Nvidia, the US chipmaker valued at about $5.5tn (£4.1tn), has pledged $250m worth of computing power to the United Kingdom to bolster research into artificial intelligence safety. The gift, which is spread over five years, is intended to help researchers test and evaluate high‑performance AI systems.
The donation coincides with a UK government announcement that its newly confirmed AI Security Institute (AISI) will have the capability to scrutinise advanced AI models developed by firms such as OpenAI, Anthropic and Google before those models are made publicly available. Officials say the institute will play a central role in assessing risks posed by large, powerful AI agents.
Concerns about the behaviour and potential harms of increasingly capable AI systems have intensified across governments, industry and academia. Access to large amounts of compute is a key bottleneck for independent safety research, and Nvidia’s contribution is aimed at expanding the technical resources available to UK teams working on model evaluation and mitigation techniques.
Nvidia’s pledge is an example of private sector infrastructure being made available to support public interest research into technology risks. The company has grown rapidly in value in recent years as demand for chips used to train and run AI models has surged, making its hardware a critical component in both commercial AI development and safety testing.
UK authorities and researchers will now need to coordinate how the computing power is allocated and used in practice, balancing rapid assessment of emerging systems with oversight and transparency. The move adds to a broader global conversation about how best to combine industry capacity and public safeguards as AI systems become more capable.
As governments and technology firms navigate the trade-offs between innovation and risk management, donations of compute and other forms of collaboration are likely to be part of wider efforts to ensure advanced AI systems are developed and deployed safely.