The UK Power Grid Has a Phantom Data Center Problem
Between November 2024 and June 2025 the UK grid connection queue grew from 41 gigawatts to 125 gigawatts. New data centers alone account for roughly 73 GW of that total, about 1.5× the country’s peak demand last year (per DUKES 2026, Chapter 5). That jump is not just a headline number. It creates real planning headaches for network operators and forces policymakers to weigh competitiveness for AI infrastructure against grid reliability.
What the connection queue actually is, and why it matters
The connection queue is the register of projects applying to hook into the transmission and distribution networks. Each application triggers detailed network impact studies, modelling cumulative load, fault conditions, and reinforcement needs, even if the project never completes. Those studies inform decisions on where and when to spend money on substations, transformers, and lines. If demand signals are wrong, planners either overbuild at great cost or underinvest and create risk.
Phantom data centers: what they are and why they appear
“Phantom” projects are connection applications filed without firm customers, secure funding, or a concrete build plan. They are placeholders. The incentives are simple: early grid access can raise a site’s value, make land easier to sell, or act as a cheap hedge against future demand for capacity-heavy AI and cloud facilities. Since 2024 the UK has tightened entry rules (developers must show land rights, submit planning applications and meet other milestones, per neso.energy), but speculative filings persist. Middleman developers sometimes flip sites once grid access is secured.
“It’s absolutely insane, ” said Taco Engelaar, managing director at Neara. “No one really understands … what the real grid demand will be because of these phantom projects.”
Ofgem’s proposed fix, and the policy trade-off
In July 2026 Ofgem set out proposals to clear the queue and restore predictability. The key measures are steep, nonrefundable deposits, requirements to demonstrate customers, and verification of funding. Ofgem says the reforms aim to speed the connection of viable projects and improve network planning. The consultation closes in September 2026.
“We recognize data centers are a key part of the UK’s AI ambitions and future economic growth. Enabling viable data centers to connect more quickly is an enabler of this, ”
There is a clear balancing act. Ofgem warns that for the very largest projects cumulative deposits and fees could run into the high tens or hundreds of millions of pounds, which would deter speculative filings if applied correctly but could also block legitimate projects if applied bluntly. Daniel Newton, partner at Slaughter and May, sums up the tension:
“It’s a really difficult balancing act. If you don’t set [the fee] high enough, you don’t deter speculation. If you set it too high, you risk killing viable projects, because people are unwilling to put large sums of money at risk at the development stage.”
Operational pain and strategic stakes
The immediate impact is operational. Every speculative entry forces grid planners to run full impact studies, which slows timelines and raises planning costs. Olivier Darmouni, associate professor of finance at HEC Paris, explains the consequence:
“They have to treat every project as serious when they try to study whether the system can handle them. The more speculative projects are especially damaging because they make these studies more complex, more expensive, longer.”
There is strategic risk too. Data centers are core physical infrastructure for AI compute. Large capital commitments, for example Nscale’s public pledge of £2 billion (per the company press release), signal investor interest, but capital will flow to jurisdictions with faster, lower-friction access to power and land. That is why industry voices worry reforms could deter legitimate builds even as they punish speculation.
“The frenzied data center buildout ‘is bringing a huge amount of capital investment into the UK, ’” said Alex Burgoyne, head of data centers at Knight Frank. “We don’t want to shoot the golden goose.”
Policy options that actually balance the trade-off
There is no single silver bullet, but regulators have several practical levers that can curb speculation without strangling legitimate investment. Below are options with operational examples you can expect to see discussed in the consultation and industry responses.
- Staged deposits tied to milestones. Start with a modest fee on application and escalate payments as projects clear planning, secure customers, and show funding. This shares risk between developers and the grid without forcing one large upfront cost.
- Sliding scales keyed to project size and developer track record. Hyperscalers with established balance sheets pose lower relative risk than first-time middlemen. Fee and evidence requirements can reflect that.
- Clear, objective proof-of-customer standards. Define acceptable evidence, such as signed offtake agreements, funded letters of credit, and committed cloud customers, so genuine deals are not stuck behind ambiguous tests.
- Transparency, deadlines, and penalties. Publish queue timelines, require periodic re-affirmation of intent, and apply penalties for serial withdrawals or land-flipping behaviour to raise the cost of speculative applications.
- Targeted fast tracks for strategic projects. Create an accelerated pathway or exemptions for projects tied to national AI priorities, subject to strict governance and transparency, so strategic investments are not broadly penalised.
These options trade blunt instruments for calibrated tools. They require clear rules, predictable timelines, and administrative capacity, but they reduce the risk that policy meant to protect the grid ends up protecting competitors instead of the UK market.
What executives should be doing now
- Track the consultation and respond. The details matter, since staged-deposit mechanics, evidence thresholds, and any grandfathering rules will change project economics.
- Strengthen your proof-of-demand. Secure customers, conditional power purchase agreements (PPAs), and verifiable funding commitments early. Those will lower your perceived risk under new rules.
- Stress-test site pipelines. Model scenarios where deposits rise or timelines extend. Identify which projects are marginal under tougher regimes and re-prioritise accordingly.
- Consider jurisdictional diversification. If the UK’s regime becomes blunt or slow, have contingency plans for sites in the US or EU where timelines and grid access may be more predictable.
- Engage partners. Work with legal, grid advisory, and local planning teams to structure staged commitments and to make credible submissions during the consultation window.
Questions you might be asking, short, honest answers
- Why is the connection queue so congested?
Many applicants have filed speculative, placeholder requests for grid connections. These “phantom” data centers inflate the queue even when they lack customers or funding (Ofgem data covering Nov 2024, Jun 2025).
- What is Ofgem proposing to fix it?
Ofgem’s July 2026 consultation proposes steep nonrefundable deposits, proof of customers, and funding verification. The consultation closes in September 2026 and details will be finalised after that process.
- Could these reforms scare away real investment?
Yes. If deposits and evidence tests are set too high or applied without nuance they could deter legitimate developers and push investment to other markets, a central concern voiced by industry and legal advisers.
- How big is the phantom-demand problem?
Ofgem reports the queue rose from 41 GW to 125 GW between Nov 2024 and Jun 2025, with 73 GW attributable to new data centers, roughly 1.5× last year’s UK peak demand (DUKES 2026, Chapter 5).
- When will we know the final rules?
The consultation period ends in September 2026. Implementation timing after that remains uncertain and will determine how quickly queue sizes and lead times improve.
The UK wants to be a leading base for AI infrastructure. Getting there means pruning the ghosts from the grid queue without scaring off real builders. For executives and infrastructure planners the choice is practical: prepare for tougher connection tests, lock down commercial evidence early, and, if necessary, be ready to shift where you build, because capital follows certainty as much as it follows capacity.