NextFin News - Britain’s AI growth zones are supposed to solve a real problem: how to build the power-hungry data-centre infrastructure needed for artificial intelligence faster than the usual planning and grid process allows. But the policy’s flagship example is already exposing the gap between ambition and execution. The most prominent case, an £8.2bn datacentre project in Lanarkshire backed by CoreWeave and DataVita, has been promoted as a major step in Britain’s AI build-out, yet official and internal correspondence show that power provision remains unresolved and that the public promise of a renewable-powered site by 2030 is much harder to deliver than the government’s messaging suggested.
The basic idea behind the zones is straightforward. The government wants selected areas to become fast-track locations for very large AI datacentre campuses, with planning support and coordination over energy access. The threshold discussed for these projects is 500MW or more, which is large enough to put them in a different category from ordinary commercial datacentres. At that scale, the main constraint is not whether the buildings can be designed, but whether the electricity can be secured, moved and paid for in time.
That is why the Lanarkshire project matters. It was announced in January as part of the UK’s AI infrastructure push, with a promise that it would be powered entirely from on-site renewables and built by 2030. The scale was meant to signal seriousness: a large private investment, a large energy requirement and a national policy that could make Britain look competitive in the global race for compute. Instead, the project has become a test of whether an AI growth zone is a genuine delivery mechanism or just a label attached to a difficult planning problem.
The government has also tried to bolster the broader AI strategy with national compute spending. In June it announced £1.1bn for AI hardware, including £750m for a national AI supercomputer and support for chip start-ups. That does not solve the growth-zone problem, but it shows the state is trying to treat AI infrastructure as strategic, not incidental. The logic is clear: if Britain does not host the physical backbone of AI, it risks becoming dependent on infrastructure built elsewhere. The question is whether the zone model is the right tool, or merely a politically attractive shorthand for a much harder energy-and-planning challenge.
The answer, so far, is mixed. The idea of a designated zone is not absurd. Britain does need a way to coordinate land, power and planning for compute-intensive projects that would otherwise stall in the system. But the first major example also shows how easy it is to overpromise. A datacentre can be announced quickly; power infrastructure cannot. Timelines stretch, grid queues build, and the engineering needed to support a 500MW campus is closer to utility planning than to conventional property development.
What The Policy Is Trying To Do
An AI growth zone is essentially a state-supported cluster for large AI datacentres. The point is to reduce friction: a site can be designated, infrastructure can be coordinated, and government support can make it easier to line up the electricity, land and permissions that projects of this size require. In theory, that could shorten the time between announcement and construction, which matters because global AI investment is increasingly a race for capacity, not just software talent.
That framing is sensible. AI models are becoming more compute-intensive, and the companies building them want large, reliable, and often dedicated power supplies. Britain has a chance to compete if it can offer speed, planning certainty and strategic backing. The policy also fits a wider industrial strategy that treats compute as critical infrastructure, not just another tech sector.
But the language of “growth zone” can hide how difficult the underlying task really is. A 500MW campus is not a normal data hall expanded to a bigger plot. It is a multi-layer infrastructure project that needs substations, transmission capacity, cooling, security, telecoms and, in many cases, local generation. If the grid connection is not available, or if the on-site energy plan fails, the project does not scale neatly. It stops being a growth story and becomes an engineering bottleneck.
That is the tension at the heart of Britain’s plan. The government is trying to turn a coordination problem into a policy asset. If the coordination works, the zones could unlock major private investment. If it does not, the zones are only a new name for an old problem.
Why Lanarkshire Became The Stress Test
The Lanarkshire project is the most important example because it was presented as the template for what an AI growth zone can achieve. Announced in January, it was billed as an £8.2bn investment from CoreWeave and DataVita, with completion by 2030 and power supplied entirely from on-site renewables. The pitch was attractive because it paired digital infrastructure with clean-energy rhetoric and promised jobs, investment and regional development at once.
But that combination is also what makes the project vulnerable. If the power plan is weak, the rest of the story weakens with it. The project’s scale implies very large electricity demand, while the promise of self-supplied renewables implies a huge amount of new generation and associated infrastructure. That is expensive, slow and politically sensitive. It also depends on land, weather, regulatory approvals and a long chain of engineering decisions that are much harder than a ministerial announcement.
Internal and official correspondence show that the power issue was already understood as a risk. In February, Scotland’s first minister, John Swinney, wrote to the managing director of DataVita and said:
“I recognise that power provision remains a key issue and we will continue to engage with the UK government and relevant partners to secure timely grid connections that enable and support the development to proceed at pace.”
That wording matters. It is not a dismissal of the project, but it is a public admission that the central constraint has not gone away. A growth-zone designation does not itself create grid capacity, and no amount of political optimism can substitute for a connection that does not yet exist. The promise of a renewable-powered AI campus may be directionally plausible in the long run, but the claim that it can be delivered on the original timetable looks much less secure.
The broader problem is credibility. When a flagship project is sold with a simple, clean narrative — jobs, growth, renewables, speed — and the more technical reality is that power is still uncertain, the policy starts to look like it is running ahead of the infrastructure. That does not mean the project is impossible. It means the public case for it is weaker than the public relations case.
Why The Idea Is Not Entirely Bunk
It would also be a mistake to dismiss the whole policy as fantasy. Britain does have reasons to pursue AI growth zones. The country has a large technology sector, a strong financial base, a skilled labour pool and a government willing to put compute infrastructure near the centre of industrial policy. If the state can coordinate planning and energy access more effectively, it could lower barriers that currently make major datacentre developments slow and uncertain.
There is also a strategic rationale. AI infrastructure is now a competitive asset, and countries that can host large-scale compute are better placed to attract investment, talent and spin-off industries. The UK’s June AI hardware package, which included £1.1bn and a £750m national supercomputer commitment, is evidence that ministers understand the scale of the race. They are not just trying to regulate AI use; they are trying to secure some of the infrastructure that makes AI possible.
That matters because the alternatives are unattractive. If Britain cannot provide suitable sites, companies will build elsewhere. If it can provide them, but only after years of delay, investors may conclude that the UK is too slow to matter. The growth-zone model is therefore not inherently irrational. It is a response to a real bottleneck in planning and energy access.
The problem is that the first major showcase has already shown how easily ambition can outrun delivery. A zone can be designated quickly. A datacentre can be announced quickly. But the electricity grid, the transmission upgrades and the generation build-out all move on much longer timelines. The policy’s credibility depends on whether the government can reduce that gap. So far, the evidence suggests that the gap remains wide.
The Feasibility Test Is Still Power, Planning And Time
The question of feasibility comes down to three things: power, planning and time. The power challenge is the hardest because 500MW is a very large load by British standards. The planning challenge is difficult because even fast-track zones still need local, environmental and infrastructure coordination. The time challenge is the most politically dangerous, because ministers can announce a zone immediately but cannot make substations, transmission links or generation projects appear on command.
That is why the phrase “complete bunk” captures a real frustration, even if it overstates the case. The policy is not bunk in the sense that AI infrastructure is unnecessary. It is bunk only if the government is pretending that designation alone solves the problem. The Lanarkshire case suggests that the power story is still unresolved, and that unresolved power makes every other promise less convincing.
There is also a reputational risk. If one flagship zone struggles, the government’s claims about future zones will be harder to believe. Local communities, investors and energy planners will all read the same signal: Britain wants the optics of an AI build-out, but it has not yet proved it can execute one at this scale. That is a serious problem because infrastructure policy depends on trust as much as capital.
The most realistic reading is therefore not that Britain’s AI growth zones are doomed, but that they are under-designed as a delivery story. The concept can work if it is narrowed to sites where power can genuinely be secured and if ministers are more honest about the timetable. It fails if it remains a broad promise attached to hard-to-build projects whose energy requirements have been politically softened.
The next test will be whether the government can translate the zone label into concrete progress: grid offers, planning permissions, power purchase arrangements and visible construction. If that happens, the policy may yet look like serious industrial strategy. If it does not, critics will keep calling it bunk — and the Lanarkshire case will be the reason why.
For now, Britain’s AI growth zones are best understood as an attempt to speed up a real infrastructure bottleneck, not a solved plan. The idea is plausible. The execution is not yet.
Explore more exclusive insights at nextfin.ai.
