It has become remarkably difficult to discuss the NHS Federated Data Platform without discussing Palantir. For many critics, the two have become almost interchangeable. There are understandable reasons for that. Palantir leads the consortium supplying the platform, the contract is worth up to £330 million over its maximum term, and the technology sits close to some of the most sensitive information held by the state. Scrutiny is inevitable and appropriate. But concentrating too heavily on the identity of the supplier risks missing another question that may ultimately prove more important: what would the NHS retain if Palantir were no longer involved?

There is no simple answer yet, and that is partly the point. Technology suppliers are temporary, even when their contracts are large. National infrastructure cannot be treated in quite the same way. The long-term credibility of the FDP will depend on whether the NHS retains sufficient control over the data foundations, models, products and institutional knowledge being created through it, rather than becoming dependent on whichever company provides the underlying technology at a particular moment.

That distinction is particularly relevant as the current contract approaches a decision point. The initial three-year term runs to 2027, with extension provisions thereafter. The Commons Science, Innovation and Technology Committee has recommended using the 2027 break clause and seeking an alternative supplier, while the Government has not accepted that recommendation and has said the contract will be considered through normal contract management processes. Whatever happens, the possibility of a future change provides a useful test of the claims being made about the FDP's architecture.

What does the NHS actually own?

The detail here matters. Palantir retains the intellectual property in Foundry, its existing software platform on which the FDP is based. NHS England does not own Palantir's underlying technology, and claims about NHS ownership should not blur that distinction. NHS England does, however, state that intellectual property created specifically for the NHS under the programme is owned by the NHS. This includes the Canonical Data Model and NHS-funded products and components developed through the FDP. Patient data also remains under NHS control and governance, with Palantir operating as a processor rather than acquiring rights to use NHS patient data for its own purposes.

On paper, that provides an important degree of protection. In practice, ownership of intellectual property and genuine technological independence are not quite the same thing. A public body can own its data and commissioned products while still becoming heavily dependent on the software, technical expertise and implementation knowledge of a particular supplier. Anyone assessing the FDP seriously should recognise that distinction.

Public-sector technology has repeatedly encountered this problem. An organisation procures a system, embeds it deeply into operations and discovers years later that changing supplier is technically possible but operationally expensive. Knowledge accumulates around the incumbent platform. Integrations multiply. Staff become accustomed to particular tools. The theoretical freedom to leave can become less meaningful with every year that passes.

The FDP should therefore be judged against a demanding standard. Its lasting assets extend beyond software licences and contractual ownership. They include common structures for organising NHS data, operational products developed around real problems, workflows tested inside hospitals, integration experience and a growing understanding of how fragmented information can support decisions. If those capabilities genuinely remain usable outside the current technology environment, the NHS will have created something with lasting strategic value.

NHS England says the FDP contract includes measures intended to reduce vendor lock-in, including exit planning, migration requirements and technical documentation designed to support the transfer of services. These provisions matter, but contractual portability is easier to demonstrate on paper than during an actual migration. Until the NHS has tested that portability at meaningful scale, some caution is warranted.

Changing the underlying provider could involve substantial work. Data pipelines might require adaptation, integrations would need testing, staff could require retraining and some functionality might have to be rebuilt. There would be cost and operational risk. The crucial question is not whether switching would be seamless. Few serious technology transitions are. It is whether the NHS could make such a change without losing the underlying capability and knowledge accumulated through the FDP.

 The NHS should be the constant

This is where the FDP debate connects to the larger question of NHS digital sovereignty. The term can easily be overstated. Digital sovereignty does not require the NHS to develop every piece of technology internally, nor would that be desirable. The health service needs specialist technology companies and will continue to rely on private-sector expertise.

The more useful definition is the ability to make meaningful choices about technology without being trapped by decisions made years earlier. The NHS should be able to continue with Palantir if its technology, performance and value justify doing so. It should also be able to choose another provider if the market develops in a different direction. Technology will change considerably over the next decade, and no procurement decision taken today can identify with confidence which company will offer the strongest platform in 2031 or 2036.

That makes architecture important, but architecture alone will not guarantee independence. The NHS also needs sufficient internal technical knowledge, strong commercial management and a clear understanding of which components it must retain if suppliers change. Without those capabilities, nominal ownership of data models or commissioned intellectual property may offer less protection than it appears.

There is nevertheless a significant opportunity here. NHS technology has too often been purchased as a succession of systems addressing individual problems. The FDP offers the possibility of building reusable data and operational capability beneath those systems, allowing some technologies to change while the NHS retains more of what it has learnt and created. Whether it ultimately succeeds in doing so remains something to prove.

The case will also stand or fall on outcomes. NHS England has reported operational benefits from FDP-supported capabilities, including more patients receiving procedures, support across cancer pathways and reductions in delayed discharge among patients with longer hospital stays. Those results are encouraging, but they deserve continued examination as adoption expands. The programme is sufficiently large, expensive and strategically important that evidence of benefit needs to become progressively harder, more comparable and more transparent.

The same standard should apply to security, procurement and portability. Supporters of the FDP should not dismiss legitimate concerns about supplier dependence simply because contractual safeguards exist. Critics, equally, should be careful not to treat the involvement of a controversial supplier as proof that the underlying strategy is flawed.

The more revealing assessment will come later. If, five or ten years from now, the NHS possesses the data foundations, products, institutional knowledge and operational capability created during this period, while retaining a credible ability to adopt different technologies, the FDP will have achieved something important. If changing supplier proves prohibitively expensive or requires large parts of the capability to be reconstructed, the claims of digital sovereignty will look considerably weaker.

That is a more useful test than whether one approves of Palantir. The NHS does not need to own every piece of technology it uses, and attempting to do so would probably constrain innovation rather than protect it. But it does need enough control, knowledge and negotiating power to ensure that today's technology choices do not quietly become tomorrow's permanent dependencies.