.”
Asking the wrong question
Asked if AI data centers are really causing local warming, or if this phenomenon is overstated, Sanchit Vir Gogia, chief analyst at Greyhound Research, said, “the signal is real, but the industry is asking the wrong question. The research shows a consistent rise in land surface temperature of around 2°C following the establishment of large data centre facilities.”
The debate, however, “has quickly shifted to causality: whether this is driven by operational heat from compute, or by land transformation during construction. That distinction matters scientifically, but it does not change the strategic implication.”
Land surface temperature, said Gogia, is not the same as air temperature, and that gap will be used to challenge the findings. “But dismissing the signal on that basis would be a mistake,” he noted. “Data centers concentrate energy use, replace natural surfaces with heat-retaining materials, and continuously reject heat into the environment. Those are known drivers of thermal change.”
He added, “the uncomfortable truth is this: Even if the exact mechanism is debated, the outcome aligns with first principles. Infrastructure at this scale alters its surroundings. The industry does not yet have a clean way to separate construction impact from operational impact, and that ambiguity makes the risk harder to model, not easier. This is not overstated, it is under-interpreted.”
Location strategy must change
But will the findings change how hyperscalers and enterprises should choose data center locations? Gogia said that location strategy is no longer just about access — it is about sustainability of conditions over time.
“The research indicates that temperature impact can extend several kilometers from a facility,” he said. “That immediately reframes the issue from site-level optimization to regional system dynamics. [But] the industry is still optimizing for individual facilities, while the risk is emerging at cluster level. As more data centres concentrate within a geography, their combined effect can begin to degrade the local operating environment. Cooling becomes less efficient. Resource competition increases.”
This leads to a critical but under-discussed outcome, he observed: “A region can become economically less viable for new data centres because of existing ones. This is the beginning of thermal saturation.”
What looks like a prime location today, he said, “can become a constrained environment tomorrow, not because of land or power alone, but because the accumulated infrastructure changes the conditions it depends on.”
Enterprises that ignore this dynamic, said Gogia, are effectively locking themselves into environments with declining marginal efficiency.
He described current AI infrastructure strategies as being incomplete in ways that lead to systematic underestimation of second-order risk: “Current strategies are built on scaling compute, securing power, and managing emissions,” he said. “All of these assume that external conditions remain stable as infrastructure grows. This research challenges that assumption directly.”
Just a regular heat island
However Jeremy Roberts, senior director at Info-Tech Research Group, had a slightly different view, stating, “a paper finding a data heat island effect is provocative, but I am skeptical of its findings and conclusions.”
The atmosphere, he said, “is an excellent heat sink. Data centers don’t generate enough heat to meaningfully increase the air temperature. I am not a physicist, but you don’t have to be one to come to this conclusion when you realize how voluminous the atmosphere actually is. Yes, you will feel heat if you stand next to a data center exhaust, but no, it’s probably not heating the ground 9 kilometers [5.6 miles] away.”
The data heat island, stated Roberts, “seems to me like it’s just a regular heat island. They built a building, and the temperature of the ground where the building was built (since that’s effectively what they’re measuring) went up. Using electricity to run AI solutions has its own set of problems, but it’s very unlikely that compute intensity is a proximate cause of temperature increases.”
Zoning and planning for data centers might get harder, he said, “and if this paper catches on, I suppose this proposed effect could have something to do with it. But I think the real reasons people don’t like them are they’re big and ugly, sometimes loud (if they’re running generators, for example), they don’t employ that many people per square foot, and they use a lot of electricity and other resources.”
The heat island effect, he said, “happens when you pave over nature. It doesn’t matter what you put there if it absorbs heat. Buildings do that, whether they’re data centers or the headquarters of environmental charities (assuming they don’t take steps like greening their roofs).”
