The FDP Break Clause Is Dead. The Government Just Hasn't Said So Yet.

Read between the lines: the break clause is dead

On 16 June, the Health and Social Care Committee held a one-off evidence session on the NHS Federated Data Platform. Two panels of witnesses, an hour of questioning, and a minister who arrived between votes. The coverage that followed focused on the minister's apology for a DPIA error, the contract timeline, and whether the government would exercise the break clause. All of this missed the most consequential thing said in the session.

Rob Thompson, the Chief Digital, Data and Technology Officer for DHSC and NHS England, told the committee that if the government chose to exit the Palantir contract and reprocure from a different supplier, the process could take "anywhere between one and three years depending upon the complexity, size and scale of what you're doing." He then said something more striking still. Asked what the consequences would be, he said the NHS would be "arresting the benefits from the programme as it currently stands in favour of a transition." The chair asked him to clarify. He confirmed: development on the platform would have to stop to give a replacement time to catch up.

This admission has sat in the public domain since 16 June, and no one has reported it.

Now consider what it means. The Science, Innovation and Technology (SIT) Committee has recommended the government exercise the break clause and publish a costed exit plan by the end of 2026, thereby suggesting the government takes six months to create the plan and a full three months to execute something that has taken NHSE 4 years to get half way through, mid way through its abolishment as an organisation. The Lib Dems have called for a sovereign replacement - not a bad idea in itself were there a viable alternative technology. The BMA, Medact, Foxglove and others have campaigned for Palantir's removal without any sort of plan for what comes next.

All of these positions assume that the break clause is a viable decision with manageable consequences. Thompson told the committee, on the record, that exercising it means a multi-year procurement during which 139 Trusts currently using operational products on their wards would see those products frozen. To be clear, that means no new features, no new Trusts onboarding, No iteration on the cancer, discharge, or theatre scheduling products that clinicians in those Trusts are using today. The impact on patients would be severe, at a time when the headlines are dominated by the Ockenden review which stated clearly the impact of poor information systems on patients.

We already have a live example of what this looks like in practice. Yesterday, Sadiq Khan quietly reversed his decision to block a £50m Palantir contract with the Metropolitan Police. Having initially intervened on procurement grounds and faced a legal challenge from Palantir in response, Khan has now approved a 12-month extension so the Met can retain existing capability while it runs an open procurement for a long-term replacement. The pattern is instructive: a political decision to exit, followed by the operational reality that you cannot simply switch off a platform people depend on, followed by an extension to buy time for a procurement that has not yet started. It will be interesting to see what the procurement outcome there will be given the clear lack of alternative. They are in the same boat as the NHS: keen to make a step change in the capability of their information systems and being held back on principle.

Now apply that pattern to FDP at 50 times the scale. The government declares its intention to retender. Development freezes across 139 Trusts. A procurement process begins, lasting one to three years based on Thompson's own estimate, and longer than that if you account for the 15 months the original FDP procurement took. Longer still when you realise the organisation will be half the size it was when FDP was envisaged. During the procurement, the incumbent continues to run the frozen products because there is no one else to run them, exactly as Palantir will continue running the Met's systems during Khan's 12-month extension. At the end of the process, the winning bidder has to demonstrate that its platform can replicate or exceed the capabilities of a system that has been in production across the NHS for four years, including the ontology layer, the canonical data model integrations, and the operational products built on top of both. The incumbent, still running the frozen system, is the obvious front-runner in any retender because it is the only organisation with a production-ready platform. The challenger has a proposal. Palantir has a live service.

The more realistic path towards a post-Palantir future is not the break clause. It is a deliberate, government-supported programme to develop sovereign platform capability targeting 2030, giving UK technology companies and the NHS's own engineering talent the time and investment to build a credible alternative with in-production evidence before the contract's final expiry, rather than scrambling to replace an operational system overnight. That is a serious industrial strategy conversation. The break clause debate is not.

The committee never got to the right question

The reason Thompson's admission landed without scrutiny is that the committee spent almost the entire session on procurement mechanics, supplier reputation, and data governance. Whilst these might be legitimate areas of concern, they are certainly the wrong frame for understanding what FDP does and what exercising the break clause would actually cost.

The session opened with three expert witnesses: Professor Nigel Shadbolt from Oxford, Sarah Scobie from the Nuffield Trust, and Matt Hennessy, the Chief Data and Analytics Officer for NHS Greater Manchester, representing the CDAO network. With no FDP experience between them, they framed FDP as a data analytics and integration platform: a way of querying across silos, cleaning waiting lists, and producing population health dashboards. Shadbolt described it as an overlay that "allows you to pull on the different data" so that "you don't see the silos, you see a joined up response." Hennessy called it "an off the shelf all in one, jack of all trades, kind of master of none." One wonders why they were called as witnesses, and through what process. I note the presence of several anti-FDP campaigners on the panel itself including a guest: Martin Wrigley, the MP who convened the highly partisan April Westminster Hall debate on FDP and has led the parliamentary campaign against the Palantir contract.

Theirs is a fundamental mischaracterisation. FDP is an operational platform. It hosts products that clinicians use on wards to manage discharges, schedule theatres, coordinate cancer pathways, and track patients through urgent care. These products create new data through the actions clinicians take on them. They replace the spreadsheets, whiteboards, and Word documents that frontline teams have relied on for operational coordination for years, the shadow IT that no EPR was designed to handle and no information governance framework currently covers. The five-part series I wrote for Computer Weekly earlier this year explains this architecture in detail and why it represents a structural change in how clinical work gets done.

It is worth noting who was and was not in the room. Hennessy's ICB is one of only two in England that has rejected FDP. His evidence was drawn from a system that chose not to adopt the products he was being asked to evaluate. Greater Manchester's Trusts do use FDP at Trust level, but Hennessy was not speaking from that experience. He was speaking as an ICB data leader with a mature local analytics platform, which is a legitimate perspective but not one that equips you to assess the operational value of products you have never used. Martin Wrigley, who led much of the questioning on data governance and sovereign alternatives, sits on the SIT Committee rather than the Health Committee and was attending as a guest, alongside other MPs who participated in the April Westminster Hall debate on FDP. No one from a Trust that is actively using FDP's operational products was invited to give evidence.

The minister, Preet Kaur Gill, tried repeatedly to correct the framing. She described clinicians having to reprioritise patients using spreadsheets. She talked about real-time visibility of bed capacity, discharge readiness, and referral status. She described what she had seen on wards at Chelsea and Westminster and the Royal Surrey. Each time, the committee cut her off. Paulette Hamilton told her, testily, "I'm not asking you what it is." The chair said the committee already understood the platform and wanted procedural answers about the contract. The expert witnesses had been given extensive time to characterise FDP as an analytics product. The minister was given almost none to explain why that characterisation was wrong. The tone from the committee questioning was viscerally hostile, laying bare the tension between those delivering and those campaigning.

The consequence is that the committee's understanding of FDP, and therefore the basis on which it will scrutinise the contract decision, is built on a faulty premise. If FDP were an analytics platform, you could pause it, swap the supplier, and resume with a different product in a year or two. The disruption would be to dashboards and reports. Because FDP is an operational platform with products embedded in clinical workflows across 139 Trusts, pausing it means pausing the operational infrastructure that has replaced whatever those Trusts were using before. In many cases, there is nothing to go back to other than yet more risky spreadsheets and whiteboards. Thompson's "arresting the benefits" is a polite way of describing what happens when you freeze a live operational system that clinicians depend on every day.

NHSE is answering the wrong question

While the public debate has moved decisively to benefits, NHSE's communication about FDP has moved in the opposite direction.

The minister released a video explaining FDP to the public. In it, she describes the platform as helping "NHS staff spot problems sooner and fix them faster." That could describe any data product in any sector. It says nothing about what FDP specifically provides, what clinical workflows it supports, what products Trusts are using, or why the architecture underneath is different from what came before. The minister clearly understands these things. She tried to explain them to the committee and was shut down. The video retreats from that specificity into generalities that persuade no one.

NHSE's own contribution was an animation released for National Patient Data Day, focused entirely on information governance compliance: data stays in the UK, access is role-based, activity is logged, suppliers only access data when instructed by the NHS. These statements are accurate, and set a clear boundary against the recent reporting. Unfortunately for NHSE, the public debate is about whether FDP delivers value and they responded with a video about security controls. The people who distrust Palantir will not be reassured by an animation telling them the data is safe. The people who want to understand FDP's benefits will find nothing here to work with. As a piece of strategic communication, their silence concedes the benefits argument entirely by choosing not to engage with it.

Meanwhile, the evidence base for those benefits is crumbling under scrutiny. Laura Hughes at the Financial Times has driven this story hardest: her reporting revealed that Chelsea and Westminster accounts for 84% of the outpatient waiting list reductions NHSE cites across 16 Trusts (she omits to mention this product has been deprecated and is not being rolled out further), and that roughly a third of Trusts using the Inpatient CCS product performed fewer operations after adoption than before. Just days later she exposed NHSE's qualification of the FDP's benefit figures.

NHSE's response has been to retreat. Its methodology page now carries the caveat that it cannot "draw conclusions about cause and effect as other variables have not been controlled for," a statement that appears seven times in the updated document and was absent from the archived May version. The 110,000 additional operations figure that the minister cited to the committee, and that Palantir's UK leadership repeats in every media appearance, now carries a health warning from the very organisation publishing it.

The DPIA follows a similar pattern of forced concession. The minister apologised to the committee for NHS England's error in describing who has access to patient data within the platform's National Data Integration Tenant. As HSJ's Ben Clover observed, his surprise was not that the DPIA had been corrected but that the apology was for the documentation rather than for the underlying decision to grant contractors access to identifiable patient data in the first place. With NHS England's workforce cut in half and its data engineering capacity shrinking, the reliance on external contractors to build and maintain FDP is growing. That is a defensible position if you are willing to make the case for it. Staying silent does not prevent scrutiny, it just means the scrutiny goes unanswered.

Against this backdrop, the programme's own head of demand and delivery, Nirav Patel, said in the June FDP Bulletin that FDP is "here to stay." But asserting permanence while walking back both the evidence base and the governance framework is not a tenable position. If FDP is here to stay, the programme needs to demonstrate why, with evidence at the level of specificity that the current debate demands.

NHSE has that evidence. It is just choosing not to use it. The people attending the build-with-FDP sessions, Trust heads of data and heads of digital transformation who are doing operational work with these products every day, have concrete examples of what has changed on their wards and in their pathways. The data engineering community calls where Trusts share what they have built, the problems they have solved, the workflows they have replaced, contain more persuasive evidence for FDP's value than anything on the NHSE website. That is the story that would shift the debate. Instead NHSE publishes aggregate numbers with flawed methodology and generic reassurance about data security. The result is that the programme's opponents set the terms of the debate and NHSE plays defence on terrain chosen by its critics.

What should happen next

The committee should invite witnesses who can speak to FDP at the operational level: Trust CIOs and CDIOs who are implementing it, clinicians who use the products daily, and engineers who understand the architecture. The session on 16 June heard from an academic, a think tank researcher, and an ICB data leader whose own system chose not to adopt the Trust-level products. That left the committee without the evidence it needs to assess what exercising the break clause would cost in practice, which is the question Thompson's admission should have prompted.

The benefits methodology needs independent review, as Coleman has called for. But the metrics also need to change. Counting additional procedures performed is the wrong measure for a platform whose primary value lies in reducing clinical and IG risk, increasing operational efficiency, and reducing the burden of admin on overstretched clinical staff. FDP's contribution at Trust level is in replacing ungoverned clinical coordination with governed, auditable, shareable products. That does not show up in a before-and-after count of operations. The people best placed to articulate what does show up are the Trusts that have gone furthest with adoption, and neither the committee or the public has heard from any of them.

The conversation about sovereign alternatives needs to get serious. The break clause is a blunt instrument wielded by people who have not grappled with the operational consequences. A credible path away from Palantir, if that is what the government wants, requires a funded programme with a realistic timeline, likely targeting the contract's final expiry around 2030, giving UK technology companies and NHS engineering teams the time and investment to build platform capability that can genuinely compete with what Foundry provides. That is an industrial strategy question, not a procurement question, and it deserves to be treated as one.