The Devon AI Mega-Campus: No Consent Without the Numbers.

A 1.5GW AI data centre is not a business park. It is a power-station-scale demand placed into the countryside.

North Devon has suddenly become the test case for one of the biggest questions Britain now faces: who gets priority access to land, water, electricity and grid capacity , local communities, food production and ordinary bill-payers, or the new AI industrial machine?

Xlinks has unveiled plans for what it calls the Devon Data Campus at Alverdiscott, North Devon. The company describes the scheme as a proposed AI data and energy storage development comprising two separate planning proposals: a large-scale AI-optimised data centre and an on-site battery energy storage facility. The proposed data centre would provide up to 1.5GW of AI-optimised compute capacity, supported by a 1.8GW battery energy storage system. Xlinks also claims the campus could represent £12.2–13.8 billion of capital investment, support thousands of construction jobs and create hundreds of permanent roles.

That is the sales pitch.
But behind the glossy language sits a much harder reality. A 1.5GW data centre is not a normal commercial development. It is not simply a warehouse, a technology park or a local employment scheme. It is a permanent, round-the-clock industrial electricity demand on a scale normally associated with power stations, heavy industry and national infrastructure.
And that is why the public must not be distracted by phrases such as “clean energy infrastructure”, “AI growth” or “regional investment”. The central question is brutally simple:
Where is the firm power coming from?
The scale is staggering
The House of Commons Library says the UK had approximately 1.6GW of data centre capacity in 2024. That means the Devon proposal, at up to 1.5GW, is almost equivalent in power capacity to the entire UK data-centre sector only two years ago.

That comparison should stop every councillor, MP, regulator and planning officer in their tracks.
This is not a modest addition to local infrastructure. It is a single rural project that could match nearly the whole existing national data-centre power base.
If operated continuously at full 1.5GW load, the arithmetic is stark:
1.5GW × 8,760 hours = 13.14TWh per year
Ofgem’s revised medium Typical Domestic Consumption Value from 1 July 2026 is 2,500kWh of electricity per year. On that benchmark, 13.14TWh is equivalent to the annual electricity consumption of around 5.25 million typical households.

That does not mean the site would necessarily operate at full load every hour of the year. But it does show the proper scale of the issue. This is not merely a planning application. It is a national energy allocation decision.
A battery is not a power station
The most important distinction in this entire debate is this:
Battery storage does not generate electricity.
It stores electricity that has already been produced somewhere else. A 1.8GW battery may provide balancing, backup, arbitrage, grid services or temporary support. But unless the public is shown the full duration, cycling assumptions, charging source, winter resilience model and grid-impact assessment, the battery cannot be treated as proof that the data centre has a secure 24/7 power supply.
That distinction matters even more because Xlinks says the Devon campus comprises two separate planning proposals: the AI data centre and the battery energy storage facility.

So the public must ask: if these are separate applications, can one proceed without the other? If the data centre is consented, what legally guarantees the matching firm power? If the battery is delayed, reduced, curtailed, constrained or operated for wider grid-balancing revenue, what happens to the AI load?
A battery beside a data centre is not the same as a dedicated power station. It is not equivalent to a nuclear reactor, a gas turbine, a hydro reservoir or any other source of dispatchable generation.
The campaign line should therefore be clear:
No consent without the numbers. No gigawatt data centre without a gigawatt power plan.
The countryside is being treated as industrial spare capacity
Farmers Guardian reports that the proposal would span around 850 acres between Great Torrington, Weare Giffard and Huntshaw, with campaigners warning about the loss of productive farmland, rural heritage, wildlife habitats and the character of the landscape.

That is why this cannot be dismissed as local nimbyism. The people objecting are raising issues of national importance: food security, grid capacity, land use, water use, planning fairness and democratic consent.
Britain is already watching large areas of farmland and countryside targeted for solar farms, battery storage sites, pylons, substations and transmission corridors. Now, on top of that, rural communities are being asked to absorb hyperscale AI infrastructure.
The pattern is familiar. First, communities are told the project is essential. Then they are told it will bring jobs. Then they are told the impacts can be mitigated. Then they discover that the decision was never really local at all, because the project is wrapped in the language of national need.
That is not proper planning. It is industrialisation by narrative.
Critical national infrastructure must not become a blank cheque
In September 2024, the UK Government designated data centres as part of the country’s critical national infrastructure. The House of Commons Library notes that data centres are now seen as central to AI, public services, innovation and digital activity.

But “critical” must not mean “untouchable”.
If anything, the designation should impose a higher test, not a lower one. If data centres are now critical national infrastructure, then they must be judged against national infrastructure standards: power security, water resilience, cyber resilience, environmental impact, land-use compatibility, emergency planning and grid deliverability.
The public should not be asked to accept a project of this scale on the basis of economic promises alone.
A 1.5GW AI campus must answer, in public, the questions any serious infrastructure project should answer:
Where will the electricity come from?
Is that power firm, dispatchable and available in winter?
What happens during Dunkelflaute conditions when wind and solar output are low?
How much grid reinforcement is required?
Who pays for that reinforcement?
Will local households and businesses face higher network costs?
How much water will be used for cooling?
What is the fire and emergency plan for the BESS?
What are the impacts on farmland, hedgerows, habitats and rural roads?
What happens if the data centre proceeds but the battery does not?
What happens if the battery is used for wider grid services rather than dedicated resilience?
Until those questions are answered, this proposal should not be treated as consent-ready.
The Xlinks pivot raises even bigger questions
Data Center Dynamics reports that Xlinks was previously behind the Morocco–UK Power Project, a proposed 3.6GW high-voltage direct-current interconnector that would have brought solar and wind-generated electricity from Morocco to the UK, landing at Alverdiscott. DCD reports that UK Government support for that cable was withdrawn, and that Xlinks has now pivoted toward the Devon data-centre proposal.

That is politically significant.
The original concept was about bringing large volumes of electricity into Britain. The new proposal appears to create a vast new electricity demand at or near the same grid geography.
That raises a simple public-interest question: has this location been chosen because it is genuinely suitable for a hyperscale AI campus, or because earlier energy infrastructure plans created a development pathway around Alverdiscott?
Communities deserve transparency. They should not be left to piece together the relationship between old interconnector plans, new AI demand, battery storage, grid infrastructure and private energy networks after the major commercial decisions have already been made.
AI growth cannot come before energy sovereignty
Britain does need computing capacity. AI, cloud services, public data, defence systems, hospitals, finance, logistics and research all rely on resilient digital infrastructure.
But digital infrastructure cannot be allowed to outrun physical infrastructure.
There is no serious AI strategy without a serious energy strategy. A country that cannot provide affordable, reliable electricity to households and manufacturers should not casually approve gigawatt-scale AI loads in rural areas without proving how those loads will be powered.
The national priority should be firm generation first: nuclear, small modular reactors, gas resilience, industrial-site generation, coastal energy hubs and upgraded grid infrastructure. Data centres should be located where firm power, cooling, transport, emergency access and industrial infrastructure already make sense.
They should not be imposed on productive countryside simply because land can be assembled and planning language can be made to sound strategic.
The real test: who pays?
The developer talks about investment. But every large electricity demand has system consequences.
A new 1.5GW load does not exist in isolation. It interacts with transmission capacity, distribution networks, balancing costs, reserve requirements, constraint costs and future reinforcement. If the project requires grid upgrades, the public must know whether those costs sit with the developer, the network, consumers, taxpayers or some mixture of all four.
This is the hidden problem with Britain’s current infrastructure model. Private projects are presented as private investment, but the grid consequences are often socialised through bills, standing charges and network costs.
That is why local communities must not merely ask whether the project creates jobs. They must ask whether the project increases system costs for everyone else.
If an AI campus needs power on the scale of millions of homes, then the public has a right to know whether ordinary consumers will be expected to subsidise the infrastructure needed to serve it.
North Devon should not be the sacrifice zone for AI
The Ecologist has reported campaigners describing the proposal as a “monstrous megalith” AI centre.

It is a powerful phrase because it captures the public instinct that something is out of proportion. But the stronger argument is not emotional. It is factual.
A project of this scale must not be assessed as though it were a normal local development. It must be tested as a power-station-scale demand, a land-use transformation, a water-resilience issue, a grid-capacity issue and a national planning precedent.
North Devon is not empty space. It is farmland, villages, hedgerows, wildlife, roads, homes and communities. Once countryside of that scale is industrialised, it does not easily come back.
And once the precedent is set, other rural areas will follow.
No gigawatt data centre without a gigawatt power plan
The demand must now be clear.
Before any consent is granted, Xlinks and the relevant authorities must publish a full, independently scrutinised power and infrastructure plan. That plan must show the source of firm electricity, the role and duration of the battery, the grid upgrades required, the cost allocation, the water demand, the emergency plan and the impact on farmland and local communities.
Without that, the proposal should not proceed.
Britain cannot build an AI economy on wishful thinking, battery slogans and rural sacrifice zones.
AI may be the future. But electricity is the foundation. And no country can digitise its way out of an energy crisis by pretending that batteries are power stations and farmland is industrial spare capacity.
The public test should be simple:
No consent without the numbers.
No AI campus without firm power.
No countryside sacrifice for infrastructure that has not proved its case.
Shane Oxer.   Campaigner for fairer and affordable energy