Foxglove / Global Action Plan challenge to UK hyperscale data centres (2025–ongoing)

The UK government's mid-2020s programme of fast-tracked planning approvals for hyperscale data centres — the building-scale facilities on which the current generation of frontier AI models and large-scale cloud computing physically depend — has produced the corpus's first sustained public-law challenge to AI-era infrastructure as a category. The campaign is led by Foxglove and the environmental charity Global Action Plan, supported by Leigh Day as solicitors and a counsel team led by David Wolfe KC of Matrix Chambers, with Ruchi Parekh of Cornerstone Barristers, Anna Stein of No5 Chambers, and Rowan Smith of Leigh Day. It is the first time a UK ministerial approval of a hyperscale data centre has been challenged in court, and the first of the seed's UK strategic-litigation campaigns whose primary frame is environmental rather than algorithmic-accountability.

Campaign actors and tactics

The lead case targets the 90 megawatt Woodlands Park data centre proposed by developer Greystoke on green-belt land at Iver in Buckinghamshire. The local planning authority had refused permission; then-Deputy Prime Minister Angela Rayner called the application in and approved it, the third such ministerial override of a local refusal of a data-centre application in eight months, following Court Lane in December 2024 and Abbots Langley in May 2025. On 21 August 2025 Foxglove and Global Action Plan launched a planning statutory review under Section 288 of the Town and Country Planning Act 1990 against the Secretary of State, the public-law route by which a minister's planning decision can be quashed in the High Court.

The grounds, set out in Foxglove's case-launch post, were that the Secretary of State had screened the application out of needing an Environmental Impact Assessment despite the facility's energy demand of around 90MW (comparable in scale to a small town), had relied on a factual error in stating that the planned "Iver Power Station" would supply the site directly when in fact it was a substation connecting to the local grid, had not published Affinity Water's consultation response on the impact of a facility whose hyperscale peers consume between four and nineteen million litres of water a day, and had granted permission without disclosing who would actually operate the data centre — making it impossible to know which company's environmental commitments were being relied on or how they would be enforced.

The legal action ran alongside a public campaign whose tactical shape closely tracks Foxglove's earlier UK strategic-litigation cases. Funding was raised through a Crowdjustice page that hit its initial £20,000 target and progressed to its stretch goals through autumn 2025. Foxglove paired the litigation with an evidence campaign: an October 2025 report estimating that ten UK hyperscale data centres in planning or construction would generate roughly 2.7 million tonnes of CO2 a year — almost exactly cancelling out the Climate Change Committee's 2.9-million-tonne projection of 2025 UK carbon savings from electric vehicles — which became the campaign's headline framing in Westminster and press coverage. The campaign expanded in November 2025 to cover the Ealing Council approval of the 1,610,528 MWh-a-year International Trading Estate / Trident Way data centre in Southall (planning reference 250949FUL), whose annual energy demand was roughly equivalent to powering 530,000 homes or about four times Ealing borough's residential electricity consumption, with Foxglove's advocacy director Donald Campbell calling for the Mayor of London to refuse the consent referral.

Outcome

On 19 January 2026, four days before a scheduled High Court permission hearing on 22 January, the Government Legal Department wrote to the claimants conceding the case. The letter, as reported by Local Government Lawyer, accepted that the Secretary of State's decision contained a "serious logical error" and that the Woodlands Park approval should be quashed. The legal point that landed was the one the claimants had pressed on the EIA: the government had screened the application out of needing an Environmental Impact Assessment on the basis of mitigation measures, but had then failed to secure those measures as enforceable planning conditions, with the result that the reasons for considering the development did not need an EIA were, in the Government Legal Department's own words, "inadequate." Leigh Day's press release noted the government had "failed to put in place measures to ensure the developer Greystoke and the future operator would be held to the commitments they made." The Woodlands Park permission was set to be quashed and the application returned to the Secretary of State for a fresh decision under proper EIA scrutiny.

The campaign continues. The Ealing / Southall application sits with the Mayor of London on a stage-2 referral; a second large data-centre application in Ealing on the former Honey Monster site is also pending mayoral comment. Foxglove and Global Action Plan have positioned the Woodlands Park case as a template — an early indication, in the words of Foxglove's January 2026 post, that "it shouldn't take us having to drag the Government to court for them to admit their decision to back Big Tech's polluting data centres was fundamentally wrong" — and have signalled that further public-law challenges against approved or pending hyperscale sites are in scope as the UK's data-centre acceleration programme continues to clash with local planning refusals.

Significance for the broader AI-good movement

The campaign extends the Foxglove cluster's UK public-law working method into a new register: rather than challenging an automated decision-making system deployed by a government body (the Ofqual A-level algorithm, the Home Office visa-streaming tool, or the DWP General Matching Service), it challenges a government planning decision enabling the physical infrastructure on which large-scale commercial AI systems run. The frame is therefore environmental and planning-law rather than algorithmic-accountability, but the working theory carries over: that public-law disclosure pressure can extract operational information about a system whose externalities (energy, water, grid, climate) the public is being asked to absorb, and that strategic litigation paired with crowdfunded public mobilisation can move a UK government decision that an opaque expert-led process would otherwise leave undisturbed. The shape of the win — a four-month sprint from filing to a Government Legal Department concession on EIA-screening grounds — closely echoes the Ofqual U-turn arc of August 2020, though in this case the concession was extracted in correspondence rather than under public protest, and the practical effect is a redo of the planning decision under EIA conditions rather than the abandonment of the contested system.

The campaign also opens a constituency the seed had not previously been mapped against: communities living near, or downstream of, the physical infrastructure of large-scale AI — Buckinghamshire and west-London households whose grid capacity, water supplies, and local-planning processes the data-centre build-out runs through. That positioning makes the campaign an early example of how AI-era infrastructure can be contested in the same public-law and community-organising registers that have framed earlier UK strategic-litigation cases about welfare algorithms, immigration enforcement, and educational assessment, and is plausibly the template a later wave of UK and international cases will draw on as the energy and water demands of frontier-AI infrastructure become a sharper public issue.