What if the AI revolution was also a place-making revolution?

Infrastructure has always made places

Every infrastructure revolution in Britain's history has also been a place-making revolution, whether it meant to be or not.

The railways didn't just move goods. They invented towns: Crewe and Swindon were fields until the engine sheds arrived. The great industrialists went further and built places on purpose, Saltaire, Bournville, Port Sunlight, and a hundred humbler company villages, because it seemed obvious to them that an industry and its place should prosper together. You can argue with their paternalism. You cannot argue with their belief that infrastructure owed something to the ground it stood on.

Then, somewhere in the twentieth century, we unlearned it. The motorway age gave us infrastructure that made nowhere in particular: distribution sheds, business parks, edge-lands that no one loves and no one defends. Efficient, necessary and placeless.

Now the next revolution has broken ground. Since September 2024, data centres have been classified as Critical National Infrastructure, alongside water and energy. The largest campuses are now consented nationally, like power stations. Five AI Growth Zones are designated, tens of billions of pounds are committed, and the buildings going up will still be running in the 2060s.

The AI revolution will remake British places either way. That is not a prediction, it is what infrastructure does. The only question is which tradition it joins: the one that made towns, or the one that made nowhere.

The default is already failing

Early signs suggest nowhere, and the places have noticed.

A recent review of 33 disputed UK data centre applications found planning delays averaging 490 days. Look at what delayed them longest. Not noise. Not traffic. The objections were about inadequate community engagement and the absence of any clear, tangible community benefit. In February, campaigners across Britain coordinated national days of action against hyperscale data centres. In the United States and Ireland, the pattern is further along and worse.

It is tempting for the industry to read this as a communications problem. It isn't. Communities are not anti-technology; they are declining a deal. A hyperscale campus asks a place for land, grid capacity, water and consent, and offers in return some jobs, some business rates and a fence. When residents push back, they are responding rationally to an exchange whose terms they can see perfectly well.

Which means something has quietly changed for developers too. Social licence is no longer the soft stuff. At 490 days of idle capital per disputed scheme, it is the binding constraint on the AI build-out. The fastest data centre to build in Britain is the one its town actually wants. And you cannot buy that after the design is fixed. You can only design it in.

The better deal is sitting in the racks

Here is the strange part: the thing these buildings could offer is already inside them, being thrown away.

Every data centre makes two things. The first is the product it sells: the models, the cloud, the answers. The second is heat. Almost every unit of electricity a server uses becomes warmth, so a big campus is a town-sized heater that happens to think. In Britain, banks of fans and chillers run day and night pushing that warmth into the sky. Paid for once as electricity, paid for again as cooling.

It could warm homes instead. Server halls run at about the temperature of bathwater; heat pumps lift that to what radiators need; insulated pipes carry it under the streets the way mains carry water. Stockholm has done this at city scale for a decade. Meta's campus warms eleven thousand homes in Odense. Denmark simply requires it. Research published in January found that heat from Britain's expanding fleet could warm between 3.5 and 6.3 million homes by 2035. The low end is more homes than there are in Scotland.

A child could have this idea, standing between a humming shed and a cold estate. So the interesting question is not whether it is good. It is why it almost never happens here.

Nobody owns the joins

The answer is not engineering, and it is not money. It is how Britain decides what gets built.

When a major project is planned, each expert examines one slice. An energy consultant models the power. A transport consultant models the junctions. Separate specialists assess health, jobs, carbon and flood risk. Seven consultants, seven methodologies, seven documents, one place.

Each report is competent. Together they are blind, because a town is not seven slices. It is one living system, and the value sits on the joins, exactly where nobody is paid to look. Follow one winter on one street and watch it cross five professional boundaries: a warm, dry house ends a child's cough; the child stops missing school and his mum stops missing shifts; the surgery has a quieter winter; the money that used to leave town through the meter is spent at the butcher; the gas that did all this badly never gets burned. That chain is the real return on infrastructure. No document in the current system contains it.

And notice what we do about the blindness: we prescribe more reports. Social value assessments. Health impact assessments. Climate strategies. Local area energy plans. Each one sensible, each one asking the same underlying question, what happens if we do this?, and each one answering it alone, in its own silo, on its own budget.

Learning to see

At Generation 7 we've spent years building the instrument this problem needs. We call it HAUK: a place-based intelligence resource that views a place the way a hawk does, from above, where the joins between housing, energy, health, money and nature disappear, because on the ground they never existed. It reads every place through three lenses, people live well, communities thrive, planet heals, and beneath them eighteen connected dimensions that are less abstract than they sound. They are real life, named: the back bedroom above 18 degrees in January, the cough that doesn't come back, the gas bill money spent locally.

The challenge is not whether infrastructure can create this value. The challenge is whether we design it to. That comes down to three questions, asked before any masterplan is drawn.

We know this works because we recently lived it, on a project in the north of England. The place: an estate of several thousand people with something precious under its streets, a district heat network, decades old, already connecting nearly two thousand homes. SEARCH told us what was true: an ageing energy centre, households exposed to the gas market, a network a previous generation had bought and paid for. SEE showed us what it touched: bills, health, a school and a pool within reach, renewal costs competing with rent for every pound. The plan for its low-carbon future was heading in a sensible, familiar direction, a water-source solution, well studied and entirely reasonable.

Then, in FORESEE, we tested one more scenario, almost as due diligence. What if compute came to the estate? Everything reorganised. A modular data centre could sit at the energy centre and plug almost straight into the existing pipes. Two thousand households cut loose from the gas market. A revenue stream that could fund the network's renewal. Headroom to reach the school, the pool, the streets just outside the boundary. We went in to appraise a heat source and came out having found the estate a new story to tell about itself: not a place heavy industry left behind, but a place the digital economy moves into, warmth first.

The engineering had been available all along. What changed the map was the way of looking. The process changed our minds too.

And there is a bigger inversion waiting behind it. Today, data centres choose their sites by grid economics, and heat is an afterthought. But councils are now mapping heat demand street by street, and regional energy plans arrive this year. A local plan that says "this town needs a large, constant source of warmth" is, though nobody has noticed, a data centre prospectus. The places that see this first won't wait to be chosen. They will do the choosing.

The rules are being written now

Three things are converging, and none will wait.

The Treasury's rewritten Green Book, published in February, concedes the argument: a benefit-cost ratio is not a verdict, place-based effects belong at the heart of appraisal, and whole-place business cases are being piloted in four cities. Heat network zoning arrives in England this year, with the power to require large heat sources, data centres named among them, to supply local networks. And the National Policy Statement for data centres, the rulebook for a decade of consents, is due for public consultation imminently, having slipped from February. Whatever tests it sets will govern what every campus must demonstrate to be approved.

So the asks are short. Developers: design heat offtake and local benefit into the masterplan, not the appendix; at 490 days per dispute, the gift is cheaper than the fight. Councils and housing providers: your heat maps and your estates are the negotiating table; ask for the whole-place picture before consent, not after. And everyone else: when the NPS consultation opens, one sentence would transform it. Ask every campus to show what it will create for the place that hosts it.

The hundred-year test

One of those old company villages still stands on the Bedfordshire clay. In 1926 the London Brick Company dug its pit at Stewartby and built houses, a school, a hall, playing fields and a pool around it, because that is what industry then understood itself to owe. The brickworks closed in 2008. The village is still there. People still live inside the decision.

Now a campus of national significance is proposed on that same pit: 143 acres, 720 megawatts, more warmth than the whole area could use, within a few kilometres of thousands of homes that will need clean heat within twenty years. Nobody has yet been asked to put those two facts in the same document. That sentence is this entire article in miniature.

There is an old principle, and our company is named for it, that a decision should be weighed by its effect on the seventh generation to come. Heat networks can hold that horizon; they are hundred-year assets, and pipes laid by one generation become the opportunity the next one plugs into. The future of infrastructure will not be judged only by what it delivers to the system. It will be judged by what it gives back to the places that host it. The test is simple: not whether a town tolerates its data centre, but whether it would fight to keep it.

The railways made towns. The motorways made nowhere. The AI revolution gets to choose, and it is choosing right now, in documents most people will never read.

Imagine what Britain could build once it does.

Oliver Riley is the founder and Chief Engineer at Generation 7, where infrastructure, energy and regeneration projects are shaped using HAUK, a place-based intelligence resource that sees every place through three lenses (people live well, communities thrive, planet heals) and eighteen connected dimensions, asking three questions: what is true here, how does it connect and what could this place become?