How much heat does an AI data centre produce, and where are they located?

Tech giants are racing to build the infrastructure that powers artificial intelligence.

But a growing body of evidence suggests that AI hyperscalers – large-scale cloud service providers like Google, Amazon and Microsoft – are warming the ground around them as well.

A study [ PDF ] by Cambridge-led researchers found that land surface temperatures around AI data centres rise by an average of 2 degrees Celsius (3.6 degrees Fahrenheit), with some areas recording increases as high as 9C (16.2F).

Researchers have called this the “data heat island effect”. Al Jazeera explains what that is, where AI data centres are concentrated and the effects on those living near these facilities.

Every time someone uses ChatGPT, Gemini or Claude, the request is handled in a data centre, a vast facility full of specialised computers that run 24 hours a day.

AI data centres use powerful chips that perform thousands of calculations in parallel and running large models continuously makes them much more energy hungry than typical servers used to browse the web.

According to the International Energy Agency (IEA), data centres consumed about 415 terawatt hours (TWh) of electricity in 2024, about 1.5 percent of global supply, growing at about 15 percent a year over the last five years. That figure is projected to nearly double to 945 TWh by 2030.

Among the most energy-intensive are hyperscale data centres – the largest facilities of their kind, built by major tech companies to support cloud computing and AI at a global scale. According to IBM, they typically house at least 5,000 servers and occupy a minimum of 10,000sq feet (930sq metres).