收听本期播客
阅读正文
A recent investigation by The Guardian newspaper has cast doubt on the UK’s ambitious multi-billion-pound drive into artificial intelligence (AI), suggesting that a significant portion of these investments may be based on ‘phantom investments’ and ‘shaky accounting’. Since 2024, both Conservative and Labour governments have proudly announced substantial deals aimed at developing new data centres, creating thousands of jobs, and building supercomputers, presenting these as vital for boosting the economy through technology.
However, The Guardian’s findings paint a different picture. One notable example involves Nscale, a company associated with AI giant Nvidia, which in January 2025 announced a $2.5 billion investment to construct what was described as the largest UK sovereign AI data centre and supercomputer in Loughton, Essex, with completion anticipated by 2026. Yet, the investigation revealed that the proposed site remains merely a scaffolding yard. A planning application was only submitted recently, following inquiries from the newspaper. The government itself admitted that this $2.5 billion commitment represents an ‘intention to commit capital’ rather than a formal contract for new infrastructure.
Similarly, in 2024, the government celebrated a £1 billion commitment from CoreWeave, another Nvidia-backed firm, for ‘two new data centres’ in the UK. CoreWeave later stated these were operational. However, The Guardian discovered that CoreWeave had actually become a client of *existing* data centres, renting space and deploying its Nvidia chips within facilities that were built years ago. Essentially, this investment involved transferring computer chips into the UK, not constructing new buildings. CoreWeave defended this approach as ‘industry-standard’.
The Department for Science, Innovation and Technology (DSIT), while rejecting some allegations, acknowledged limitations in its oversight, stating it was ‘not playing an active role in auditing these commitments’. Experts, such as Professor Cecilia Rikap from University College London, argue that these are ‘phantom investments’ where big tech companies inflate claims to appease governments eager to demonstrate economic growth. This raises concerns that the actual beneficiaries of these vast sums might be companies and investors primarily based outside the UK.
Another project, an ‘AI growth zone’ in Lanarkshire, backed by CoreWeave and DataVita, promises a £2.5 billion investment, 3,400 construction jobs, and one gigawatt of on-site renewable energy. Experts have questioned the feasibility of such a large energy supply, which is equivalent to a substantial share of Scotland’s total electricity demand. Again, the government indicated it would not audit these claims. Ultimately, these revelations highlight a significant lack of transparency and accountability in large-scale AI investments. Without proper oversight, there is a risk that ambitious declarations may not translate into tangible economic benefits or new infrastructure for the UK.
