The United Nations Economic Commission for Europe (UNECE) warned that AI data centres and other energy-intensive facilities are being built faster than the power grids needed to supply them. That gap in timing puts pressure on how reliably electricity systems run, and how well they cope with shocks, around the world. UN News carried the warning on Sept. 8, 2026.
The International Energy Agency, in an outlook published in April 2026, projects that data centres' electricity use will nearly double, from 485 terawatt-hours in 2025 to 950 TWh in 2030, or around 3% of the world's electricity demand by then. The IEA says electricity use by AI-focused data centres grew 50% in 2025 and is expected to triple between 2025 and 2030, while overall data-centre demand grew 17% in 2025.
The core risk is timing. UN News says a large facility can often be built and connected in two to five years, while new power lines and other grid expansion can take more than 10 years because of long planning, approval and construction processes.
UNECE describes how grids could fail under stress; it does not say a collapse is inevitable. UN News reports concerns that overloading electricity systems could cause voltage swings, unintended disconnections and cascading failures, and says this is already happening in some extreme cases in systems that depend on renewable energy, where sudden swings in AI demand can be hard to balance in real time. Ireland has introduced connection restrictions in Dublin, and the Netherlands has limited where data centres may be built.
There is also the question of who pays. UN News reports there is no consistent framework for dividing grid-upgrade costs among data-centre developers, electricity companies and consumers. The IEA adds that uncertainty about peak demand, and the gap between fast-moving projects and slower energy investment, can leave grid spending out of step with demand and push up electricity prices in some places if nothing is done.
UNECE is calling for more regulation, information-sharing and coordination, including better real-time visibility of electricity use and more strategic choices about where data centres are built. The IEA goes further than a warning about demand. It proposes managing project pipelines and grid investment ahead of time. It also backs more flexibility: non-firm connections, power that can be cut back when the grid is tight; demand response, where users cut their power use when the grid is strained; and storage. And it calls for removing policy barriers so AI can also help make energy systems more secure and sustainable.
How quickly countries close the gap in build times, and who pays when concentrated data-centre demand lands on strained grids, is still unsettled.