OpenAI shelves Stargate UK — what it means for the UK’s ‘AI superpower’ ambitions

OpenAI’s recent decision to pause its ‘Stargate UK’ project — part of the £31bn ‘Tech Prosperity Deal’ aiming to position the UK as an ‘AI superpower’ — has brought renewed attention to a growing challenge facing the UK’s AI ambitions: energy.

As demand for AI continues to accelerate, the infrastructure required to support it — particularly access to affordable, reliable and sustainable energy — is becoming a critical constraint on where and how AI can scale.

Building on discussions from our Future of Tech conference, Maya Tajuddin and Colin Bell explore what Stargate UK reveals about the hidden energy cost of AI, outline how infrastructure constraints are influencing development and break down what this means in practice, including key takeaways for businesses planning AI-enabled growth.

Stargate UK: a case study in infrastructure pressure

Stargate UK formed part of a wider strategy to expand global AI computing capacity. It was intended to establish a large‑scale data centre in the North East to support advanced AI development within the UK, including the kind of computing infrastructure needed to train and operate advanced AI systems. 

Projects of this scale matter because they provide the technical foundations needed for AI to develop and scale. However, they also highlight the practical limits of existing infrastructure. AI‑driven data centres place substantial and sustained demands on electricity supply, bringing energy pricing, grid resilience and long‑term sustainability sharply into focus. 

This is becoming increasingly pressing as the UK moves towards a greater reliance on renewable energy where electricity generation can be variable. Against that backdrop, Stargate UK illustrates how decisions about AI investment are now closely tied to energy policy, requiring confidence that future power demands can be met in a way that is both cost‑effective and consistent with wider decarbonisation goals.

Why energy cost is becoming the limiting factor

While OpenAI has reiterated its long-term commitment to expanding within the UK market, the pause of Stargate UK underscores the reality that the data centres required to train and operate advanced AI models are among the most energy‑intensive forms of digital infrastructure. That makes them particularly sensitive to electricity pricing and therefore commercial risk.

Large AI data centres require sustained access to affordable power over many years. Without confidence that energy pricing, grid capacity and regulatory frameworks will remain supportive, projects of this scale become harder to justify.

These pressures are particularly acute for large hyperscale data centres that rely on enormous, uninterrupted power supply and often face lengthy planning processes and grid connection delays.

Environmental & community considerations

Similar issues have arisen overseas. For example, Elon Musk’s company xAI — which develops the AI model Grok — has faced sustained local opposition and regulatory scrutiny in Memphis, Tennessee, where it established one of the world’s largest AI training data centres. Community concerns have included the facility’s potential environmental impact due to large AI data centres requiring substantial volumes of water for system cooling.

While the regulatory context differs from the UK, this example helps to illustrate a broader point:  energy‑intensive AI infrastructure can give rise to planning, environmental and community considerations where energy demand outpaces existing capacity.

A greener & more practical solution? Rethinking scale & location

In response to these challenges, companies like Deep Green argue that part of the solution may lie in rethinking the scale and location of data centre infrastructure.

One alternative approach is the development of smaller, urban‑based data centres where energy consumption is lower and facilities can be more readily integrated with existing local infrastructure. Proposals advanced by such companies point towards decentralised models that make more efficient use of power and heat — including the reuse of waste energy rather than relying on the large‑scale — continuous electricity demand associated with hyperscale facilities. 

While these models won’t replace hyperscale data centres entirely, they’re a useful indicator of where the market may head as demand rises: more flexible, localised infrastructure that reduces waste and supports wider sustainability goals.

One particular advantage of smaller, urban‑based data centres is their potential to recover and reuse excess heat generated through AI processing. Unlike remote hyperscale facilities, urban data centres can be designed to capture waste heat and redirect it to nearby buildings, district heating networks or industrial uses — reducing overall energy inefficiency and supporting wider decarbonisation objectives.

Practical takeaways: what this means for businesses planning AI-enabled growth

For businesses planning AI-enabled growth, the key takeaway is that infrastructure can no longer be treated as a background consideration. Energy availability, pricing and long-term sustainability should form part of early-stage strategy rather than an afterthought, particularly where projects rely on scaling data or computing capability. 

This may mean rethinking where and how AI solutions are deployed, including exploring more localised or energy-efficient models, building in flexibility around location and capacity and factoring regulatory and planning constraints into planning timelines from the outset. Ultimately, AI investment decisions are becoming as much about access to energy, planning and power as they are about technological innovation itself.

Nscale — the British AI unicorn backed by Nvidia (and others) — is looking at alternative energy sources and supplies following concerns over delays to grid connection for one of the UK’s flagship AI infrastructure projects. On-site and off-grid energy solutions are options to avoid the delays of awaiting permanent grid connections.

The government proposals for delivering AI Growth Zones include discounts on electricity costs for eligible data centre projects, helping to address the concerns around rising energy bills.

These will inevitably be priority issues for Andy Burnham as the newly appointed Prime Minister, as well as the government’s Energy, AI and Treasury department and offices, with particular pressure on energy costs — not just for industry but also the impact on costs for the general public.

Talk to us

With energy costs and infrastructure constraints beginning to slow or reshape AI investment, businesses working in AI, data infrastructure and emerging tech aren’t just navigating innovation but the commercial, regulatory and sustainability risks that come with it.

As a certified B Corp, we’re committed to responsible, transparent and genuinely sustainable business practices and we bring that same approach to the organisations that we support. 

Our technology team combines deep sector expertise with a practical, commercial focus, helping businesses to manage complex regulation, protect and commercialise innovation, secure investment and build models that support both growth and long-term resilience. 

These topics were the focus of our Future of Tech conference in March 2026 — and our team is well placed to advise on the issues affecting the industry.

Talk to our team by calling 0333 004 4488, emailing hello@brabners.com or completing our contact form.