Rented Sovereignty As A Monthly Data Subscription

In 2010, if you wanted to see the cost of the Private Finance Initiative, you could drive to a hospital. The building would be there, usually handsome, sometimes impressive, invariably more expensive to occupy than anyone had been told when the ribbon was cut. The contract was made of concrete and glass. You could kick it.

The modern equivalent is harder to find. It sits behind logins. It runs inside dashboards. It organises the way data flows between wards and waiting lists, between GP practices and operating theatres, between discharge coordinators and bed managers. It is not a building the state is locked inside. It is an operating layer the state can no longer see around.

Britain spent decades paying inflated rent on hospitals and schools it theoretically owned. It is now doing something structurally similar with software, data platforms and digital infrastructure.

The difference is visibility.

A bad PFI building at least provoked public outrage once the annual service charges were published. A bad software dependency may never be fully understood by the ministers who approved it, the officials who manage it, or the patients whose records travel through it.

What PFI Was, And Why It Matters Here

For readers unfamiliar with the abbreviation: PFI, or the Private Finance Initiative, was a John Major scheme under which private companies built public infrastructure (hospitals, schools, prisons, roads) and the state paid them back over decades through annual service charges.

The appeal was political: the upfront borrowing stayed off the government's balance sheet. The cost was economic: the state ended up paying far more than if it had borrowed and built directly.

PFI was formally retired in 2018, its brand too contaminated to save. But the underlying logic never disappeared. Use private capital. Avoid visible debt. Bind future taxpayers to long-term payments. It simply migrated into new forms. Regulated utility bills. Captive care markets. And, most invisibly of all, technology contracts.

The analogy is not perfect. Software procurement is not structured like a PFI hospital deal. But the dependency pattern is close enough to be dangerous: public need, weak state capacity, private delivery, long-term reliance, expensive exit, supplier knowledge advantage, and a political class unable to reverse course.

£330 Million, Seven Years, One Platform

The centrepiece is the NHS Federated Data Platform. In November 2023, NHS England awarded a contract to a consortium led by Palantir Technologies, the American surveillance firm co-founded by Peter Thiel to build a unified data system across up to 240 NHS organisations. The consortium includes Accenture, PwC, Carnall Farrar, and the North of England Care System Support. The contract runs for up to seven years, with a reported value of up to £330 million, and the initial three-year commitment comes up for review in early 2027.

Careful readers will remember this saga from a previous article:

You Cannot Govern Through Purchase Orders

A £1 pandemic contract became £330 million. A £75 million defence deal became £240 million. Palantir now holds at least 34 UK government contracts worth over £670 million. Most hospitals refused to use its platform. The MoD cannot switch providers without it.

The Restorationist®Oliver Kerr

The platform is built on Palantir's Foundry software, and its purpose is sensible enough: integrate the fragmented data systems across NHS trusts so hospitals can track bed availability, manage waiting lists, coordinate discharges and allocate theatre time more effectively.

NHS England says the system has supported over 110,000 additional operations since launch, reduced discharge delays by more than fifteen per cent, and increased the proportion of cancer patients diagnosed within 28 days of referral by nearly seven per cent.

Those are real numbers, and they deserve to be taken seriously. It's most likely, as Palantir are an exceptionally competent US company.

The NHS has appalling data fragmentation. Clinicians routinely work across incompatible systems. Trusts operate in information silos while patients fall through the gaps between them. Anyone who dismisses the case for a better data infrastructure is not engaging with the problem.

But urgency is precisely when dependency becomes most dangerous. A desperate institution buys badly. And the questions now being raised by Parliament, by doctors, by data-protection lawyers, by the government's own review, are not about whether the NHS needs better technology. It does.

The questions are about who controls it, on whose terms, and what happens when the contract ends.

Parliament: "Unacceptable Point of Weakness"

In June 2026, the partisan House of Commons Science, Innovation and Technology Committee published a report describing Palantir's growing presence across the UK public sector as an "unacceptable point of weakness." The committee urged the government to exercise the 2027 break clause in the NHS contract and either develop an in-house replacement or seek an alternative provider. Its chair, Dame Chi Onwurah, said Britain should be:

aiming for technology sovereignty in critical parts of our public sector.

Correct.

This was, however, to be expected, and it comes with the caveat its only a problem if the technology providers aren't highly progressive and left-wing like the culture of the civil service they are due to work with.

Palantir gets the left hopping mad like Palestine. If the answer to socialist NHS problems is not the victory of pure religious socialism over them, or perceived enemies of socialism are allowed into the stable to disrupt the process of socialism demonstrating its purity, the inmates go beserk. The controversy is now something to do with Palantir's relationship with US immigration enforcement.

The committee identified Palantir as the most concerning example of a wider pattern: the public sector's growing dependence on a small number of major foreign technology providers, including Microsoft and Amazon Web Services. It called for a strategy to end vendor lock-in, diversify suppliers, and strengthen digital resilience. Vendor lock-in, the committee argued, should not be treated as inevitable.

This was not a campaign leaflet. It was a cross-party parliamentary committee telling the government its own procurement had created a strategic vulnerability.

Separately, the government itself has been reviewing the contract. Reuters reported in June 2026 the review would consider patient confidentiality, trust and reliance on a US supplier. Health Minister Zubir Ahmed said ministers would decide later in 2026 whether to trigger the extension clause.

The British Medical Association, a left-wing trade union representing over 190,000 socialist doctors, passed a resolution in June 2025 calling Palantir an "unacceptable choice of partner" and urged members to limit usage of the platform.

Greater Manchester's Integrated Care Board refused to adopt the system entirely, saying its own analytics stack was more advanced. Several other trusts questioned whether the platform offers genuine improvements over existing tools.

An open letter from the NHS Chief Data and Analytical Officer Network (whatever that is supposed to be) noted many trusts:

already have similar tools in use presently exceeding the capability and application of what the FDP is currently trying to develop.

How Lock-In Dependency Actually Works

Vendor lock-in is widely misunderstood as a contractual clause. In practice, it is far deeper and more difficult to reverse. A platform like Foundry does not merely store data. It organises it. It creates the data model through which information is categorised, linked, and retrieved. Other systems begin to connect around it. Staff redesign their workflows to match the dashboards.

Training becomes vendor-specific. Management reporting becomes dependent on the platform's version of reality. Procurement decisions begin to assume compatibility with the existing system.

Over time, the vendor understands the operating environment better than the public body does. The institution's own knowledge of how it functions becomes mediated through a private company's architecture.

One can see why the left really don't like this. It looks a lot like institutional capture, which in turn looks a lot like religious conversion to another ideology as apostasy from socialism.

A vendor does not become irreplaceable when the contract says so, but when the institution forgets how to operate without it.

With PFI, the state was trapped inside a building. With platform procurement, it risks being trapped inside a way of seeing. The dashboards define what managers monitor. The data model defines what questions can be asked. The integration layer defines what other systems can connect.

This is not a passive utility. It is the camera through which an institution the absurd size of the NHS understands itself.

Rotting Foundations Under The AI State

The government is not buying platforms from a position of strength. It is buying them because its own systems are old, fragmented and chronically underfunded. Government assessments have found almost 80 critical IT systems classified at the highest risk level.

• The Public Accounts Committee's 2025 report on digital transformation noted piecemeal patching of legacy infrastructure rather than genuine modernisation.
• A 2025 NAO report on government cyber resilience found departments had multiple fundamental control failures. Legacy systems, some dating to the 1970s, underpin essential services including state pensions, border management and tax collection.
• A Re:State report published in May 2026 warned the legacy problem is getting worse, not better, with digital expenditure across government running thirty per cent below benchmarks in comparable countries.

The government's very silly 2025 AI Playbook sets out plans to embed artificial intelligence across public services. Good luck with that.

The PAC's AI report acknowledged the potential but warned government faces "significant challenges" in grasping the opportunity.

These are confessional euphemisms stating the obvious: our country is governed by Boomers who are clueless about technology.

The danger, however, is layered.

AI systems will not merely store data. They will help interpret, prioritise, classify and, eventually, recommend decisions. A platform dependency today may become a decision-making dependency tomorrow. The system will not simply tell a hospital how many beds are free. It may flag which patients should be seen first, which referrals are urgent, which discharge pathways should be followed. If the models, training data and evaluation criteria are proprietary, the institution cannot fully audit the logic behind its own operations.

Britain is attempting to build an AI-enabled state on top of a digital estate some of its own officials describe in terms normally reserved for condemned buildings.

Switzerland Said No

Not every country has responded to this dilemma the same way. In December 2025, a Swiss Army risk assessment concluded Palantir's platform posed unacceptable risks to national data sovereignty. The assessment was not based on technical performance: Swiss officials acknowledged the software's analytical power. It was based on architecture.

The core concern was jurisdiction: as a US-based company, Palantir operates under American law, including the CLOUD Act, which can compel US firms to hand data to American authorities regardless of where it is stored. The Swiss review found no technical or contractual mitigation sufficient to eliminate the risk.

Switzerland recommended alternatives. Palantir responded by suing the Swiss magazine which reported the story.

One country ran a formal sovereignty assessment and walked away. Britain awarded contracts worth over £670 million across at least ten government departments, including health, defence, policing, nuclear weapons management and financial regulation. A December 2025 Ministry of Defence contract alone was worth £240 million and was awarded without competitive tender.

The question is not whether foreign firms build good software. Many do. It is whether a state can still function if they stop. Or if diplomatic, legal or commercial conditions change. Or if US export controls, sanctions policy, or corporate strategy shift in ways no contract anticipated.

Anthropic discovered the same problem this week when the US government simply shut it down.

Revolving Doors And Institutional Capture

Matthew Swindells served as Deputy Chief Executive and Director of Operations at NHS England until July 2019. Within months, he joined Global Counsel (the lobbying firm co-founded by Peter Mandelson) and began advising Palantir on NHS strategy.

From April 2022, Swindells simultaneously held the role of joint chair of four major hospital trusts in north-west London, including Chelsea and Westminster, which served as a flagship site for Palantir's technology.

The Financial Times reported Swindells privately urged colleagues to add more patient data to a Palantir-built platform while in his NHS role. Chelsea and Westminster said it would exclude Swindells from "any decision making in relation to Palantir," though the full scope of influence is, by its nature, difficult to trace. Swindells is understood to have stepped down from his NHS position following the scrutiny.

An investigation by The Nerve, published in early 2026, identified at least 32 former officials, ministers, intelligence chiefs and peers who had taken roles connected to Palantir. Global Counsel and Palantir jointly hosted events promoting NHS data strategies.

The revolving door did not create the contract. But it created the conditions in which a company with deep intelligence-community roots could embed itself in the administrative machinery of British healthcare without sustained public debate.

What A Palantir Exit Would Actually Require

Suppose the government does trigger the 2027 break clause. What happens then? The platform is built on Palantir's proprietary Foundry software. The NHS does not own Foundry and cannot replicate it. The data itself remains NHS property under the contract terms, but the structure around the data, the models, workflows, integrations, dashboards and institutional habits, is shaped by the platform.

Eighty trusts were reportedly using the system by early 2026. Staff have been trained on it. Procurement assumptions have been built around it. NHS England paid KPMG £8 million simply to promote adoption of the platform across the service.

An exit is not impossible. But it would require transition planning, alternative systems, retraining, data migration and enough in-house capability to manage the process. It would, in short, require exactly the kind of internal capacity the NHS did not have when it bought the platform in the first place.

This is the PFI echo. The original PFI hospitals were difficult to exit because the state had no alternative delivery mechanism. The state had hollowed out its own construction and maintenance capacity, then found itself dependent on the private sector to fill the gap. Software procurement is following the same arc: capability is outsourced, internal knowledge erodes, and by the time the contract is up for renewal, walking away has become more expensive than staying.

No major public-sector software contract should be approved without an exit plan as detailed and as funded as the implementation plan. Open standards, data portability, interoperability requirements, in-house technical teams and a credible alternative pipeline are not extras. They are the minimum conditions for sovereign procurement.

Public Weakness As Private Revenue

PFI created private annuities from public buildings. Regulated Asset Base models create private returns from captive bill-payers. Care markets create private income from statutory duty. Software platforms create private dependence from administrative incapacity.

The mechanism is always the same: public weakness becomes private revenue.

The software version is more dangerous than its predecessors because the asset is not external to the state. A hospital PFI contract put the state inside an expensive building. A software dependency puts the state inside an expensive way of thinking. The platform does not sit alongside governance. It becomes part of how governance happens. How decisions are framed. How performance is measured. How risks are surfaced. How priorities are set.

The NHS needs a data spine. But a spine should not be rented. And the institution supplying it should not be one whose co-founder describes the National Health Service as suffering from "Stockholm syndrome," regardless of whether he is correct.

The old PFI made Britain pay rent on hospitals it thought it owned. The new version risks making Britain pay rent on its own capacity to see, to decide, and to govern.

Next in The New Rentier State: how Britain built a permanent contractor class no scandal can dislodge and no reform can reach.