King's and Cambridge to build NHS cancer AI on the new Zenith supercomputer

The AI Centre for Value Based Healthcare, led by King’s College London and Guy’s and St Thomas’ NHS Foundation Trust, is partnering with the University of Cambridge to put the new Zenith supercomputer to work on NHS cancer care. Zenith, launched with a visit from a DSIT minister, is now described as the UK’s largest AI-for-science platform, and the collaboration also draws in Cambridge University Hospitals and other NHS trusts.

The plan is specific rather than a vague pledge to “use AI in the NHS”. Through a programme it calls MOSAIC, for Multimodal Oncology Sovereign AI Collaboration, the centre wants to build cancer foundation models, large models trained on data from several NHS trusts, and use Zenith’s compute to develop tools for earlier diagnosis, choosing the right treatment, and predicting patient outcomes. The machine learning here is trained on real clinical data, which is both the point and the hard part: it is what could make the models genuinely useful, and what makes governance, consent, and data protection central rather than an afterthought.

The recurring phrase is “sovereign AI asset”. The ambition is that the resulting models are owned and controlled within the NHS and UK research system, deployed across the health service and even licensed for new treatments, rather than rented from a foreign cloud provider. It is the same instinct now visible across UK AI policy, from sovereign compute to defence infrastructure, applied to health data, which is among the most sensitive and valuable a country holds.

It complements the wider push we covered in NHS England’s AI rollout. That story was about deploying AI tools into frontline care at scale; this one is further upstream, about building the models and the computing muscle to create them in the first place. Whether the sovereign ambition holds, or the economics eventually pull the work back toward the big commercial clouds, is the question worth watching.