As AI data centres guzzle energy, can we harness their wasted heat?

What the report says

Al Jazeera reports that the fast expansion of AI data centres is adding pressure to global power systems, but engineers in Denmark are showing how some of that energy use can be captured instead of lost. The article focuses on the Bitten supercomputer at the University of Southern Denmark in Sonderborg, where a liquid-cooling system designed by Danfoss recovers all of the facility’s waste heat rather than venting it into the air.

That recovered heat is fed into Sonderborg’s district heating network, helping warm homes in the city and illustrating a broader effort to turn data centres into a useful energy source. The report says the system uses coolant directly on the processors, allowing the heat to be reused for residential heating and potentially industrial applications. Al Jazeera places this in the context of a wider AI buildout that is increasing electricity demand, water use and emissions concerns around the world.

The article cites International Energy Agency figures showing data centres used 485 terawatt hours of electricity in 2024, with demand expected to rise further by 2030 as AI grows. It also notes that the United Nations Economic Commission for Europe has warned about risks to electricity systems from large cloud providers and the data centres that support them.

Beyond Denmark, the piece points to other examples in Scandinavia and describes waste heat recovery as a possible way to reduce carbon emissions, cut energy waste and ease pressure on the grid, especially if new AI infrastructure is built where its heat can be reused.