Two letters reached Andy Burnham's desk within days of him entering Downing Street. One came from techUK and the Startup Coalition, warning that breaking up the Department for Science, Innovation and Technology would send a damaging signal to a sector expanding rapidly and depending on Whitehall for clarity. The other came from Article 19, Open Rights Group and the Balanced Economy Project, arguing that the new prime minister should use his first days in office to end the NHS's contract with Palantir and cut the health service's reliance on American technology firms. Neither letter mentioned the other directly. Read together, they describe the same problem from opposite sides: a machinery of government change that has become entangled, almost by accident, with the most contested data contract in the NHS.

DSIT houses the AI Security Institute, the Sovereign AI Fund, UK Research and Innovation and the Government Digital Service. Reports over the weekend suggested most of this would move into an enlarged Department for Business and Trade under Jonathan Reynolds, with other functions transferring to the Department for Culture, Media and Sport. None of these bodies sits inside the Department of Health and Social Care's org chart, yet all of them shape the conditions under which NHS trusts adopt artificial intelligence, procure data infrastructure and govern the tools now moving from pilot to deployment across acute and primary care. The Federated Data Platform, ambient voice technology in outpatient clinics, and the algorithms increasingly used in diagnostic triage all depend on a regulatory architecture that assumes DSIT continues to exist in something like its current form.

That assumption is now in question at precisely the wrong moment. Trusts have spent the past two years building the operational habits, procurement routes and clinical governance structures needed to work with the Federated Data Platform, much of it under sustained scrutiny over Palantir's role and the terms of data access it granted a single vendor. The digital rights groups writing to Burnham this week are not raising a new concern; they are asking whether the transition to a new government offers a chance to revisit an arrangement that has never fully settled. Whether or not that contract survives the reshuffle, the institutions responsible for adjudicating it are themselves about to be dissolved and rebuilt.

Liz Kendall, the science secretary expected to be replaced, said recently that the UK had roughly two years to establish a foothold in AI before the technology was, in her words, done to us. That timeline was already tight before anyone suggested dismantling the department she runs. According to industry estimates, the reorganisation itself will cost about six months of official attention. During this time, civil servants who are knowledgeable about data governance, AI safety, and health technology purchasing will be too busy creating new organisational charts to respond to the questions that NHS leaders are currently posing to them. For a health service trying to scale ambient scribing tools safely, resolve the governance gaps the Federated Data Platform exposed, and give integrated care boards a clear steer on what AI procurement should look like after consolidation, a prolonged institutional vacuum at the centre is not a minor inconvenience.

NHS leaders have grown accustomed to political churn at the top of government affecting the pace of reform, from the repeated restructuring of NHS England to the uncertain status of workforce plans drawn up under previous administrations. What is different here is that the churn is occurring one level removed from health policy altogether, inside a department most NHS executives rarely have cause to think about, yet whose survival or dissolution will determine how confidently trusts can proceed with the technology decisions already underway. Burnham's team may conclude that folding DSIT into a larger department improves coordination. They should also reckon with the fact that the NHS's most consequential and most contested digital transformation programme now depends on institutions that do not yet know what they will look like in six months.