The rapid growth of artificial intelligence is creating unprecedented pressure on European power grids, as data centers race to connect and consume massive amounts of electricity. While Europe can generate enough power overall, the limiting factor is no longer supply—it’s the ability to move that energy efficiently. This bottleneck is slowing down the deployment of new data centers, undermining ambitions to capitalize on the AI boom, and even causing some projects to be cancelled outright.
The Connection Crisis: A 30GW Backlog
National Grid, the operator for England and Wales, reports over 30 gigawatts (GW) of proposed data center demand currently stuck in the connection queue—roughly two-thirds of Great Britain’s peak electricity usage. Even acknowledging that not all projects will proceed, the existing infrastructure is simply insufficient. The backlog has swelled since 2024 when data centers were classified as “critical national infrastructure,” triggering a surge in applications that has tripled in size.
The situation is not unique to the UK. Across Europe, projects are failing because grid access is unavailable. Grid operators are under intense pressure from governments to resolve the congestion, but solutions require complex upgrades, which are slow and costly.
Slow Infrastructure, Fast Demand
Building new transmission lines is the obvious long-term fix, but it can take seven to fourteen years due to planning, legal hurdles, supply chain issues, and construction delays. Meanwhile, Europe’s geography presents additional challenges. Renewable energy generation is heavily concentrated in regions like Scotland and North England, while demand, including from data centers, is concentrated in densely populated areas further south.
Given these constraints, grid operators are experimenting with ways to maximize the capacity of existing networks. These include switching materials in power lines, rerouting energy around congested areas, and dynamically adjusting energy flow based on weather conditions.
Squeezing More Out of Existing Lines
National Grid is piloting “dynamic line rating” (DLR), which uses sensors to adjust energy transmission based on real-time weather conditions. Cooler temperatures allow for higher energy throughput without exceeding safety limits. According to grid optimization company Neara, up to 75% of the UK network could potentially carry more energy with this approach.
An EU study suggests that “grid-enhancing technologies” like DLR could increase overall network capacity by 40%, but deployment remains slow. As of now, National Grid has only applied DLR to 275 km of lines, citing the risk of blackouts if pushed too aggressively.
The Flexibility Factor: AI’s Potential Advantage
A key challenge is that data center demand peaks during heatwaves, precisely when grid capacity is lowest. However, AI data centers, unlike traditional facilities, may have more flexibility in their workloads. If they can adjust consumption during peak demand or divert to on-site batteries, they could be prioritized for connection.
National Grid is signaling that flexibility will be rewarded. Hyperscale AI facilities willing to adapt their energy usage may gain faster access to the grid. However, current regulations prevent grid operators from formally factoring this flexibility into connection planning, creating a mismatch between policy and practicality.
The long-term solution remains infrastructure upgrades, but short-term fixes are proving crucial. National Grid estimates it has expanded capacity by 16GW in the last five years through a combination of grid-enhancing technologies and replacing older lines. This patchwork approach buys time, but the underlying issue—the mismatch between rapidly growing AI demand and slow infrastructure development—remains a significant threat to Europe’s ambitions in the AI race.
