Europe weighs AI data-centre growth against climate goals
Germany is easing data-centre heat rules as Europe tries to expand AI computing without abandoning its emissions targets.
Europe's push to build more data-centre capacity for artificial intelligence is running into a problem that cannot be solved by adding more servers: the infrastructure has to fit around the continent's climate goals. Germany wants to expand its AI computing capacity and reduce reliance on the United States, but its attempt to make new data centres reuse waste heat has exposed the tension between speed, cost and environmental policy.1
Germany's data-centre compromise
Earlier this year, the German government set out plans to double the country's data-centre fleet by 2030 and quadruple the computing capacity available for AI models.1 That ambition sits alongside the European Union's commitment to cut greenhouse-gas emissions by 55% by the end of the decade.2
Germany's 2023 Energy Efficiency Act required new data centres to reuse part of the heat produced by their equipment, ideally by supplying nearby district-heating networks or industrial users.1 The logic is straightforward: a facility that consumes vast amounts of electricity should not simply dump its waste heat when homes or businesses could use it.
The practical difficulty is location. A data-centre operator may be able to capture heat, but it cannot create a district-heating network, greenhouse or industrial customer where none exists. Developers argued that the original rules could make projects impossible to approve, while environmental campaigners said access to a heat customer should be a basic site-selection requirement.
In June, Germany's federal government softened the obligation before the rules were due to take effect. Under the amended approach, developers would have to assess whether waste-heat reuse is technically and economically feasible, but could avoid the requirement when the conditions are not there.1
The wider European trade-off
The change could make Germany more attractive to companies building AI infrastructure. Its central position makes it useful for routing data around Europe, and fewer hard-to-control requirements may reduce the risk that projects are abandoned before construction begins. Several proposed facilities have already fallen through, although the precise cause of each cancellation is not always clear.1
That does not make the climate question disappear. It shifts responsibility from a blanket rule to individual planning decisions, where the pressure to approve more compute may be stronger than the pressure to design genuinely efficient sites. Europe is trying to build digital sovereignty while also limiting emissions, and those goals will increasingly collide wherever electricity, water and heat networks are already under strain.
Our opinion
The sensible answer is not to choose between AI capacity and climate policy as if one must be sacrificed. Waste-heat reuse should be a serious default test for every major data-centre proposal, but the rule needs to recognise local reality rather than demand an impossible connection. The bigger failure would be approving facilities first and asking how they fit into the energy system afterwards. If Europe wants AI infrastructure that lasts, climate constraints need to be treated as part of the engineering brief, not as paperwork added at the end.