Nvidia Commits $250M of Compute to UK AI Safety Research
American science secured four times as much funding just five days earlier, while the UK commitment remains uncontracted.
Nvidia has pledged $250 million over five years for UK AI safety and security research, offering access to its compute resources rather than direct cash. The provider selection and commercial terms are still under negotiation.
This amounts to $50 million in compute power annually, delivered through UK-based Nvidia Cloud Partners with infrastructure hosted domestically.
While most of the compute will benefit the wider UK safety community, specific allocations will go to:
- BOLD at Oxford
- SOFAIR at UCL
- Imperial College London, focusing on safety engineering and robotics.
In contrast, American science received $1 billion from Nvidia over five years, distributed among universities, quantum computing, and government cloud providers.
Jensen Huang, Nvidia's CEO, emphasized the need for capability and safety to advance together, praising the UK's world-class scientific rigor in this essential engineering challenge.
This announcement follows a week of European investments, with London and Berlin expected to follow. However, it's unclear if this $250 million contributes to the promised $2 billion.
AMD had previously committed up to £2 billion over five years, including direct investments in British startups, while Nvidia offered nothing at the time.
Nvidia's funding will support work on secure agents, open-model security, and threat mitigation, utilizing its OpenShell runtime and Sentry reference design.
This initiative builds on existing progress, such as the Isambard-AI supercomputer in Bristol, which already runs on Nvidia Grace Hopper chips, and aims to inform safeguards across the broader computing ecosystem.
Cristian Dina, CRO at The Next Web, contributed to this report. He has interviewed over 300 industry leaders and authored the book King of Networking, establishing himself as a connected and respected voice in the tech scene.