Europe is in the middle of a data center sprint. AI labs around the world are gobbling compute, and national governments want a piece of the action. The snag is not how much electricity Europe can produce. The snag is how well the wires can move that electricity to the places that need it.

The queue is enormous and getting worse

In the UK alone, proposed data centers requesting grid connections represent more than 30 gigawatts of demand. That is roughly two thirds of Great Britain’s peak power use. Even if some projects never get built, there simply is not enough spare transmission capacity right now to plug them in without risking blackouts.

Things accelerated at the end of 2024, when the government labeled data centers as critical national infrastructure. Connection applications then surged and the queue has roughly tripled in size. The shortage of grid access has already forced some projects to collapse, denting Europe’s plans to attract a share of the billions being spent on AI compute.

Why building new lines is no quick fix

Building new transmission lines is the straightforward answer, but it is expensive and slow. Depending on scale and local pushback, building new transmission can take seven to fourteen years. Planning issues, legal objections, supply chain delays and the actual construction all add up.

Geography does not help. Much of the UK’s wind and hydro generation sits in Scotland and northern England, while most demand, including data centers, sits much farther south. Hard terrain in the west means new routes often need to go along the east coast or offshore, limiting options for growth.

Squeezing more out of the existing network

Because waiting a decade for wires is not popular, grid operators are testing a range of upgrades and operational tricks to increase capacity on existing lines. The goal is to connect big customers without always having to build large amounts of new infrastructure.

  • Dynamic line rating (DLR) - Sensors measure weather and conductor temperature and adjust how much power a line can carry in real time. Cooler, windier conditions let lines safely carry more current.
  • Better conductors - Replacing old lines with more conductive materials increases capacity.
  • Flow diversion software - Tools that route energy around congested circuits to free up bottlenecks.
  • Flexible demand from data centers - Encouraging data centers to reduce or shift consumption during stressed periods using on-site batteries or workload scheduling.

An EU study suggests that technologies like DLR and other grid-enhancing measures could, in theory, raise network capacity by up to 40 percent. National Grid says it has already added about 16 GW of capacity in the last five years by combining such technologies with conductor upgrades. Still, the practical rollout is cautious. So far DLR has been installed on a limited portion of the network while operators balance safety and reliability concerns.

Flexibility is the new currency

Historically, data centers were treated as inflexible, always-on loads. AI workloads complicate that picture. While peak AI computing is power hungry, some parts of that work can be scheduled or paused. Trials show that hyperscale data centers that offer flexibility - reducing power draw at short notice or shifting loads - can be easier to connect.

There is a catch. The times data centers need the most cooling - hot weather - are often when line capacity is lowest. So flexibility must be paired with smarter grid control and routing to be effective during stress events.

Rules and reforms

Regulatory rules today do not always reward flexibility. In some cases, grid planners cannot factor a data center’s ability to flex its demand into connection decisions. That limits how much these operational improvements can help in the short term.

Regulators are trying to clear the logjam. Reforms are being prepared to discourage speculative connection requests and prioritize serious projects. There are also warnings of financial penalties for grid operators that fail to increase capacity and meet connection targets.

Bottom line

Europe can generate the electric power AI data centers want. The hard part is getting that power to the machines. A mix of upgrades, operational tools and smarter consumer behavior can buy time and free capacity. But over the long term, new transmission infrastructure will still be needed. For now, grid operators are doing what regulators and governments ask: try a lot of different fixes at once while the long lead time work gets under way.