The AI Competition is Pushing Utilities to Maximize Output from Europe’s Power Networks.

European nations are vying to deploy new data centers as AI laboratories globally are increasingly demanding additional computing resources. The main constraint is energy—and specifically, its transportation.

While Europe is poised to produce sufficient energy, utility experts point out that grid operators generally lack the necessary infrastructure to deliver it where it’s required. This limitation is curbing grid capacity and, consequently, the number of energy-intensive data centers that can connect without the risk of blackouts.

According to National Grid, which manages the transmission network in England and Wales, there are proposed data centers seeking connection to its grid that represent over 30 gigawatts (GW) of power demand, which amounts to two-thirds of Great Britain’s peak demand. Even with the expectation that some of these data centers may never materialize, the current capacity is insufficient to accommodate them all.

The lengthy wait for grid connection approvals is leading some data center initiatives to fail, jeopardizing Europe’s goal of capturing a slice of the hundreds of billions of dollars AI labs are allocating to computing needs. “Throughout Europe, projects are being cancelled because access to the grid is lacking,” states Taco Engelaar, managing director at grid optimization firm Neara.

Faced with government pressure to resolve the bottleneck, grid operators are exploring various strategies to extract additional capacity from their existing networks—from altering the materials used in power lines, to circumventing congested areas, to adjusting the energy flowing through lines based on shifting weather conditions.

“There isn’t one simple answer,” explains Steve Smith, President at National Grid Partners, the venture capital arm of National Grid. “What you need to do is a little bit of everything.”

The influx of data centers waiting to connect to the UK grid surged dramatically toward the end of 2024, coinciding with the government’s designation of these centers as “critical national infrastructure.” Since that point, connection requests have “vastly exceeded even the most optimistic projections,” as reported by UK energy regulator Ofgem, and the queue has tripled. “We were aware of this new demand wave stemming from the electrification of transport and heating,” says Smith. “Now, AI has added to that.”

One clear solution is to construct new power lines, but this process is both costly and time-consuming. Depending on the project scale, it can take anywhere from seven to fourteen years to establish new transmission infrastructure, considering planning challenges, legal hurdles, supply chain, and labor constraints, as well as construction. “It requires time to lay the groundwork, make connections, and get the line workers to execute all that work,” notes Jack Presley Abbott, deputy director for strategic planning and connections at Ofgem.

The unique geography of the UK also adds to the complications. A significant portion of the UK’s renewable energy is produced in Scotland and Northern England, while energy consumption—including by data centers—occurs primarily at the opposite, more densely populated end of the country. Additionally, the challenging terrain on the UK’s western side necessitates that transmission lines be routed along the eastern portion of the country or offshore, restricting options for network expansion.

In light of these challenges, National Grid is investigating technologies that can retroactively increase grid capacity and potentially facilitate connections for additional data centers. “Large clients willing to pay for access to your network are excellent. The challenge is determining whether you can connect them without needing to construct extensive new infrastructure,” remarks Smith.