On a Monday morning in an acute trust operations room, the meeting opens with a screen. Theatre utilisation, discharge delays, the longest waiters by specialty. Few people in the room know or care which vendor sits behind the dashboard. It tells them which lists to chase and which wards are stuck. That ordinariness is what makes the government's reported plan to retender the Federated Data Platform so consequential. A contract that began as a procurement decision has become operational plumbing, and plumbing is hard to replace while the building is occupied.
The reported facts are these. Ministers are understood to be preparing a competitive tender for the platform, with the existing arrangement extended by two years to avoid disruption. The £182m contract has a break clause in February 2027 and options that could take it to seven years and £330m in total. An NHS review earlier this year concluded that replacing the supplier in March 2027 would be impractical. The original tender took more than nine months, and a new exercise is expected to invite expressions of interest in January. Officials have reportedly asked a handful of British firms whether they could run a platform of this scale. The health secretary has told MPs that sovereignty matters for NHS technology while declining to discuss the contract.
The retender sits comfortably inside a wider government programme. Procurement rules are being revised under the digital technologies sector plan to give weight to commitments on British jobs, and the NHS is the largest single buyer of health technology in the country. Using that buying power to build a domestic industry is a coherent policy. It would be a mistake to assume, though, that the word sovereignty means one thing. It can describe the nationality of the supplier, the physical location of the data, the legal jurisdiction that governs access, the ownership of the code and configurations, or the ability of the state to walk away. A British prime contractor can still run on foreign cloud infrastructure and assemble foreign software components. A foreign supplier can, in principle, be bound by contract terms that leave the NHS in full control. The label tells a minister something useful for a speech and a procurement team very little.
The meaning that matters most in practice is the last one. The NHS has discovered that a seven-year contract, sold as flexible, cannot be exited on its first break date. Trusts have built workflows, reports and operational habits on the platform, and a source close to the process put it plainly: you cannot drop the incumbent overnight. Whoever wins the next contract will inherit the same dynamic unless the NHS insists on terms that prevent it. Those terms are unglamorous. They include documented data models, open interfaces, portability of the configurations that trusts have built, and a funded migration path written into the contract at the outset. A supplier of any nationality that refuses those terms is selling dependency.
Trust is the second constraint, and it is the one that procurement cannot easily fix. More than 3.5 million patients had registered the national data opt-out by October 2026, which keeps their data out of uses beyond direct care. Opposition from doctors, patients and MPs has centred on the incumbent's military and surveillance work. The national data guardian has now written to the health innovation minister advising that public and professional trust should carry substantial weight when any supplier is evaluated, and warning that without it the system is unlikely to be adopted widely enough to deliver its promise. That warning deserves to be read literally. Opt-outs do not affect individual care, but they thin the datasets that planning, research and population health analysis depend on. A platform built to join up information across the NHS loses value with every patient who withdraws from it.
Swapping the supplier will not on its own restore that trust. Patients tend to ask who can see their data, for what purpose, and who checks. A British registered address does not answer those questions. Published access controls, independent audit, clear limits on secondary use and a visible route for redress do. The retender gives ministers a chance to make trust a scored criterion with real weight, and to show their working when bids are assessed. If it is handled as a quiet exercise in supplier substitution, the opt-out figure will keep climbing.
Timing adds a further complication. Expressions of interest open in January and a competitive process of nine months or more points to an award late in 2027. The three-year independent evaluation of the current platform, commissioned from Imperial College, is due to report in 2029. The government will therefore probably choose a successor before it has independent evidence on what the existing platform has delivered for patients and for productivity. That is an uncomfortable position for a state that wants to demonstrate value for money while waiting lists remain high and trust finances are tight. Trusts have little spare capacity to absorb a migration, and the clinical risk of disrupting live waiting list management is real.
The practical implications follow from this. NHS leaders should press now for clarity on who carries migration costs and clinical safety assurance, so that local teams are not left to fund a national decision from their own budgets. Policymakers should define sovereignty in contract language, publish the criteria and weightings, and treat the data guardian's advice as a requirement rather than a courtesy. British firms approached about the work face a demanding test. The NHS data estate is fragmented and the operational stakes are high, so a credible challenger will need a delivery record at scale, and a domestic origin will count for little without one. Speculation about preferred bidders is premature, and the incumbent remains free to rebid with the considerable advantage of already being embedded.
The retender will ultimately be judged by the exit terms of the contract that follows it. A supplier the NHS could replace within twelve months, with its data, its configurations and its public trust intact, would be sovereign in the sense that matters, whatever the address on its letterhead.
