The UK faces a bigger growth hit from the US-Iran war than any other major economy. So said a study this week from the Organisation for Economic Co-operation and Development.
The British economy will expand 0.7pc during 2026, say OECD forecasters, just half the 1.4pc prediction the Office for Budget Responsibility posted as recently as November. And with lower growth comes sluggish tax revenues and higher benefit spending, further weakening Britain’s already fragile public finances.
Rachel Reeves continues to argue that everything in the UK’s economic garden is rosy – despite plenty of conflicting evidence. Consider just one aspect of the Chancellor’s ongoing cognitive dissonance – she argued earlier this month, in the annual Mais Lecture to City grandees, that Britain will “achieve the fastest AI adoption in the G7”.
Yes, artificial intelligence (AI) presents big opportunities for some, alongside job-crushing upheaval for others. But it is, overall, clearly a potential source of economic growth, productivity gains and broader wealth creation.
And yes, the UK is home to some serious tech talent which has attracted sizeable international investment – although I know plenty of eminently bankable AI whizz kids struggling to raise cash.
But does Reeves realise that AI relies on data centres. And the mass of computers packed in the data centres that drive AI generative systems demand huge amounts of energy. AI-related activities used 3.6 terawatt-hours (TWh) of electricity in 2020 and 8.3 TWh in 2024.
These demands are set to rise threefold by 2030, to 26 TWh on industry estimates. But if the sector expands 20-fold over the next five years, as per the government’s target, that implies 72 TWh of electricity demand going towards AI by 2030 – a quarter of the UK’s current total consumption.
National Grid bosses have long been warning our electricity system is “constrained”, with “bold action” needed to cope with “dramatically” growing demand. In the absence of a money-no-object investment programme and a mass culling of planning laws, our creaking electricity network hasn’t a chance of coping – and that’s before Ed Miliband’s madcap scheme to “decarbonise the grid by 2030”.
To explore these issues, I spoke to John McGee – founder and Managing Director of Durata. John has spent his career delivering complex infrastructure and engineering projects across energy, fabrication, specialising in electrical engineering.
Durata is a UK-based, globally operating specialist in critical power and modular data centre solutions. The company designs, fabricates and deploys rapid-installation infrastructure, including AI factories, offering full turnkey power distribution and in-house fabrication services.
Based in Middlesborough, in the North East of England, John leads Durata’s expansion in modular data centres across the UK, Europe and North America, supporting the rapid growth of AI, cloud computing and high-performance digital infrastructure globally.
His perspective combines hands-on engineering and supply chain experience with a strategic view of how industrial capability, infrastructure and energy policy must evolve to support the next phase of digital economic growth.
In this exclusive interview, John McGee compares the “easy, straight-forward” siting, installation and roll-out of Durata’s UK-made modular data centres in the US with the “far more complex” situation in the UK.
As of March 2026, roughly 140 proposed data centre schemes in the queue require around 50GW of electricity between them – exceeding Great Britain’s total current peak demand of 45GW. Capacity constraints mean that, despite Reeves’s rhetoric to City grandees, leading operators including Microsoft have been warned they may need to wait until the mid-2030s for full power connections. That’s why securing planning permission and power access for a data centre in the UK can currently take 5-7 years.
Desperate for a pro-growth narrative, Reeves insists the UK will “achieve the fastest AI adoption in the G7” – something that John McGee also wants to see. But he is profoundly sceptical.
“I really beg to differ regarding how the UK will be placed in the AI world,” he says. “We’ve got planning constraints and power constraints – in America, they don’t seem to have those issues like we do.
AI is currently driving one of the fastest infrastructure build-outs in modern history. Unlike traditional IT deployments, AI requires enormous computing power and specialised facilities capable of handling high-density compute loads and advanced cooling systems – and that means vast amounts of steady “base-load” electricity.
“I’m from Teesside, the North East – and I’m proud of our industrial heritage,” says McGee. “Teesside could be the global home of modular data centres – at the heart of this new industrial revolution – but unless some startling happens, we simply haven’t got the power generation capacity at the moment”.
John says small modular nuclear reactors (SMRs) could be part of the answer. “But the problem with SMRs is speed to market. As things stand, we are going to really struggle power-wise, unless something changes very quickly”
“I’m all for green energy,” McGee says. “But when we are trying to build an AI revolution, the clean energy timeframe is simply too aggressive …
“It’s all very well for politicians to jump on the latest buzzword – which is AI – but do they actually understand what it will take to be a global leader in this sector, not least when it comes to buildings and energy”.
