For the record, I am hopeful that the new congress will haul a number of high profile Tech Bros before committees and question them under oath about their influence in government, the concentration of their power in political and media spaces, and the dangers of unbridled AI technology.
Put me down also as firmly opposed to sentient AIs taking control of nuclear weapons and driving humanity into extinction.
That said, I’ve done a lot of pushing back on a number of myths that have been bandied about on the more hyperbolic anti-data center spaces, on both left and right wings.
One of the talking points about Data Centers that I believe has been overdone is the idea that they create enough excess heat to actually alter weather conditions locally. Much of this traces back to an unfortunately headlined article in Forbes that allowed for a lot of misinterpretation.
The study mentioned there had a number of deficiencies, as discussed above by Andy Masley, a former Physics teacher in the Washington DC area, who has carved out a platform looking into issues real and unreal in the Data Center space.
In any case, new research has thrown some cold water on the whole idea.
A new study published in Environmental Research Letters has found little evidence that hyperscale data centers cause significant heating of surrounding land. While a previous study suggested data centers heat surrounding land by about 2 °C on average (up to 9 °C), reaching 10 kilometers (6.2 miles) away, 13 years of satellite data reveal that server waste heat causes minimal nighttime surface warming, while building materials actually cool ground temperatures during the day.
Energy balance calculations in the study show that generating such large regional temperature increases would require 6 to 43 times more heat than the facilities can release.
To examine these patterns, the study tracked 13 years of thermal satellite imagery across 12 data centres built between 2011 and 2025 in Japan’s largest data centre hub near Tokyo. Comparing operating facilities against control sites and analysing daytime versus nighttime readings showed:
- Commissioning a data centre cools local ground temperatures by 1.8°C to 2.2 °C during the day, because modern engineered roofs reflect more sunlight than the cropland or open fields they replace.
- At night, when server heat exhaust would be most noticeable, ground temperature changes directly over facilities remained negligible, while multi-kilometre cluster nighttime ground warming remained under 0.3 °C.
- Cross-climate tests in dry and continental environments confirmed that nighttime surface warming occurred during construction, before servers were installed or plugged in, matching regular commercial developments like warehouses.
Slight air warming (under 1 °C) can occur within a few hundred metres of concentrated cooling vents where exhaust air blows in dry climates, but the study suggests that server heat is not strong enough to cause widespread ground warming across neighbourhoods.
The dispute over data-centre heat has run on unvalidated models, uncontrolled observations and absent regulation. This Letter supplies the missing piece: the first ex-post, controlled satellite test of what hyperscale facilities do to the temperatures around them. The answer, for Japan’s largest cluster, is threefold.
Commissioning cools the daytime surface by
about two degrees—significant under every reference, a surface radiative-property signal opposite in sign to the viral +2 ◦C claim. Night-time surface changes are bounded within tenths of a degree of zero,closing beneath the air-temperature energy-balanceceiling at cluster scale—exactly where prospective anthropogenic-heat models put them, in their first encounter with long-term observations.
And one energy budget reconciles the conflicting literature,
predicting detectable near-source air warming in dry, air-cooled settings and none of the kilometre scale surface halos claimed from coarse pixels—a prediction the arid-climate consistency check supports in a second climate. At current resolutions, what spaceborne LST sees around DCs is the surface, not the exhaust—and the protocol that shows it doubles as the ex-post assessment tool that no siting regime yet requires.
The ASU study, unlike the aforementioned study, used local transects around data centers to measure air temperature rather than the surface temperature. David Sailor, whom I have collaborated with on urban heat island research told me, “Data centers consume massive amounts of energy, and hence emit massive amounts of waste heat, but their direct thermal impact on nearby neighborhoods has remained largely unmeasured—until now.” I first became aware of this new work through a post on Sailor’s LinkedIn page. A single 169 MW data center campus rejects waste heat equivalent to that emitted by nearly 200,000 households, concentrated in an area equivalent to less than a few hundred residential parcels,” He went on to say.
The ASU study found warming just downwind of four data centers in the Phoenix metropolitan area. Values ranged from 1.5 to 4 degrees Fahrenheit as much as 500 meters from the facilities. As an atmospheric scientist, it made sense to me that waste heat might be transported downwind and detected by air temperature sensors. The waste heat – land surface connection was more difficult for me to grasp.