The UK government made Lanarkshire one of five specially designated areas it calls AI Growth Zones. The idea was to cut through planning delays, guarantee fast access to the power grid, and attract the data centres that run artificial intelligence systems. Lanarkshire was supposed to be the flagship, with £8.2 billion in investment, 3,400 jobs, and enough green energy generated on-site to be completely self-sufficient by 2030.
None of the energy part appears to be real.
DataVita, the Scottish company developing the site alongside US firm CoreWeave, claims it will build over 1 gigawatt of wind and solar power directly connected to its data centres. One gigawatt is a large number: it is roughly the output of a small nuclear reactor and about one and a half times what the UK's largest onshore wind farm currently produces. That wind farm covers an area roughly half the size of Bristol and took years to build. DataVita currently runs two small data centres together drawing about 25 megawatts from the public grid. Its planning applications on file cover around 2 square kilometres of land. Independent analysis suggests its stated energy plans would need between 40 and 100 square kilometres.
Freedom of information requests obtained by the Guardian revealed that Scotland's First Minister privately wrote to DataVita's managing director acknowledging that "power provision remains a key issue." Officials from Nature Scotland, in a meeting shortly after the announcement, were already talking about relying on an expedited grid connection rather than self-generated power. The government's own application criteria for AI Growth Zones required sites to show either a confirmed grid connection or a credible plan to generate their own electricity. Lanarkshire had neither.
The grid connection alternative is not straightforward either. There are currently around 140 data centre projects queued up to connect to Britain's electricity grid, and their combined power requirements exceed the entire country's peak electricity demand. The queue for new connections grew by 460% in just six months during 2025, with waits of up to 15 years reported for some projects. The government is reforming the system to prioritise strategically important projects, and AI Growth Zones are on that list. So Lanarkshire may get pushed to the front. But that means it jumps ahead of hospitals, housing developments, and manufacturers who have been waiting in the same queue.
This is also not the only UK AI investment announcement that has not held up to scrutiny. CoreWeave's earlier £1 billion UK commitment, announced in 2024 with promises of "two new data centres," turned out to involve renting space in existing buildings from 2002 and 2015. The company declined to say how many jobs it had created. A separate flagship project, billed as the UK's largest sovereign AI data centre, was still operating as a scaffolding storage yard when reporters visited the site in early 2026. The government acknowledged there was no formal contract in place for a £1.9 billion investment it had publicly celebrated, and said it was not auditing the commitments it was announcing.
Two government departments are also working from incompatible numbers. The department responsible for technology forecasts that the UK will need at least 6 gigawatts of AI data centre capacity by 2030. The department responsible for energy and climate targets had, until recently, projected that the entire commercial services sector would grow by only a fraction of that. The climate emissions estimates for the same AI build-out were revised upward by more than a hundredfold after journalists asked questions.
What does this mean in practice for a business operator? If you are planning anything that depends on large-scale UK AI infrastructure being available by 2027 or 2028, treat government announcements as aspirations until there is physical evidence of construction and a confirmed grid connection date. The UK has genuine advantages: Scotland has abundant wind energy, electricity costs in growth zones will be discounted, and the planning rules are being simplified. The underlying case for building data centres in northern Britain is sound. The gap between what is being promised and what is actually being built is the problem, and that gap is currently wide.