Nvidia has announced a five-year compute donation valued at $250 million to support artificial intelligence safety and security research across the United Kingdom, according to a company statement published 14 October 2026.

The commitment provides access to accelerated computing rather than cash, directed at academic and other researchers working on technical safeguards for more capable AI systems. Nvidia frames the initiative as a way to expand the UK's research and compute capacity for agentic AI.

According to Nvidia's newsroom, the company plans to provide $50 million in compute access each year over five years. The majority is allocated to the wider UK AI safety and security research community, with a defined share earmarked for three named efforts: the British Open-Ended Learning and Discovery Lab (BOLD), the Science of Fundamental AI Research Lab (SOFAIR) and Imperial College London.NVIDIA Newsroom

The company characterises the initiative as focused on "agentic safety and security," a term it links to research on AI agents that can act, adapt and operate in open-ended environments. Compute resources under the program are to be delivered via UK-based Nvidia Cloud Partners, with infrastructure hosted in the UK. Nvidia states that this is intended both to support the domestic research ecosystem and to provide a "domestic platform for research," though it notes that provider selection, technical architecture and commercial terms "are to be agreed."NVIDIA Newsroom

Scope of the UK research support

Nvidia says the initiative will connect several strands of UK work in AI safety and security. BOLD, which it describes as led by the University of Oxford, is said to focus on approaches that help AI systems learn, adapt and operate alongside people. SOFAIR, described as led by University College London (UCL), is presented as exploring new open-source AI architectures and ways to make AI development and deployment more accessible. At Imperial College London, the work under this initiative is described as including safety engineering and robotics.NVIDIA Newsroom

Within the broader programme, Nvidia states that research will target "secure agents, open-model security and real-world threat mitigations." It says this includes support for secure sandboxes, reproducible research and safety assessment benchmarks, and it links this work to its own technologies, including the Nvidia OpenShell open secure runtime and the Nvidia Sentry reference design.NVIDIA Newsroom

On a separate product page, Nvidia describes OpenShell as "an open, secure runtime for agents" that governs agent execution and access and enforces policies outside the agent's own process.www.nvidia.com The company presents this as part of a broader "Open Agent Safety Platform" in which OpenShell provides sandboxing, policy enforcement, credential handling and audit logging, while Nvidia Sentry and Nvidia BlueField hardware are used for in-silicon security enforcement.www.nvidia.com

Nvidia's announcement says the findings from the UK research programmes supported by the compute donation are "intended to inform safeguards across the wider computing ecosystem," but it does not specify any publication requirements, governance arrangements for allocating compute, or oversight mechanisms for how projects will be selected or evaluated.NVIDIA Newsroom

Relationship with UK government and infrastructure

The statement includes supportive quotations from UK ministers but does not describe any formal government policy instrument linked to the donation. Nvidia quotes UK AI Minister Kanishka Narayan describing the announcement as "a vote of confidence" in the UK's strengths in AI research and frontier AI safety, and UK Business Secretary Jonathan Reynolds as saying that Nvidia's investment "will help British researchers continue to lead the world" in efforts to deploy AI safely.NVIDIA Newsroom

Nvidia positions the new initiative as building on existing collaborations with UK institutions, highlighting the University of Bristol's Isambard-AI supercomputer, which it says is powered by Nvidia Grace Hopper Superchips.NVIDIA Newsroom The press release does not detail how the new donation will interact with that system or with other national compute investments.

Under the programme, Nvidia says compute access will be funded through UK-based Nvidia Cloud Partners, with infrastructure hosted domestically.NVIDIA Newsroom It does not name the cloud partners involved and notes that technical architecture and commercial terms are still to be agreed. The company also does not set out eligibility criteria for the "wider U.K. AI safety and security research community" it plans to support.

Nvidia emphasises in its forward-looking statements section that expectations regarding the "value, funding, timing, delivery and use of compute contributions" under the UK AI safety and security initiative are subject to risks and uncertainties, including global economic conditions, technological developments, market acceptance of its products and changes in applicable laws and regulations.NVIDIA Newsroom

The company does not disclose in the announcement how it valued the $250 million figure for the compute donation and does not specify per-user or per-project limits on access.