Data Centres and the Politics of Scarcity

Understanding the contemporary revolt against the infrastructure of the new economy

In his first major economic speech as Green Party leader, delivered at the New Economics Foundation on 18 March 2026, Zack Polanski outlined what “Zack-o-nomics” entailed. He set out the fundamentals familiar from the British Left under Corbyn – borrowing for climate industrial policy, taxing wealth, renationalising natural monopolies, prioritising inequality – but there was a new element that is less familiar but increasingly fundamental. Like the left in America, left-populism is beginning to find a language to talk about AI.

Zack had clear words about where and for whom the benefits of AI growth will fall. He argued in his speech that, “AI in many ways has potential to be a force for good, but is already causing people to lose their jobs, consuming huge amounts of energy and water. Planned data centres will produce little employment, and blight communities, plus further jeopardise our climate targets.”

Polanski clearly articulates an increasingly salient concern of citizens across Europe, Britain, and most acutely, the United States: whether new data centres, the most visible symbols of the AI revolution, are actually in their interest. Whether they will make their lives better or not. What began as localised murmurs of concern now resembles something closer to organised, cross-spectrum opposition to the physical infrastructure of AI, as it manifests in people’s communities. What makes this new political moment distinctive is not the opposition itself – NIMBYism around large-scale infrastructure is nothing new – but the breadth of the coalition forming against data centres and the speed at which it is coalescing across the political spectrum appears to be as fast as the technological change itself. What began as smaller nodes of concern now resembles something closer to organised, cross-spectrum opposition to the physical infrastructure of AI, as it manifests in people’s communities.

I share some of these concerns. It is true that often data centres do not, as currently configured, provide enough meaningful employment to the local communities who are not otherwise bound into the technological economy that those data centres actually enable. But in Britain, the factual and geopolitical context in which this argument exists is almost completely absent.

So, do data centres create jobs?

The House of Commons Library has noted that data centres are “highly automated facilities” with modest direct employment, a point that Polanski himself has seized on and that local communities confronting proposed developments understand intuitively. Indeed, I expect that most data centres within 10 years will only be maintained by a handful of engineers, or even they could be completely autonomous.

However, it is also important to note that not all data centres are equal, and that there are many different forms of data centre from which different volumes and forms of employment can be generated, from enterprise data centres (which are purely operated by the company that owns it for internal purposes), co-location data centres, edge data centres, co-hosting data centres, hyperscale data centres, and AI data centres for training AI models.

Regardless, data centres have become legible symbols of a particular kind of economic model: capital-intensive, globally mobile, locally extractive. They arrive with vast power demands and modest employment offers, and draw on shared public infrastructure, electricity grids, water systems, planning capacity, and return their value primarily to shareholders and users elsewhere. For populists of left and right, data centres are the physical embodiment of an economy that takes more from places than it leaves behind.

This is the “family capital versus global capital” divide that runs through popular American oppositional rhetoric to data centres, and it is the “caring majority versus the wealthy elite” framing that structures Polanski’s. The data centre argument has become a proxy for a much older argument about who the economy is for.

Water scarcity?

The issue of water consumption by data centres, the rightfully most emotive element of the anti-data-centre narrative is fundamentally different in Britain, than in America, where a huge amount of the coverage and concern emanates from. The British picture of what data centres actually use for their everyday running is materially distinct, for structural and climatic reasons, and runs counter to the left-wing politics of scarcity in which Polanski’s argumentation is framed.

Indeed, industry groups have argued, and the House of Commons Library has acknowledged, that water consumption by British data centres is significantly lower than in the United States, because British data centre facilities tend not to use water-intensive cooling methods. The British climate allows many data centres to rely on air cooling or free cooling for most of the year. For example, a 2026 report commissioned by MOSL (the market operator for England’s non-household water market), prepared by WRc, provides the most granular picture available. It found that total potable water consumption by data centres in England is approximately 1,879,000 m³ per year, equivalent to just 0.2 per cent of the non-household water market. Two-thirds of English data centres use less than 1,000 m³ per year, comparable to a medium-sized office building. Indeed, TechUK also found that 51% of their surveyed data centre sites use waterless cooling systems entirely.

The problems which may arise from data centre water consumption are likely, therefore specific and manageable, concentrated in a small number of large, recent facilities, rather than systemic, or inherent to data centres in Britain themselves. The distinction matters enormously because the political narrative currently being constructed treats water consumption as an inherent and universal feature of data centres. It is not; rather, it is a function of cooling technology, climate, and regulatory environment. Thus, the British data centre rollout is unlikely to produce the kind of horror stories coming out of some of the USA, where entire towns have their water supply hamstrung by new data centres. However, the energy question for data centres is likely more critical.

The energy question is harder

On energy, Britain cannot claim the same degree of exceptionalism. Data centres currently consume around 2.5 per cent of Britain’s electricity, and the National Energy System Operator expects this to quadruple by 2030. In a country already grappling with grid capacity constraints, elevated energy prices, and net zero commitments, overconsumption of energy in a scarce environment is not a marginal concern, and it should not just be dismissed out of hand.

The honest reality of the argument is that data centres present a genuine energy planning challenge that has not yet been adequately addressed, but the benefits of hosting this infrastructure, in economic value, in sovereign capability, in the capacity to develop and deploy AI domestically, are substantial if realised. Here, the government’s AI Growth Zones, which operate as an fastraking system for data centre buildout, offer a genuine opportunity, but only if they are conceived ambitiously. The energy infrastructure required to power clusters of data centres should not be built merely to serve those facilities. It should be designed as a springboard: expanding grid capacity, accelerating renewable deployment, and bringing down energy costs for the surrounding region. If an AI Growth Zone in South Wales means cheaper, more reliable energy for local businesses and households, the political case transforms. If it means a fenced-off compound drawing power from an already strained grid, the politics will be corrosive, and the politics of scarcity, represented by Zack Polanski, will win.

Infrastructure sovereignty

There is a communication and policy gap between the industry and the government for how data centres will benefit communities and provide the critical economic growth that this country desperately needs. Often, industry treats explaining the benefits of data centres as a numbers game, of releasing annual GVA or investment figures, but this fundamentally does not feel convincing to much of the public, who are concerned with the very basics like the cost of their energy bills or overall basics. The government has oscillated between designating data centres as critical national infrastructure and fast-tracking planning approval, without building the local case for why communities should welcome them. The result is a vacuum that populists on both sides are filling with narratives that are emotionally compelling even where they are factually incomplete.

But the biggest gap is a political narrative. The reason we should want sovereign data centre capacity is for our own defensible sovereign pieces. If you believe, as much of the British left increasingly does, that the country is too dependent on American power, too exposed to the caprice of Washington, too enmeshed in an Atlantic economic architecture over which it has diminishing influence, then you should want domestic data centre capacity. As the alternative to building sovereign compute infrastructure is not some pastoral Britain untouched by AI. It is a Britain that runs its public services, its health system, its defence systems, and its economy on infrastructure hosted elsewhere, owned by others, and subject to decisions made in boardrooms and capitals over which we have no democratic say whatsoever.

The same political tendency that rightly warns of American overreach, that objects to British procurement being locked into American platforms, that calls for greater strategic autonomy in energy and trade, is simultaneously opposing the domestic infrastructure that would make technological autonomy possible. Polanski’s framing treats data centres as extractive outposts of global capital. Some are. But domestic data centre capacity is also the precondition for any serious industrial strategy in AI, for public-sector AI deployment that keeps sensitive data under British jurisdiction, and for the regulatory leverage that comes from being a significant node in the global technology stack rather than a mere consumer of services hosted elsewhere. It is entirely possible — perhaps even likely — that some years from now, a future left-wing government will quietly thank the current one for building out this capacity rather than renting it.

It is not about data centres themselves; it is what they enable

Data centres do not create many permanent operational jobs, but the construction phase creates significant employment, the supply chains generate indirect work, and — critically — data centres enable the AI and cloud computing sectors that do create high-value employment at scale, hosted in Britain. Data centres are a necessary reality of the new industrial revolution, much like how railways were critical in the 1800s in the first industrial revolution. They are the roads, the critical national infrastructure, which will enable the next economy.

More fundamentally, data centres are the material basis of technological sovereignty. Without them, every aspiration this country has, in AI, in public-sector modernisation, in defence, in industrial strategy, is built on infrastructure we do not control. The left, in particular, should recognise that opposing domestic data centre capacity in the name of anti-corporate politics is, in practice, a vote for deeper dependence on the very American technology giants it claims to distrust.

The argument by industry and government, which needs to be made in terms of what data centre infrastructure enables, not what it directly employs. And it needs to be backed by tangible commitments to local hiring, skills investment, and community benefit agreements.

This is primarily not a technical problem, but a political problem. The populist critique of data centres is not really about water or watts. Rather, it is a politics of scarcity which does come from concerns about who benefits from technological change, who bears the costs, and whether ordinary people have any say in the process of development which unfurls in their backyard, irrespective of the broader questions of whether we as a country need a particular element of technological change. But these are legitimate democratic questions, which the Government must answer to, and make policy to quell, as falling behind in the technological race will provide more damage to the British public than data centre rollout could ever aspire to. Industry and government must instead demonstrate that the AI revolution can be built in a way that communities recognise as something they will benefit from, or accept that the politics will get worse.