UK firms increasingly move AI workloads abroad due to rising energy costs and infrastructure challenges

A new survey reveals that one in five UK companies are shifting AI workloads overseas as soaring electricity prices and infrastructure constraints hamper domestic AI deployment, raising concerns over UK’s technological sovereignty.

One in five UK firms have already shifted AI workloads overseas because electricity costs at home are too high, according to research commissioned by CUDO Compute, underscoring how quickly power prices are becoming a strategic issue for artificial intelligence deployment. The survey of more than 700 senior AI decision-makers across the UK, US and Europe found that a third of UK respondents said energy costs were limiting their ability to scale, while 43% said cost and performance mattered more than sovereignty when deciding where to run AI systems.

The findings point to a growing tension between the UK’s ambitions for AI sovereignty and the practical limits of its infrastructure. Respondents in Britain were more likely than the broader sample to say geopolitical instability was pushing them to keep workloads closer to home, yet commercial pressures still dominated: nearly a third said they were actively considering moving workloads because of geopolitical risks, while 45% said data sovereignty, regulation or national security shaped their AI strategy. Even so, 31% said they would favour sovereign or regionally controlled compute even if it cost more.

That trade-off becomes sharper for the most demanding users. Among AI-first businesses, 32% said they would consider moving workloads abroad because of power costs, compared with 18% of larger enterprise organisations. Asked where new AI cluster capacity looked most attractive, respondents ranked the United States first, followed by India and Eastern Europe, with Western Europe and the Nordics trailing behind. The pattern suggests firms are being pulled towards markets that can offer cheaper power, more available land and better access to grid capacity.

The wider industry backdrop is consistent with that reading. Data centre executives have warned that energy, rather than raw compute, is increasingly the bottleneck for AI growth in the UK, while other research has found many companies are leaning on external partners to cope with rising power costs and denser infrastructure. Analysts also say AI workloads are changing the economics of data centres themselves, with training jobs requiring far higher rack densities, specialised cooling and large, reliable power feeds. Matt Hawkins, chief executive of CUDO Compute, said the UK risked falling behind if it did not close the gap between policy goals and the physical infrastructure needed to support them.

Source Reference Map

Inspired by headline at: ^[1]

Sources by paragraph:

Source: Fuse Wire Services