Are Britain’s AI Growth Zones the New Enclosures?

How a new policy is shaping land, energy, and governance to serve the territorial demands of artificial intelligence

AI-generated image by OpenAI’s ChatGPT. As the land and resources are being enclosed, the digital commons, too, is steadily fenced off through its scraping and use as training material for AI to create images like this.

As I’ve been covering in this newsletter, the race to build the physical backbone of artificial intelligence is no longer confined to server racks and algorithms—it’s reshaping territory itself. Last week, the Financial Times reported that Stargate, the $500 billion AI infrastructure venture led by OpenAI and SoftBank, is weighing future investment in the U.K. as part of its global expansion plan, drawn by government promises to fast-track energy access for data centers. While IP battles and data scraping debates have rightly focused attention on how AI feeds on the digital commons, this story signals the other, often quieter form of enclosure at work—the reorganization of land, energy, and infrastructure to secure the material foundations of computation.

Building on my earlier research on Google and Amazon’s infrastructure in Oregon, I’m now developing a set of case study geographies tracing how planning regimes are being retooled—or dismantled—in response to the rapid expansion of AI and the digital economy. If there’s a geography that is of interest, please let me know! In this essay I examine how recent planning reforms in the U.K. are reshaping territory to make space for data centers, compute clusters, and the energy demands of AI at scale.

Lastly, I’m excited to share that this newsletter recently crossed 1,000 subscribers—thank you for reading and supporting this project! If you haven’t subscribed yet, I hope you consider it, and please share if the work resonates.

---

Do AI Dream of Electric Sheep?

“Sheep, which are naturally mild, and easily kept in order, may be said now to devour men and unpeople—not only villages, but towns...”

— from Thomas More’s Utopia, 1516.

A few centuries before the rise of machine learning models and compute clusters, it was sheep that quietly reorganized the British countryside—through the slow, profitable work of grazing. As landowners fenced off common pastures to expand wool production for a global market, entire communities were uprooted to make way for a more compliant form of life. Enclosure, emerging in fragmented form during the Tudor period and intensifying under the Enclosure Acts of the eighteenth and nineteenth centuries, remade both the geography and social fabric of rural England. Backed by parliamentary sanction, millions of acres were surveyed, parceled, and transferred to landlords, while stone walls and hedges marked the erasure of customary rights. What was dismantled was not simply land use, but a subsistence economy rooted in shared access to pastures, woods, and meadows—replaced by ownership, rent, and economic productivity. Wool for the global market, not food for the village.

Enclosure was never just about agricultural modernization. It was a deliberate spatial strategy, enforced through law and state violence, to transform land from a means of livelihood into an engine of accumulation. Yet, enclosure is less a singular historical episode than a recurring political technology—a template for how capital, enabled by the state, remakes space, reorganizes social relations, and reasserts control through new regimes of extraction.¹ Its logic persists today, encoded in planning systems, infrastructure corridors, and special economic zones—contemporary enclosures that continue to reorganize extraction under the banner of progress.

Today, this logic returns—this time in the language of “growth zones” and “innovation corridors.” The UK government’s AI Growth Zones, pitched as infrastructure for the age of artificial intelligence, promise to “unleash innovation,” attract private investment, and jumpstart regional revitalization. But beneath the questionable promise of employment and the techno-optimist gloss lies a more familiar operation: the remapping of territory to meet the material demands of computation at scale. Fast-tracked planning approvals, rerouted grid capacity, and loosened environmental oversight mark these zones as spaces where the rules of land use and governance are rewritten to clear the way for digital capital.

In this essay I discuss how AI Growth Zones may be best understood as a contemporary enclosure movement—one that redirects public resources and ecological capacity toward private technological development while sidelining democratic participation and local control. The future may be framed in the sleek language of innovation—but as before, the costs of progress may very well remain unevenly distributed.

The Rise of AI Growth Zones in the U.K.

In 2024, planning applications for data centers in the UK rose by over 40%, with at least 38 new proposals—fewer than the 2020 peak of 50, but pointing to much larger compute demand, energy consumption, and physical scale as newer designs reflect the intensifying resource needs of AI infrastructure. Yet grid constraints and limited power capacity remain critical obstacles, restricting where and how quickly new development can proceed. The head of the UK’s National Grid has warned that energy demand could grow by 500% over the next decade. This combination of rising proposals and mounting energy pressures has exposed deeper strains in the planning system where local communities are wary of the impacts, prompting high-level calls for fast-tracked approvals—but without addressing these power challenges, planning reform alone risks shifting bottlenecks rather than solving them.

In the winter of 2024, with bipartisan enthusiasm and little public scrutiny, the UK government unveiled its latest industrial strategy: AI Growth Zones (AIGZs). Framed as the spearhead of a national plan to position Britain as an AI “superpower,” these zones promise to “unleash innovation,” attract private investment, and accelerate digital infrastructure development. Yet beneath all of this rhetoric lies a familiar politics of land, energy, and state power. The AIGZ initiative is not merely about fostering artificial intelligence—it is a project to reconfigure territory itself, remapping authority and ecological limits in service of computation.

Outlined in the AI Opportunities Action Plan and reinforced by a separate Planning Reform Working Paper, the policy casts the UK’s planning regime as a “brake on economic growth,” proposing to streamline approvals for data centers and related infrastructure by designating them as “Nationally Significant Infrastructure Projects.” While already designated as “Critical National Infrastructure,” this extra-significant mechanism, inherited from past planning policy but now repurposed for digital expansion, enables the central government to override local objections and bypass standard deliberative processes. In its cross-party promotion—from the outgoing Conservative government to the incoming Labour administration—the AIGZ model crystallizes a state-led strategy of accelerated accumulation which includes fast-tracking energy-intensive data infrastructures while softening the usual constraints of land use, grid access, and democratic oversight.

The AIGZs are thus engineered as enclaves of strategic exception, where infrastructural sovereignty trumps local governance. Building on the legacy of enterprise zones and freeports, AIGZs are not zones of deregulation but of targeted re-regulation—spaces where the state intervenes directly to shape conditions for capital investment. Their design reflects a basic material fact about AI at scale—it demands not only algorithms, but land, water, and energy on an industrial scale. High-voltage grid connections, sprawling sites for data campuses, cooling systems with massive water draws—these are the lifeblood of machine learning infrastructure. And it is the state, through zoning, permitting, and subsidy, that is tasked with clearing the way. Predictably, corporate beneficiaries are already in place. Companies like Vantage, Nscale, and Kyndryl have committed over £14 billion in investment, with the promise of 13,250 jobs. For these companies, the point is not to create employment per se but to consolidate control over infrastructure as a revenue stream.

The first flagship site, announced in early 2025, is Culham in Oxfordshire, home to the UK Atomic Energy Authority. There, the government brokered plans for a 100 MW “secure compute” facility with ambitions to scale fivefold, positioning it as one of Europe’s largest AI-focused data centers. Marketed as a testbed for green energy integration and next-generation data governance, the project also signals a willingness to suspend local planning control to meet national digital ambitions. Following Culham’s designation, the government issued a national call for Expressions of Interest, inviting proposals from local authorities, developers, and utilities for additional sites. The selection criteria reveal the underlying territorial calculus: access to high-capacity substations, availability of “underutilized” land, proximity to renewable generation, and alignment with regional growth narratives. But the targeted geography is telling—coastal towns, post-industrial zones, and brownfield peripheries where land is cheap, regulation is pliable, and communities are positioned as awaiting revitalization.

A key political promise of AIGZs is their potential to help rebalance regional inequalities in the UK. The government has explicitly linked AIGZs to the leveling-up agenda, pledging a “particular focus on deindustrialised areas” to ensure that “every corner of the country has a real stake in our AI-powered future.” This language draws on long-standing geographic divides between the prosperous Southeast (London, Oxford/Cambridge) and regions such as the North East, North West, South Wales, and coastal Scotland that have faced industrial decline. By early 2025, proposals emerged from sites across Scotland, Wales, and northern England, with Keir Starmer endorsing the city of Glasgow—and Scotland more broadly—as a potential host for future AIGZs. MPs from Fife similarly issued a letter to the Secretary of State for Science, Innovation and Technology, advocating for their region’s inclusion in the program.

Map of data centers in the UK and key regions of investment identified.

This strategy, however, may still collide with material and political resistance. Many eligible sites for AIGZs lie at the rural-urban fringe or adjacent to greenbelt land, where data center development has already provoked public opposition. Recent refusals in Buckinghamshire and Hertfordshire, where councils cited environmental damage and landscape harm, were some of the reasons why the government reclassified data centers as critical infrastructure, which centralized approval power in Whitehall. Such moves exemplify the governance logic at work—infrastructure follows investment, and planning follows behind. Meanwhile, the ecological contradictions remain acute. Moreover, the energy footprint of each zone—hundreds of megawatts per facility—rivals that of mid-sized towns, while water consumption for cooling raises alarms in regions like Oxfordshire, already grappling with drought risk and proposals for new reservoirs.

As these few examples illustrate, these policies confront real places, ecological limits, and social forces that may resist the scale of transformation that AI demands. To understand the deeper stakes of the AIGZ policy, let us now consider how these zones reprise the structural patterns of enclosure—remaking space, flows, and social relations to create a new infrastructural order.

The Territorial Logic of Digital Enclosure

Like the enclosure movements that redefined the British countryside centuries ago, AIGZs mark a territorial intervention designed to remake land, infrastructure, and governance in the service of a new regime of accumulation. Beneath the rhetoric of innovation lies a structural grammar of seizure and redirection—an economic strategy that repurposes state power to clear pathways for capital-intensive computation. Far from merely addressing technological undercapacity, these zones operate as contemporary enclosures, reshaping the material and political conditions of life to accommodate the needs of data-driven capitalism.

Classical enclosure commodified land by dismantling customary rights and displacing communal use, transforming shared landscapes into rent-generating assets. The AIGZ policy follow this playbook with different materials but the same premise. Here, it is not pastureland but “underutilized” territory, selectively reclassified to host energy-hungry data centers and compute clusters. Regulatory frameworks are refashioned to grease this transformation. Under proposed planning reforms, these facilities may soon qualify as “nationally significant infrastructure,” allowing the state to override local planning authorities and suppress community opposition. Ecological value, social need, and historical character are made secondary to the land’s capacity to perform computationally—to host server farms, substations, and water-intensive cooling systems. In Karl Polanyi’s terms, land remains a “fictitious commodity,” and subjecting it to market logic without regard for social or ecological consequence invites systemic unraveling. The AIGZ policy does precisely this, converting territory into throughput, indifferent to what is displaced or foreclosed along the way.

But enclosure has never been only about land. It also seizes flows—wood, water, energy, labor—commandeering the systems that sustain life. The AIGZ policy extends this logic to the infrastructures of the grid and the utility. Here, the frontier is not ownership but priority. The government’s criteria for zone selection—“500 MW-ready” sites with expandable grid capacity and access to cooling resources—foreground a political economy of scarcity, not of knowledge but of power (and water). In this calculus, infrastructural privilege is reallocated. Utilities historically managed as public goods are “splintered”² or redirected to service the logistical demands of machine learning, framed as matters of national competitiveness. The Culham site, the flagship pilot zone, exemplifies these stakes. Planned within a water-stressed region already targeted for new reservoir development, its infrastructure needs arise not from community demand but from computational modeling. Like the enclosures of the past, this is a project of infrastructural capture—repurposing collective systems to serve capital-first logics.

This process of reordering territory also hinges on legibility—the state’s capacity to make space calculable and governable. As James C. Scott argues, control begins with visibility—drawing borders, quantifying assets, standardizing usage.³ AIGZs operationalize this through calls for “high-potential” sites, soliciting local authorities to identify and discipline their landscapes according to compliance with growth-centered metrics. This mirrors David Harvey’s notion of the spatial fix—capital’s search for new spatial configurations to resolve its crises. Yet this legibility remains asymmetrical. What is rendered visible to policymakers and investors—the grid node, the vacant lot, the cooling pathway—simultaneously erases histories of use, ecological limits, and community claims. The AIGZ framework recasts territory as an open canvas for digital futures, positioning local populations as passive recipients of innovation rather than active participants in shaping spatial outcomes.

Like enclosure before it, AIGZ policy threatens not only displacement but new forms of dependency and uneven development. The zones promise “jobs and regeneration,” but the infrastructures they enable—highly automated, capital-intensive data campuses—offer limited direct employment and weak local multipliers. Instead of broad-based investment, AIGZs risk concentrating resources—public planning capacity, infrastructure budgets, political attention—into select geographies deemed ripe for digital enclosure, leaving others further marginalized. What is framed as “levelling up” may, in practice, recalibrate inequality rather than resolve it. Absent robust public mandates, ecological governance, and democratic planning, these zones institutionalize a politics of acceleration without consent, redistribution without redress. In Polanyi’s terms, they deepen commodification without countervailing protection, advancing accumulation at the expense of the social and ecological commons.

What Animates the Future

Like the “town-devouring” sheep of the enclosures, today’s animate constructs—the compute clusters, grids, and cooling towers of digital capitalism—reorder the land on behalf of the alienating logic. Philip K. Dick (author of Do Androids Dream of Electric Sheep?) once reflected on our tendency to project life onto artificial environments, imagining machines and systems as quasi-alive.⁴ But what really animates these is the design of human institutions and policies that reorganize land, energy, and governance around the imperatives of growth, surveillance, and control.

The AIGZ policy is not just a bid to secure Britain’s place in the global AI economy—it marks a deeper transformation in how territory, infrastructure, and political authority are reorganized to serve the imperatives of digital capital. Marketed as a leap into the computational future, AIGZs are another form of enclosure tethering local land and resources to global circuits of capital and information flow—and like their historical counterparts, AIGZs depend on legal abstraction, spatial control, and political exclusion. While marketed as an economic development strategy for lagging regions, it is difficult to see how the benefits would not concentrate in existing centers of investment while the burdens would simply be displaced onto peripheral communities and fragile ecologies.

More than industrial policy, AIGZs represent a mode of techno-statecraft—an infrastructural project designed to ensure the scalability of computation and the reproducibility of profit. What further distinguishes AIGZs from the early enclosures (and even traditional industrial policy) is their ideological cover. Where historical forms of enclosure were justified by appeals to improvement and efficiency, today’s are framed by narratives of inevitability, the fear of falling behind in a global technological race, the promise of regional regeneration, and the fantasy of carbon-neutral growth through machine intelligence. These narratives obscure the central political dynamic—the transfer of public authority to private corporations, as governments reorient planning, regulation, and infrastructure around the speculative demands of the tech sector. Further yet, rather than setting independent terms for technological development, Britain’s AI strategy reflects a defensive posture shaped by fear of exclusion from U.S.-dominated AI ecosystems. This anxiety over being left behind has led the UK government to prioritize alignment with American corporate interests, reinforced by concerns that diverging from pro-growth orthodoxy might jeopardize engagement with firms like OpenAI or Google DeepMind. As Starmer declared, AI will deliver “a decade of national renewal,” but only if government is “on their side… [and] won’t sit back and let opportunities slip through its fingers.”

The AIGZ policy thus crystallizes a broader politics of surrender, where the state not only clears the way for digital enclosure but formalizes its dependency through new institutional channels. The recently announced AI Energy Council, tasked with coordinating the energy demands of AI infrastructure, will give major corporate stakeholders—data center operators, cloud providers, and utilities—a privileged seat in shaping national energy strategy. Rather than governing these systems in the public interest, the council institutionalizes the priority of compute over community, embedding corporate influence directly into the planning and management of critical resources.

In sum, “national renewal,” in this vision, arrives not through shared prosperity but through deeper infrastructural dependency—binding land, energy, and governance to the demands of capital-intensive computation, while displacing its social and ecological costs onto the margins. In this, the AIGZ policy is less a break from history than its digital continuation—enclosure by other means.

References

1. See Polanyi, Karl. The Great Transformation. Farrar & Rinehart, 1941.

2. On this logic of “splintering” and the creation of premium enclaves, see Graham, Stephen, and Simon Marvin. Splintering Urbanism : Networked Infrastructures, Technological Mobilities and the Urban Condition. London and New York: Routledge, 2001.

3. Scott, James C. Seeing Like A State: How Certain Schemes to Improve the Human Condition Have Failed. Yale Agrarian Studies. New Haven: Yale University Press, 1998.

4. Dick, Philip K. “The Android and the Human.” In Vancouver Science Fiction Convention. University of British Columbia, 1972.