UK

Sunak to spend £100m of taxpayer cash on AI chips in global race for computer power


James Titcomb
Sat, 19 August 2023 

Rishi Sunak

Rishi Sunak is to spend up to £100m of taxpayer money on thousands of high-powered artificial intelligence chips in an effort to catch up in a global race for computing power.

Government officials have been in discussions with IT giants Nvidia, AMD and Intel about procuring equipment for a national “AI Research Resource” as part of Rishi Sunak’s ambitions to make Britain a global leader in the field.

The effort, led by science funding body UK Research and Innovation, is believed to be in advanced stages of an order of up to 5,000 graphics processing units (GPUs) from Nvidia, whose chips power AI models such as ChatGPT.


£100m has been allocated to the project. However, it is believed that the outlay is seen as insufficient to match the Government’s artificial intelligence ambitions, with civil servants pushing Jeremy Hunt to allocate far more funds in the coming months.

GPUs are the critical components in building artificial intelligence systems such as ChatGPT, whose latest version was trained on as many as 25,000 Nvidia chips.

Nvidia’s chips power AI models such as ChatGPT - I-Hwa Cheng/Bloomberg

Mr Sunak has outlined plans for Britain to be an AI superpower but the UK severely lags the US and Europe in the computing resources needed to train, test and operate sophisticated models.

A Government review published this year criticised the lack of a “dedicated AI compute resource” with fewer than 1,000 high-end Nvidia chips available to researchers.

It recommended that at least 3,000 “top-spec” GPUs be made available as soon as possible.

Mr Hunt set aside £900m for computing resources in March, although the majority of that is expected to be spent on a traditional “exascale” supercomputer.

It is believed that slightly more than £50m was assigned to AI resources, but that the bill is expected to rise to between £70m and £100m amid a global race for the chips powering AI.

Officials are likely to press for more funding to be released in the Autumn Statement, which could take place around an AI safety summit in November.

Last week The Financial Times reported that Saudi Arabia had bought at least 3,000 Nvidia H100 processors, the company’s $40,000 high-end component for training AI.

Tech giants such as Microsoft, Amazon and Google are also racing to secure tens of thousands of the in-demand chips, while Joe Biden has blocked Nvidia from selling them in China under national security powers.

It was not clear what types of chips Britain is in talks to buy.

GPUs will be used to construct an AI Research Resource that the Government hopes will be operational by next summer.

Funds for the project are separate from a £100m taskforce that will conduct safety research into AI systems such as ChatGPT and Google Bard.

Officials have also been weighing up the merits of a “sovereign chatbot” – a publicly-funded language model similar to ChatGPT – and finding ways of boosting AI’s deployment in public services such as the NHS.

The Government has been in discussions with multiple microchip companies although Nvidia, whose chips are widely used to train AI, is seen as the clear frontrunner.

Mr Sunak is pushing for the UK to become a hub for setting global standards for the safe development of AI. He has spearheaded plans for the AI safety summit, expected to take place at the World War II codebreaking hub of Bletchley Park.

It is hoped the event will lead to international agreements between governments and top AI companies on developing the technology.

A Government spokesman said: “We are committed to supporting a thriving environment for compute in the UK which maintains our position as a global leader across science, innovation and technology.”

Nvidia did not comment.