Denmark Halts Grid Hookups as AI Data Center Demand Hits 60GW

Denmark Hits Pause on the Grid

Denmark has become the latest country to signal that its power grid cannot keep pace with the artificial intelligence boom. Energinet, the state-run grid operator, has stopped processing new connection requests after applications for new data centers and other large installations piled up to a staggering 60 gigawatts of potential demand 1. That figure dwarfs the country's current electricity needs and underscores how aggressively data center developers — many chasing AI infrastructure contracts — have been racing to secure power capacity, even in relatively small national grids like Denmark's.

A Global Pattern of Strained Infrastructure

Denmark's move fits a broader pattern playing out worldwide as grid operators, utilities, and governments struggle to reconcile AI's voracious appetite for electricity with the physical limits of existing infrastructure. The Danish pause is notable because it marks one of the more explicit instances of a national grid operator hitting the brakes entirely on new requests rather than simply rationing or delaying approvals, reflecting just how disconnected data center demand forecasts have become from realistic near-term power supply.

The strain is not confined to Europe. In the United States, the buildout is reshaping financial markets: rising corporate borrowing to finance data center construction has been cited as a factor pushing the 10-year Treasury yield toward a breakout, with that same buildout helping propel the S&P 500 2. Meanwhile, OpenAI has committed to a massive data center campus in Pike County, Ohio, partnering with Japan's SB Energy, the U.S. Department of Energy, and Nvidia — illustrating how deeply intertwined chipmakers, energy firms, and federal agencies have become in AI infrastructure projects 3.

Commodities and Contrasts

The physical demands of this buildout extend beyond electricity to raw materials. Copper prices have surged to record highs, with data center construction cited as a key driver of demand alongside other factors, a trend that has notably boosted revenue for major miners like BHP 4. The scale of construction required for AI infrastructure — server racks, cooling systems, and power delivery — depends heavily on copper wiring and components, tying the AI boom directly to global commodity markets.

China offers a telling contrast to Denmark's grid-driven caution. Reporting indicates that China's data center expansion faces little public or regulatory pushback, with one observer noting that infrastructure siting is simply "a matter for the state" 5. There, the binding constraint on AI growth is chip availability rather than grid capacity or community opposition, a reflection of China's centralized planning apparatus and its ability to direct energy resources toward strategic priorities.

Why It Matters

Taken together, these developments suggest AI's infrastructure demands are colliding with real-world limits at different points depending on the country: grid capacity in Denmark, capital markets and financing costs in the U.S., materials supply globally, and chip access in China. As data center requests continue to outstrip what utilities can reasonably deliver, more governments may follow Denmark's lead in slowing approvals until supply catches up with demand.