The Labour government’s 50-step artificial intelligence (AI) plan sets out a series of measures to develop the UK as a powerhouse for artificial intelligence.
Some of these steps, such as building sovereign AI capabilities, involve vast supercomputer investments. Although this work has already begun, with new facilities such as Bristol’s Isambard-AI and the Dawn supercomputer in Cambridge, the government wants to increase high-performance computing (HPC) capacity 20-fold by 2030.
Procuring AI
Data from Tussell’s public sector AI procurement tracker shows that while the headline figures for investment in HPC capacity to support AI look promising, the total invested is skewed by one immense contract.
Tussell calculated that £2.4bn worth of AI contracts were awarded in the UK public sector between 2018 and 2024. However, it seems this figure is not as big as the hype surrounding generative AI (GenAI) would suggest, with Tussell reporting that nearly half of that total stems from a single contract between the Met Office and Microsoft in 2021, worth £1bn.
It also reported that £2.4bn represents roughly 2.5% of the total value of all IT services and software contracts awarded by the public sector between January 2018 and October 2024. According to Tussell, this shows that while there is a lot of excitement around AI, it has not filtered into public procurement in a meaningful way.
Data silos
Beyond the computing capabilities required to power machine learning and AI inference workloads, Jacob Beswick, director of AI governance at Dataiku, believes there is some confusion over data ownership and an inability to share data in a meaningful way across the public sector.
In his experience – having previously been UK government assistant director for AI adoption and regulation, and before that, assistant director for industrial strategy at the Department for Business, Energy and Industrial Strategy – it is almost impossible to find out who in the public sector actually uses AI.
There are also questions as to what data exists, who has ownership of that data and who can access it, according to Beswick, along with “power dynamics between central departments pushing an AI agenda and departmental autonomy”.
He said there are data silos between departments and within departments, such as between different hospitals and NHS trusts, which hampers attempts not just to create effective AI models, but to create a generally digital, data-driven service.
Growing skills
Looking at the government’s five-year plan to build out computing and AI capabilities, Amanda Brock, CEO of Open UK, said: “Making the UK a global IT superpower is more than about AI and infrastructure. There are longstanding challenges around talent, funding and education that would need to be addressed to reverse that trend.
“I doubt these initiatives will make the UK ‘irresistible to AI firms’. Does the government think it has the right level of skills and experience to do this itself?”
Similarly, Ivana Bartoletti, founder of think tank Women Leading in AI and global chief privacy and AI governance officer at Wipro, believes there should be more focus on AI skills.
“There needs to be an emphasis on AI skills, extending beyond technical expertise to wider literacy and management skills,” she said. “Equipping business leaders to understand how AI can concretely solve their problems and enhance productivity, by leveraging employee insights, will be essential for widespread adoption.”
Bartoletti believes businesses need to move away from fearing AI and take more of an AI partnership mindset.
Regional tech
One of the steps in the government’s AI opportunities action plan is to identify AI growth zones, which it said would consider regional growth opportunities. The government’s goal is to take into account local growth plans and align with its industrial strategy for the digital and technology sector.
The location of AI growth zones must also be considered in relation to the energy requirements of the computing facilities these sites will require.
Shweta Singh, assistant professor of information systems and management at the University of Warwick, pointed out that AI appears to be heavily focused on London and the south-east, despite the government’s ambitions to build out AI growth zones.
“While growth zones aim to spread benefits, historically, tech innovation has often been concentrated in specific regions like London or the south-east, leaving others behind,” she said.
Singh also claimed the UK’s AI strengths are weakened by its reliance on foreign AI firms, such as Google-owned DeepMind, which raises concerns about domestic innovation and intellectual property retention.
While the general consensus from industry commentators is largely positive, the big challenge facing the government when it comes to utilising AI more effectively in the public sector is funding.
“It’s tough to see where funding for new technology would come from, particularly as the chancellor reiterates that departmental budgets are already tight, said Dataiku’s Beswick.
Implementing a robust digital and AI strategy requires proper levels of funding. As Beswick noted, if they are not centrally funded, funding for public sector AI initiatives will have to come from individual departmental budgets.
“That’s a tough sell for ministers already faced with making large amounts of savings,” he added.
