A new UN University report warns that AI infrastructure could consume nearly 3 percent of global electricity by 2030. The study highlights that carbon emissions are only part of the problem, as massive water requirements for cooling and land use for power generation create significant ecological strain. While training large models is resource intensive, the report finds that daily operational use now drives the bulk of the environmental cost. Generative AI searches can use up to ten times more electricity than conventional search queries, putting pressure on global power grids and local water supplies.
