A majority of UK public sector AI spending has so far gone on predictive analytics, according to research company Tussell.
The category accounted for £1.4 billion of a total of £2.4 billion (58%) of the spend between January 2018 and November 2024, according to the company’s research through its AI Procurement Tracker and ChatGPT.
The next largest category was data analysis with £394 million (16%), followed by research into AI with £179 million (7.4%), automation with £95 million (3.9%), machine learning with £51 million (2,1%), image recognition with £26 million (1.1%) and natural language processing with £8.9 million. £229 million of the spend (9.5%) was uncategorised.
The research identified the Met Office’s supercomputer contract with Microsoft in 2021 as the largest in value, worth £1.03 billion over 10 years.
This helped to make Microsoft the leading supplier by value with over £1 billion, followed by Palantir with £375 million, init with £259 million and HP with £176 million.
The contract also made the Met Office the largest buyer, followed by Transport for London at £518 million, NHS England at £223 million, the Ministry of Defence at £195 million and the University of Bristol at £176 million.
Procurement routes
A Tussell blogpost pointed out that just over a third of the identified contracts were awarded through framework agreements or dynamic purchasing systems, compared with a quarter for wider public sector procurements.
It also quoted Joe Hill, policy director at the Reform thinktank, in pointing out that many AI contracts have been awarded to non-specialist firms, large digital companies and systems integrators.
“This shows government has begun its AI adoption journey, and that it’s largely buying technical support rather than buying software directly,” he said, adding that the balance is likely to shift to the latter in the future.
The findings have been published after Prime Minister Sir Keir Starmer announced plans for a major effort to boost the development and use of AI in the UK public and private sectors.