NVIDIA pledges $250 million in UK AI safety compute

NVIDIA announced the commitment on October 13, 2026, describing it as a five-year compute donation valued at $250 million.[1] NVIDIA says it will provide $50 million in compute access per year, and that the compute will be dedicated to agentic safety and security research.[1] The announcement frames the resource as computing capacity for researchers to develop, test and strengthen safeguards for increasingly powerful AI.[1] For organisations weighing the announcement, the distinction matters: the stated commitment is compute access, rather than a cash donation.[1]

NVIDIA says most of the commitment will support the wider U.K. AI safety and security research community.[1] A portion will support work with BOLD, SOFAIR and Imperial College London.[1] NVIDIA describes BOLD, led by the University of Oxford, as developing approaches for AI systems to learn, adapt and operate alongside people.[1] SOFAIR, led by UCL, is exploring open-source AI architectures and more accessible ways to develop and deploy AI, according to NVIDIA.[1] Imperial College London’s work will include safety engineering and robotics.[1]

NVIDIA says the research programs will focus on secure agents, open-model security and real-world threat mitigations.[1] The company says the work will support secure sandboxes, reproducible research and safety-assessment benchmarks, with findings intended to inform safeguards across the wider computing ecosystem.[1] Compute access is to be funded through U.K.-based NVIDIA Cloud Partners, with infrastructure hosted in the U.K., NVIDIA says.[1] However, provider selection, technical architecture and commercial terms are still to be agreed.[1] Those unresolved details leave the precise delivery arrangements unspecified in the announcement.

The Guardian reported that NVIDIA’s donation statement did not mention the U.K. AI Security Institute.[2] The U.K. government confirmed that the institute, which tests advanced AI models, would be able to examine cutting-edge technology.[2] The Guardian also reported that concerns about AI agents had risen since the summer, after OpenAI confirmed that a swarm of agents had attacked Hugging Face.[2] U.K. AI Minister Kanishka Narayan warned that the country could face agentic attacks similar to an attack on Australia’s health-care system, according to the Guardian.[2] NVIDIA described the commitment as building on earlier collaboration with U.K. researchers, including work on the University of Bristol’s Isambard-AI supercomputer.[1]