NextFin News - The fight over a planned AI datacentre in rural Scotland has become a test of whether the country’s infrastructure pitch is an economic strategy or a political promise. In Lanarkshire, DataVita and CoreWeave are promoting an £8.2bn campus that they say will create 3,400 jobs, return up to £543m to local communities and support 1GW-plus of directly connected renewables. But the same project is now being challenged on the most basic question of all: whether its power story can be delivered on time, at scale and with public trust intact.
That question matters because the Lanarkshire site is being sold as Scotland’s first official AI growth zone and a centrepiece of Britain’s effort to build domestic AI infrastructure. The company’s own page frames the project as a sustainable AI future for North Lanarkshire, while saying the grid is already “up to 700% lower carbon than other UK regions” and that the site will go further with its own renewable buildout. Those claims sound impressive in a presentation. They are harder to translate into steel, permits and megawatts.
The anxiety is spreading beyond Lanarkshire. In Auchtertool, Fife, a separate 600MW proposal has become a flashpoint for villagers who fear a large AI facility would overwhelm a rural community. The project would span 60 hectares, run around the clock and create 120 jobs, according to the local coverage of the plan. For residents, the issue is not abstract digital growth. It is whether an industrial-scale operation can be inserted beside homes, gardens and roads without changing the character of the place entirely.
What ties the two stories together is the gap between headline ambition and operational reality. Scotland does have a cleaner grid than much of the UK, plus the land and cooling conditions that make datacentres attractive. But a project of this size still has to secure planning consent, transmission capacity, water and construction timelines that do not naturally bend to the pace of an AI investment cycle. That is why the conversation has shifted from opportunity to suspicion so quickly. The more the project is described as transformative, the more local people ask what, exactly, has already been locked in.
The power issue is central. On its website, DataVita says it will add 1GW-plus of directly connected renewables to the project. Yet critics argue that the implied energy buildout is far larger and more uncertain than the public language suggests. In February, First Minister John Swinney told DataVita that he recognised “power provision remains a key issue” and that the government would continue to work with partners to secure grid connections. That is an important admission because it shows the project’s viability is still dependent on infrastructure that has not yet been delivered.
In practical terms, that means the project is being marketed before the hardest parts of it are solved. The risk is not simply delay. It is that the renewable promise becomes the justification for approval while the actual delivery of the energy system remains unresolved.
The Promise Is Large, But So Is The Burden Of Proof
The case for the Lanarkshire project is straightforward on paper. Scotland wants investment. The UK wants AI capacity. A private developer says it can bring both while adding community money, jobs and cleaner energy infrastructure. That combination is politically potent because it connects growth, climate and regional development in one package. But it also creates an unusually high burden of proof. The larger the promise, the more detailed the evidence needs to be.
DataVita’s Lanarkshire page says the project will involve a £8.2bn investment, 3,400 jobs and a community fund of up to £543m over 15 years. It also says the company is adding natural cooling, closed-loop systems and biodiversity measures. Those features matter, but they do not answer the central question of whether the electricity supply can be made real. A datacentre campus is not green because it says green things about itself; it is green when the energy inputs are genuine, contracted and capable of serving the load.
That is why campaigners have focused so heavily on the site’s power claims. Kat Jones, director of the Association for the Protection of Rural Scotland, said:
“There is a wave of applications for hyperscale AI datacentres coming to Scotland and they all say they’re going to use renewable energy. We have examined DataVita’s plans to build renewables to power their 500MW datacentre and found them wanting.”
The quote matters because it shifts the argument away from abstract anti-development sentiment. Jones is not saying datacentres should never be built. She is saying the renewable model being attached to this one does not add up. Her criticism goes to the heart of the planning debate: if the climate case is weak, the public-interest case also weakens.
That is especially relevant because Scotland’s planning system is already under strain. SPICe, the Scottish Parliament’s information service, has noted that hyperscale datacentre decisions may need to be considered at a national level because of their cumulative impact on energy consumption. It also says the Scottish Government is giving active consideration to whether planning guidance should balance the rapid expansion of these facilities with national energy and climate goals. In plain English, the system is still catching up with the scale of what developers are proposing.
The Lanarkshire project therefore sits in a grey zone between industrial policy and local consent. Government backing helps. A cleaner grid helps. But neither one substitutes for a bankable power plan. Until the transmission and generation pieces are visibly in place, the project remains a high-value claim rather than a finished asset.
Villagers Are Reacting To More Than Noise And Traffic
The emotional force of the backlash comes from the scale mismatch between rural life and hyperscale infrastructure. A 600MW facility covering 60 hectares is not the sort of project that can be absorbed quietly into a village landscape. It would bring heavy construction activity, road traffic, round-the-clock operations and a permanent industrial presence. Even if the eventual number of operational jobs is meaningful for the region, it may not feel meaningful to the households living closest to the site.
That is why the Auchtertool reaction is so revealing. One resident, Lindsey Charnley, said:
“I can’t believe anyone would think it is appropriate to build a data centre the same size as our village on our doorstep.”
She added that the family would have no choice but to move if the plans went ahead. Her objection is not primarily ideological. It is about place, health and whether a rural community can absorb a project that large without being transformed by it. For policymakers, that distinction matters. Residents are not just objecting to technology. They are objecting to being asked to host it on terms that feel imposed.
Developers often respond by pointing to jobs and community funds. In Auchtertool, the figure on offer is 120 jobs. In Lanarkshire, the promise rises to 3,400 jobs and up to £543m for local communities. Those numbers are designed to reframe a datacentre as a regional asset rather than a local burden. But they do not solve the planning problem by themselves. A handful of permanent roles can never fully offset years of construction, landscape change and uncertainty if residents believe the development will fundamentally alter where they live.
There is also a wider policy issue. The number of hyperscale proposals in Scotland has risen fast enough that the government is now being pushed toward a more strategic approach. That matters because the cumulative effect of several giant projects can easily exceed the assumptions made when each is considered alone. One site may seem manageable. Ten sites can change energy demand, land use and public opinion dramatically. This is the point at which local planning becomes national infrastructure policy.
And that is why the Scottish government’s own language is important. In a statement tied to Lanarkshire, the First Minister acknowledged that “power provision remains a key issue.” That sentence is a quiet admission that the political story has outrun the operational one. The public was told to expect a breakthrough; the officials closest to the project are still talking about securing the basics.
The Real Test Is Whether The Energy Story Survives Scrutiny
The Lanarkshire project will ultimately be judged on whether its energy claim can survive the next round of scrutiny. If DataVita and CoreWeave can actually secure the renewables, grid connections and delivery schedule they describe, then the project could become a template for how AI infrastructure is built in a lower-carbon region. If they cannot, the project will expose a deeper problem: that major AI investments are being announced before the hardest engineering constraints are settled.
That is not a niche concern. Datacentres are capital-heavy assets that depend on power reliability, cooling, land and permitting. When any one of those inputs becomes uncertain, the economics can shift quickly. The Scottish pitch is attractive precisely because it bundles several advantages together. But bundling also increases fragility. If the power plan weakens, the whole story weakens with it.
The broader market implication is that AI infrastructure remains highly sought after, but the cost of credibility is rising. Communities are more alert, planning bodies are more cautious and the public is less willing to accept environmental claims at face value. That means the winners in the next phase are likely to be the developers who can prove delivery first and advertise ambition second.
For Lanarkshire, the question is no longer whether an AI growth zone sounds visionary. It is whether the project can still look visionary after the numbers are checked, the grid is mapped and the local consent test is applied. In infrastructure, optimism is easy to announce. The harder task is making it real. That is the difference between a growth zone and smoke and mirrors.
Explore more exclusive insights at nextfin.ai.
