Denmark Halts New Grid Connections Due to 60 GW Backlog from AI Data Centers

Denmark’s grid operator Energinet paused all new grid connection agreements in March, citing a 60-gigawatt queue that has overwhelmed the system. The queue, nearly nine times Denmark’s peak demand of roughly 7 gigawatts, stems largely from surging demand by AI data centers.

Hyperscalers chose Denmark for its clean grid and cool climate, where more than 80 percent of electricity comes from renewable sources, including onshore and offshore wind farms.

AI data centers are the proximate cause of the grid connection pause, as their power needs strain infrastructure not designed for such loads. Traditional server farms were built around modest, flexible power loads, but top AI clusters now demand far more.

The power capacity of these clusters increased from around 13 MW in 2019 to an estimated 280–300 MW for xAI's Colossus in 2025, comparable to the demand of roughly 250,000 European households.

Data centers house supercomputers and associated components that underpin the increase in AI, consuming vast amounts of energy. ChatGPT-4 training consumed around 46 GWh in total energy, equivalent to a sustained 20 MW draw over three months. That amount is enough to power the entire Brussels Capital Region for over four days.

Europe's electricity grid, consisting of power lines, substations, and transmission infrastructure that moves electricity from generation to demand locations, faces serious strain from this demand. Denmark had around 398 megawatts of installed data center capacity at the start of 2026, with an additional 208 megawatts under construction. 2 gigawatts by 2030.

Hyperscale facilities operated by Microsoft, Google, and Apple account for 60 percent of Denmark’s current data center footprint. Microsoft committed $3 billion to data center construction in Denmark between 2023 and 2027. Apple operates a data center in Viborg, Denmark, while Google has expanded its Danish operations.

Euronews reported that the long-term value and acceptability of large AI compute clusters will depend on whether they are conceived, regulated and operated as critical energy infrastructure distinct from traditional data centers.