New to this? Read it in simple words
- Nvidia will give British researchers computing power worth $250 million over five years.
- That is $50 million a year, for research on keeping AI agents safe and secure.
- Labs at Oxford, UCL and Imperial College London get a share; most goes to other UK researchers.
- The value is Nvidia’s own estimate, and the terms are still to be agreed.
- AI agent
- An AI program that acts on its own, for example by clicking, typing or running code.
- Compute
- Computing power: time on the powerful computers that train and test AI models.
- Sandbox
- A closed space on a computer where a program can be tried without reaching anything else.
AI agents are programs that act on their own, for example on a computer or on the web. On October 14, Nvidia said it would give British researchers computing power to study how to keep them safe.
What Nvidia is giving
Nvidia calls it a “compute donation valued at $250 million”, spread over five years. That means computing power worth $50 million a year, not cash.
It is meant only for research on the safety and security of AI agents. Most of it goes to the wider UK research community. A share supports three labs: BOLD, led by the University of Oxford, SOFAIR, led by UCL, and Imperial College London.
The computers will run in the UK, through British cloud companies that work with Nvidia. Which companies, how it is set up and the commercial terms are “to be agreed”, Nvidia says.
Nvidia’s own figures; the $250 million is its estimate of the computing’s value, and the terms are still to be agreed.
What the research covers
Nvidia names secure agents, the security of open models and defences against real threats. The work will also build safety tests and secure sandboxes, the closed spaces where agents can be tried without harm.
“AI capability and safety must advance together,” Huang said. Kanishka Narayan, the UK’s AI minister, called the pledge “a vote of confidence” in British AI research.
Britain has no major AI developer of its own, Reuters notes, but wants a leading role in testing AI, mainly through its AI Security Institute.
Huang on AI incidents
“I’m willing to accept no incidents of cybersecurity. We should not accept any of them,” Huang told Fortune in London. He said the incidents so far could have been avoided with better containment and monitoring of AI agents.
Huang is on a European tour that started in Paris on October 13. He said Nvidia would announce $2 billion of new investments in Europe within a week, Fortune reports. Nvidia has not yet said what they are.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · October 13, 2026NVIDIA Commits $250 Million to Advance UK Agentic Safety and Security Research Over the Next Five Years Nvidia
- 2Research · October 14, 2026Nvidia pledges $250m in computing power to UK AI safety research Reuters (via The Daily Star)
- 3Research · October 14, 2026Nvidia’s Jensen Huang pledges $2 billion for Europe, $250 million in compute for UK safety labs The Decoder
- 4Research · October 14, 2026Nvidia CEO Huang says society shouldn’t tolerate AI cyber incidents, says AI doomers “ought to stop working on” the technology Fortune
The $250 million is Nvidia’s own value for the computing it will give, and the terms are still to be agreed. The $2 billion for Europe is a plan Huang described, not yet a list of deals.
Nvidia’s $250 million computing pledge boosts UK research on AI agent safety, though its terms are not yet agreed. How we rate the mood