An economic think tank has cast doubt on the employment benefits of AI data centres, suggesting the number of permanent roles created by these facilities will fall well short of industry expectations.

Think tank Verdant said on Wednesday that the UK government and leading companies have approved a wave of new data centres based on widely cited figures from the tech industry association techUK. However, Verdant argues these numbers significantly overstate the scale of long-term employment such facilities can generate.

Verdant, which advocates for stricter planning controls on data centre development, estimates the UK's current data centres directly employ around 4,400 people. In contrast, techUK's own estimate puts the figure at 24,300. Looking ahead to projects in the pipeline, Verdant predicts roughly 10,400 permanent operational roles will be created by 2035, only about a quarter of techUK's forecast of 40,200 jobs.

techUK's calculations are based on assumed staffing levels and computing capacity for standard data centres, arriving at a figure of 8.5 jobs per megawatt. Verdant contends this is too high and bases its own estimate on older, smaller facilities with greater staffing demands. After reviewing planning documents and public personnel data for 20 proposed data centres across Britain, the think tank found a capacity-weighted average of merely 1.2 operational roles per megawatt.

Newer, larger data centres are more automated and require fewer permanent on-site staff, Verdant noted. The think tank also pointed out that techUK's broad estimate includes off-site positions, such as outsourced contractors and supply chain workers. The report further highlights that data centres generate fewer jobs per unit of electricity compared to large facilities like steel plants, car factories, hospitals, or schools, some of which create hundreds or even thousands of additional roles per megawatt.

techUK defended its figures, stating they were calculated transparently using industry data and standard economic impact assessment models based on a publicly available methodology. "The figures were derived openly using the best available industry data at the time. We remain confident they represent a reasonable estimate of the sector's employment contribution," the group said.

techUK added: "Verdant's sample size is far too small to produce reliable employment estimates. Furthermore, this report reduces the industry's economic value to simply a headcount of workers at data centre sites, overlooking the vast number of jobs supported by data centres and their associated services."

A UK government spokesperson disputed Verdant's conclusions, saying: "The 'jobs per megawatt' metric used in this report is highly misleading, the comparisons with hospitals and schools are inappropriate, and the report appears to confuse power capacity with actual electricity usage."

The spokesperson continued: "More critically, the report undervalues the wide-ranging economic benefits and overlooks the importance of building critical digital infrastructure in the UK for national security and digital sovereignty."