NVIDIA announced a five-year, $250 million compute commitment to support UK research into agentic AI safety, security and robotics.
NVIDIA announced October 13 that it will provide $250 million worth of computing access over five years to support agentic AI safety and security research in the United Kingdom. The commitment amounts to $50 million in compute access annually, with infrastructure hosted in the UK and delivered through UK-based NVIDIA Cloud Partners.
Most of the capacity is intended for the wider UK research community. A portion will support work involving the British Open-Ended Learning and Discovery Lab (BOLD), the Science of Fundamental AI Research Lab (SOFAIR) and Imperial College London. NVIDIA says research areas will include secure agents, open-model security, threat mitigation, safety assessment and robotics. The company also points to work on secure sandboxes, reproducible research and benchmarks for assessing safety.
The announcement describes compute access rather than a direct cash grant. NVIDIA says provider selection, technical architecture and commercial terms remain to be agreed. The company says locating the infrastructure in the UK is intended to support domestic research and the country’s compute ecosystem.
The initiative brings together research efforts with different emphases. BOLD, led by the University of Oxford, is developing approaches for AI systems that learn and adapt alongside people. SOFAIR, led by University College London, is exploring open-source architectures and more accessible development and deployment. Imperial’s work includes safety engineering and robotics.
NVIDIA said the program will use its OpenShell secure runtime software and Sentry reference design in some research focused on safeguards. Findings are intended to inform protections across the wider computing ecosystem. The company framed the commitment as building on previous UK research collaborations, including the University of Bristol’s Isambard-AI supercomputer. The announcement does not specify which researchers will receive access first, how applications will be selected or when the compute program will begin operating.