NVIDIA said on 13 October 2026 that it will give UK researchers computing power that it values at $250 million over five years, to study how to make AI agents safe and secure. The money is not a cash grant. It is access to NVIDIA-powered computers, and the $250 million figure is NVIDIA’s own valuation of that access.
What NVIDIA is giving
In its announcement, NVIDIA describes the pledge as “a five-year compute donation valued at $250 million”, which works out at “$50 million in compute access annually”.
“Compute” is the processing power needed to train and test AI systems. It is expensive and hard to get, especially for university labs, so access to it can decide what research gets done at all.
NVIDIA says most of the computing time will go to “the wider U.K. AI safety and security research community”. A share will go to three named groups:
- the British Open-Ended Learning and Discovery Lab (BOLD) at the University of Oxford;
- the Science of Fundamental AI Research Lab (SOFAIR), which NVIDIA says is led by UCL;
- Imperial College London.
Many details are still open
The computing will be supplied through UK-based NVIDIA Cloud Partners, with the hardware hosted in the UK. But NVIDIA’s release is plain that the deal is not finished: “Provider selection, technical architecture and commercial terms are to be agreed.” In other words, which companies will run the computers, how the system will be set up and the commercial arrangements have not yet been decided.
What the research is about
An AI agent is a system that does not just answer questions but takes actions, such as running software, browsing or handling tasks for a person. That raises new safety questions, because a mistake or a hijacked agent can do real damage.
NVIDIA says the programme will focus on “secure agents, open-model security and real-world threat mitigations”. It mentions secure sandboxes (sealed-off test environments), reproducible research and safety benchmarks, including work using NVIDIA’s own OpenShell runtime software and its Sentry reference design for monitoring agents.
What people said
NVIDIA chief executive Jensen Huang said: “AI capability and safety must advance together.”
UK AI Minister Kanishka Narayan called the pledge “a vote of confidence in our world-leading strength and expertise in AI research”. Business Secretary Jonathan Reynolds said: “NVIDIA’s investment will help British researchers continue to lead the world in that effort.”
Jakob Foerster, founding director of BOLD, said: “We are experiencing a critical moment where AI safety as an afterthought is no longer appropriate for the open-ended, evolving world.”
What this means
- It is computing time, not cash. The headline figure is NVIDIA’s valuation of the computing access it will provide.
- The terms are not settled. Providers, technical set-up and commercial terms are still to be agreed.
- It also helps NVIDIA’s ecosystem. The work will run on NVIDIA hardware through NVIDIA’s cloud partners and will use NVIDIA’s own agent-safety software.
Sources
- NVIDIA Newsroom, “NVIDIA Commits $250 Million to Advance UK Agentic Safety and Security Research Over the Next Five Years”, 13 October 2026. https://nvidianews.nvidia.com/news/nvidia-commits-250-million-to-advance-uk-agentic-safety-and-security-research-over-the-next-five-years
Related stories
- UK AI Security Institute Sets Out Its Priorities for Technical Best Practice in Frontier AI Development
- Agents Hit Real People: Why the UK Paused — Then Restarted — Its Riskiest Cyber AI Tests
- Hundreds of Scientists, Including Geoffrey Hinton and Yoshua Bengio, Call for an International Body on AI Containment and Incident Reporting
