Prime minister Andy Burnham has pledged to champion data centres, but his government's own machinery split energy, planning and digital policy across four departments, local government and two regulators in July 2026. For UK buyers chasing capacity before 2030, that fragmentation — not mayoral goodwill — is the real obstacle to a grid connection.
View the data behind this chart
| Phase | Starts (week) | Duration (weeks) |
|---|---|---|
| DSIT remit scattered | 0 | 6 |
| Ofgem queue fee reform | 6 | 24 |
| NSIP planning route applies | 12 | 36 |
| Clean power grid target | 155 | 1 |
| 6GW AI capacity target | 155 | 1 |
A reshuffle that scattered the problem, not solved it
Burnham's abolition of the Department for Science, Innovation and Technology in July 2026 did not create a single owner for data centre infrastructure — it created five. AI strategy sits with the Cabinet Office, digital functions moved to the Department for Culture, Media and Sport, energy and grid policy sits with the Department for Energy Security and Net Zero (DESNZ), water falls to the Department for Environment, Food and Rural Affairs, and planning decisions rest with local government, while Ofgem and the system operator make the actual connection calls.
For buyers planning deployments through 2030, this matters practically: there is no single Whitehall desk that can see a project's full path from land to power. Bill McCluggage, a former Cabinet Office director of IT strategy, put it plainly: a Burnham government "must reconcile localism with national priorities by accepting that datacentres are now critical national infrastructure". Until that happens, buyers should assume cross-departmental delay is the default, not the exception.
Mayoral enthusiasm can't move a grid queue
Burnham's record as Greater Manchester mayor — including backing Kao Data's £350m, 40MW facility in Stockport — shows devolution working well for convening investment, skills and local buy-in. But electricity pricing, connection queue position and planning consent at scale are all national or regulatory levers that no mayor controls.
TechUK's assessment, shared by Burnham's own former colleagues, is that electricity prices and connection timelines are the actual determinants of where data centres land — not regional ambition. Buyers should treat any mayoral or regional pitch as a starting point for site selection, not a guarantee of delivery. To explore colocation as an alternative location strategy that sidesteps some of this exposure is increasingly a sensible hedge rather than a fallback.
Speculative applications are inflating the queue — and the risk
Grid connection waits of up to 10 years were described by TechUK's Katie Davies as threatening project financing outright, rendering some schemes impossible to invest in. A large share of that backlog is speculative: applications lodged to hold a queue position rather than to build, which distort national demand forecasts for both electricity and water.
Ofgem's proposed fix is an upfront fee of up to £712,500 per megawatt to claim a queue slot. TechUK supports the principle but argues the price is wrong, warning it would deter genuine investors alongside speculators, and has instead proposed a proportionate fee, a financial capability test, and a look-back mechanism that temporarily bans applicants shown to have gamed the system. Buyers evaluating sites now should factor queue-position costs — in whatever form they finally take — directly into project economics, and finance long-term infrastructure capacity decisions around a realistic, not optimistic, connection date.
Why the 2030 power and climate targets now share one bottleneck
The government's 6GW AI-capable data centre capacity target for 2030 already looks out of reach, and it is tied to the same queue blocking the UK's clean power grid pledge. The National Energy System Operator has said a clean power system is achievable by 2030 even if data centre demand quadruples — but only if regional networks expand four times faster than over the past decade and connection delays are cleared.
That dependency means buyers can't treat grid policy and climate policy as separate risks. A stalled connections reform stalls both targets simultaneously. Teams should understand the UK data centre power landscape before committing capital to any single region, and build contingency into deployment timelines that assumes queue reform slips past 2026.
- •Water companies face the identical visibility problem as grid operators: speculative schemes block forward planning for supply
- •Clearing genuine speculative applications would shorten the real queue well below current headline figures
What buyers should do now
With no single department owning the datacentre pipeline, operators and enterprise buyers need their own risk-management approach rather than waiting for Whitehall coordination. Datacentres now fall under the Nationally Significant Infrastructure Projects regime, which gives large schemes a defined national decision route — a route worth understanding even for buyers not building at gigawatt scale, since it signals which projects Whitehall now treats as strategic.
Practical steps for 2026–2030 planning include diversifying site shortlists across regions with different queue positions, stress-testing project financing against 10-year connection scenarios rather than optimistic ones, and building power and cooling efficiency into designs from day one so that allocated capacity goes further. Buyers should also plan your data centre power and cooling around realistic allocation rather than headline capacity figures, and track UK AI data centre buildout activity to see where genuine (not speculative) capacity is actually progressing.
The superministry question buyers should watch
The strongest structural argument to emerge from this episode is for a dedicated critical infrastructure department — one with a national view across demand, grid alignment and planning timelines, able to filter speculative applications before they distort forecasts. Burnham's first reshuffle moved in the opposite direction, splitting responsibility further rather than consolidating it.
Until that changes, buyers carry the coordination risk that Whitehall hasn't resolved. Enterprises with AI workloads should treat power, not compute, as the binding constraint on deployment plans, and address AI infrastructure power bottlenecks directly in procurement and site-selection decisions through to 2030.