NVIDIA pledges compute valued at $250m over five years to UK research on AI agent safety and security

NVIDIA has committed a "compute donation valued at $250 million over five years" to UK research on the safety and security of AI agents, the company said in a release dated 13 October. Two UK ministers, AI Minister Kanishka Narayan and Business Secretary Jonathan Reynolds, are quoted in the announcement.

What NVIDIA is offering

The donation is compute, not cash. According to NVIDIA's release:

• The commitment "will provide $50 million in compute access annually" for agentic safety and security research across the UK.
• "The majority will support the wider U.K. AI safety and security research community", with "a proportion" supporting work with three named groups (below).
• Access will be funded through UK-based NVIDIA Cloud Partners, "with infrastructure hosted in the U.K."
• "Provider selection, technical architecture and commercial terms are to be agreed."

The release does not say how researchers outside the three named groups will apply for or be allocated the compute.

Who is named

The release names three recipients and describes them as follows:

• BOLD (the British Open-Ended Learning and Discovery Lab), "led by the University of Oxford, is developing approaches that help AI systems learn, adapt and operate alongside people."
• SOFAIR (the Science of Fundamental AI Research Lab), "led by UCL, is exploring new open source AI architectures and more accessible ways to develop and deploy AI."
• Imperial College London: "Imperial’s work will include safety engineering and robotics."

Jakob Foerster, founding director of BOLD, said in the release that the compute "will help us explore transparent and human-centric approaches to AI". Alessandra Russo, director of Imperial Frontier AI Research Lab, said Imperial is working with NVIDIA and industry partners on safety and assurance in physical AI.

What the research will cover

NVIDIA says research programmes "will focus on secure agents, open-model security and real-world threat mitigations", supporting secure sandboxes, reproducible research and safety assessment benchmarks. NVIDIA says that includes research using its own OpenShell runtime software and its Sentry reference design. NVIDIA says the findings "are intended to inform safeguards across the wider computing ecosystem."

The company describes the commitment as building on its existing work with UK researchers, including the University of Bristol's Isambard-AI supercomputer, which NVIDIA says is powered by its Grace Hopper Superchips.

What the ministers said

Narayan called the commitment "a vote of confidence in our world-leading strength and expertise in AI research — and, in particular, frontier AI safety that is becoming ever more important as AI’s capabilities develop". He said building "a distinctly British AI future that is safe, secure and that serves people’s best interests" was among "objectives that will be central to the U.K.’s G20 presidency."

Reynolds said: "To realise the opportunities of AI, people need to feel the technology is safe and secure." He added that NVIDIA's investment "will help British researchers continue to lead the world in that effort".

NVIDIA's founder and chief executive, Jensen Huang, said that "AI capability and safety must advance together".

We could not find a matching government press release on gov.uk at the time of writing; the ministers' words above come from NVIDIA's release. Narayan set out the government's wider approach to AI and security in a speech on 12 October, covered in our piece UK will use frontier AI to find weaknesses in government systems before attackers do, AI minister says.

Why it matters

On our reading, the headline figure is a valuation of compute access rather than a grant, and two details that decide how useful it is to researchers, which cloud providers and on what commercial terms, are by NVIDIA's own account still to be agreed. The release also does not say how the "majority" share for the wider research community will be allocated. Part of the research is tied to NVIDIA's own OpenShell runtime software and Sentry reference design. For UK groups working on how AI agents can be tested and secured, it is a stated five-year source of computing capacity hosted in the UK.