As AI facilities around the globe continue to ramp up demand for high-performance computing power, European nations are in a frantic race to bring new data centers online. The single biggest limiting factor to this expansion is not total energy generation—it is energy transmission, or more specifically, the lack of infrastructure to move electricity where it is needed.
While utilities experts confirm Europe is on track to produce enough overall electricity to meet this growing demand, most grid operators lack the transmission infrastructure required to deliver power to the sites where new data centers are being developed. This shortfall is squeezing available grid capacity, and in turn capping how many new energy-hungry data centers can connect to the network without raising the risk of widespread blackouts.
National Grid, which manages the high-voltage transmission network across England and Wales, reports that proposed data centers with a combined total power demand of more than 30 gigawatts are currently stuck in a connection queue. That figure equals two-thirds of Great Britain’s entire peak electricity demand. Even after accounting for the fact that some of these proposed projects will never break ground, the network currently has no spare capacity to accommodate the viable projects that remain.
Long waits for grid connection approval have already caused dozens of data center projects to collapse, undermining Europe’s ambition to capture a significant share of the hundreds of billions of dollars AI labs are investing in global computing infrastructure. “Across the continent, projects are being canceled purely because there is no grid access available to connect them,” says Taco Engelaar, managing director of grid optimization firm Neara.
Facing growing government pressure to clear the bottleneck, grid operators are testing a range of approaches to eke extra capacity out of their existing networks. These tactics range from switching to more conductive metals for power lines, to rerouting power around congested network sections, to adjusting the volume of electricity flowing through lines based on real-time weather conditions.
“There is no one simple silver bullet for this problem,” says Steve Smith, president of National Grid Partners, the venture capital arm of National Grid. “What we have to do is a little bit of everything, across multiple fronts.”
The backlog of data centers waiting for UK grid connection began swelling exponentially toward the end of 2024, shortly after the UK government designated data centers as critical national infrastructure. Since that change, connection applications have “far exceeded even our most ambitious growth forecasts,” according to UK energy regulator Ofgem, and the total size of the connection queue has tripled. “We already knew a new wave of demand was coming from the electrification of transport and heating,” Smith explains. “Now we’ve got AI demand layered on top of all that.”
The most obvious fix for capacity shortages is building new power lines, but this process is both extremely costly and notoriously slow. Depending on the scale of the development, new transmission infrastructure can take anywhere from seven to 14 years to complete, once you account for planning disputes, legal objections, supply chain delays, labor bottlenecks, and construction. “It takes time to lay new infrastructure underground, connect it to the existing network, and get line crews out to complete all the work,” says Jack Presley Abbott, deputy director for strategic planning and connections at Ofgem.
The UK’s unique geography adds extra complications. A large share of the UK’s renewable energy generation is located in Scotland and northern England, while energy consumption—including that from new data centers—is concentrated in the more densely populated south. Meanwhile, rugged terrain along the UK’s western coast means transmission routes must be routed down the eastern side of the mainland or built offshore, drastically limiting options for network expansion.
Against this backdrop, National Grid is testing retrofittable technologies that can squeeze extra capacity out of existing infrastructure, potentially allowing more data centers to connect much faster, without massive new construction. “Having large customers willing to pay to use our network is fantastic,” Smith says. “The trick is, can you find ways to connect them without having to build huge amounts of new infrastructure from scratch?”
One of the most promising options, per National Grid, is a sensor-based system called dynamic line rating (DLR), which adjusts the maximum amount of electricity that can flow through a power line to match real-time local weather conditions. When more energy passes through a line, it generates more heat, which causes the line to sag lower toward the ground, creating a safety hazard. But on cold, windy days, natural environmental cooling allows lines to safely carry more energy, boosting their usable capacity.
“A lot of operators build very cautious assumptions about how much power their lines can carry,” Engelaar says. “We estimate that around three-quarters of the UK network is capable of transporting more energy than it currently does. A relatively small increase in the amount of heat a line can handle translates to a large increase in total energy throughput—it’s a nonlinear relationship.”
An EU study found that rolling out grid-enhancing technologies like DLR could increase overall network capacity by as much as 40%, theoretically clearing enough space for waiting data centers and other large new power users to connect.
Despite these promising projections, National Grid has so far only deployed DLR to 275 kilometers of its lines, though it plans to roll the technology out to many of its busiest circuits over the next two years. “We’d love to be able to move fast and break things to speed this up, but when we cut corners on grid work, the lights go out,” Smith says.
There is another key limitation: during the periods when data centers need the most power (for example, during heat waves, when they must work overtime to cool server hardware), it is often unsafe to push extra energy through the grid. “It’s kind of the opposite of what you need,” says Keith Bell, a professor of electrical engineering at the University of Strathclyde and co-director of the UK Energy Research Center. “Data center demand peaks on hot days, but that’s exactly when your network capacity is at its lowest.”
National Grid aims to offset this issue by pairing DLR with two additional tools: technologies that divert power around congested circuits, and demand response systems that let data centers flex their consumption to match nationwide grid supply and demand. Data centers can cut usage or switch to on-site battery power when the grid is under strain. Traditional data centers, which require 24/7 uninterrupted computing, have long been treated by grid operators as inflexible sources of fixed demand. But trial data suggests AI data centers—while extremely energy-intensive—often have more intermittent workloads, and can adjust consumption without disrupting critical work. “The big unlock for AI data centers is flexibility,” Smith says. “If a hyperscale data center can provide flexibility in the periods we need it … it’ll get connected faster.”
Predicting exactly how soon grid-enhancing technologies will deliver measurable capacity gains, and how much extra power they can ultimately squeeze out of existing networks, remains an imprecise science. National Grid estimates that over the past five years, it has expanded network capacity by 16 gigawatts through a mix of grid-enhancing technologies and replacing old lines with new, more conductive alternatives.
The idea that this collection of technologies can act as a short-term stopgap—allowing more demand to connect while new long-term transmission infrastructure is built—is further undermined by current regulatory rules, which bar National Grid from factoring data center flexibility into grid connection planning decisions.
Most of the extra network capacity needed to accommodate the wave of AI data centers will ultimately have to come from newly built infrastructure. “Our plan over the next five years to double the amount of energy that flows over the network requires intervention—it requires building new overhead lines,” says David Adkins, head of network architecture and innovation at National Grid. “We are going to need to build more physical infrastructure, there’s no way around it.”
Meanwhile, to shrink the bloated grid connection queue, Ofgem is drafting reforms designed to sort the most viable, serious proposals from speculative projects submitted by developers betting on the AI boom. The regulator has also threatened financial penalties for grid operators that fail to expand capacity and meet connection deadlines.
“Getting connected sooner—that’s the name of the game, right?” says Presley Abbott. “We need to connect these data centers as soon as possible to secure our competitive advantage.”
