MAIN ARTICLE
Electricity becomes a strategic economic input
A modern AI data centre draws electricity continuously for servers, cooling and supporting equipment. Its operators must price in the cost of power over many years, not merely the expense of constructing a building. That makes an apparently small difference in electricity tariffs important when translated into thousands of machines operating around the clock. Weckesser's warning is fundamentally about this long-term arithmetic: expensive computing makes the region less attractive for some future workloads.
Electricity price alone is not a full explanation of investment decisions. Access to specialist workers, dependable telecommunications, local data rules, financing and proximity to customers also matter. Some AI applications require data to remain near users or within particular jurisdictions. But these advantages cannot entirely offset a severe or unpredictable power disadvantage. Europe's challenge is to make its existing strengths compatible with reliable, affordable energy.
The grid bottleneck behind the AI boom
The International Energy Agency said in its 2025 Energy and AI study that the United States accounted for roughly 45% of data-centre electricity consumption in 2024, against 25% in China and 15% in Europe. Its subsequent analysis has emphasized how quickly concentrated new demand can collide with the slower timetables required to upgrade transmission lines, substations and generation capacity.

















