‘Slowly at first, then all at once.’ Data centre developers in the UK should bear this in mind as they look across the Atlantic. At GRID UK at the ExCeL a couple of weeks ago, data centres dominated the conversation about electricity networks and demand. For all the industry’s enthusiasm, there was little reflection on public appetite or the risks of development on this scale.
The American numbers give you pause. A September New York Times/Siena poll found 61% opposed building data centres for AI. A separate Annenberg poll recorded 61% opposition to local schemes, up 12 points since spring. Between April and June alone, 45 projects worth $68 billion were blocked or delayed.
The American power market is different and regulatory expectations are in many cases much lower. Images of hectares of data centres alongside flickering lights and water barely dribbling from kitchen taps are simply not possible here.
In Scotland, DataVita, facing hostile coverage, has invited opponents onto its site and warned that the argument appeared to be borrowing from American politics.
Britain is still in the foothills of its own argument, but the field of play is being drawn for it. It is a familiar risk that shared language enables quick cultural contamination – particularly on social media. Stories about electricity prices, water use and infrastructure travel easily into a development-sceptical UK market.
There are already signs of movement. Telehouse polling found awareness of ‘data centres’ rose from 40% in 2024 to 52% this year, while the proportion believing they positively affect digital services fell from 48% to 37%. Concern about electricity use rose from 45% to 56%. Yet Public First found more people supported further construction nationally than opposed it.
Looking at Government output, much communication designed for industry is being asked to do a second job with the public. Megawatts, billions of pounds of investment, compute capacity and Britain’s position in the AI race have their place, but they are not a convincing case to Mrs Norris of Number 64 watching a warehouse muscle its way into the sky. Nor, it has to be said, does she find it convincing that the warehouse is keen to play an ‘active and positive role in her community,’ whatever that might mean.
The Government’s language around AI Growth Zones illustrates the problem. The policy talks about resilience, security, scientific research and public services, but the public presentation leads with investment, capacity, jobs and AI leadership. The means become more visible than the ends.
The Centre for a Better Britain recently observed that regulatory failures are visible, while opportunities forgone are not. Similarly, the electricity, water and land consumed by data centres are readily counted; but the case is not being effectively made to the public about the scientific, economic and public services capacity that is built and, crucially, how that will change lives. As with Net Zero, the mechanism risks swallowing the objective.
The industry’s representative body, techUK, has looked increasingly out of its depth. Its 2025 survey found 51% of sampled sites used waterless cooling; a June explainer put the sector at just 0.2% of England’s non-household water use. But so much of its efforts are reactive and, to be honest, performative for members rather than actually shifting the dial. Opponents turn water and electricity into everyday concerns; the industry has yet to make benefits as tangible.
The argument goes: Britain needs to compete in AI, therefore Britain needs more data centres. But neither ‘AI’ nor ‘data centre’ explains why somebody should want one near them. Only 35% of Britons now say AI is having a positive impact on the world. The divergence between ‘super-users’ and those using no AI at all is growing – they are talking a different language.
Yet demand for digital infrastructure extends beyond the AI boom. Every online GP appointment needs a data centre somewhere, as do the scans, records and results doctors rely on. Annual NHS CT and MRI examinations in England rose from 10.6 million in 2021/22 to 13.2 million in 2024/25. Surely there is a case for what greater capacity could mean for patients. Developers should make it, while explaining why their projects are necessary and how local costs will be managed.
The industry cannot leave this argument to individual developers fighting planning battles site by site. In my view, developers should seek less, rather than more, coverage, and focus on finding advocates and case studies who speak about their needs rather than the data centre world focusing on supply. Similarly, government should stop trying to face in both directions, tying itself in knots. Without a simple, honest national argument, every planning application starts from scratch, and opponents are winning at providing the clearest and most comprehensible explanation of what is at stake.
America is showing the risk when the costs become politically established before the benefits are understood. Britain is nowhere near as far along that road, but data centre developers should not wait until they return to find the room already on fire.
