Start a trial

Operational notes Observatory

Palantir vs London: when the supplier becomes “a point of weakness”

4 min read

Period engraving of the Palace of Westminster seen from the Thames
Once a supplier becomes embedded in public infrastructure, leaving it is a political decision.

A stand-off is under way in London that anyone working on public-sector technology should be following line by line. On one side is Palantir Technologies, the American data-analytics giant serving governments and defence; on the other, Britain’s institutions, which in recent months have begun asking out loud what the efficiency the company sells actually costs — in terms of dependency. In July the tension turned into a formal legal dispute: Palantir notified the Mayor’s Office for Policing and Crime of its intention to mount a legal challenge against the blocking of its £50 million contract with the Metropolitan Police, while the government launched a review of the £330 million NHS contract, weighing up the 2027 exit clause. Here we set out the facts — verified — and then the lesson, which concerns anyone who buys critical technology.

The facts, in order

  • The Met Police contract: two years, £50 million, AI systems for automating and analysing evidence in investigations. Mayor Sadiq Khan’s office refused approval, arguing that the police had not run an open tender. Palantir maintains that the block stems from “perceived values” rather than procedural grounds, and has launched legal action.
  • The parliamentary front: a committee described British public infrastructure’s dependency on the company as “an unacceptable point of weakness”, noting “a clear misalignment with UK values”, and recommended that the government exercise the termination clause on the healthcare contract. Palantir’s UK chief executive called the recommendation “irresponsible”.
  • The NHS review: ministers have opened a full-scale review of the Federated Data Platform — the data platform linking England’s hospitals — with the 2027 exit window on the table.

No court has yet ruled on anything; the company stands by the quality of its work and has its own arguments. But the point, for anyone watching, is not who wins: it is that the debate has got this far.

Lesson one: lock-in only becomes visible when you try to leave

The UK is not debating whether the platform works — it is discovering how hard it would be to do without it. When a country’s health data lives inside a single foreign supplier’s platform, the exit clause exists on paper, but exercising it means migrating years of integrations, training and processes. It is the same pattern we saw in the tug-of-war between the Pentagon and its AI supplier: a buyer’s real contractual power is measured at the moment it wants to change course, not when it signs. The question to ask beforehand — of every supplier, ourselves included — remains the same: how much would it cost to replace you? If the answer is not written down and verifiable, the true price of the contract is not the one on the invoice.

Lesson two: “values” have become a procurement criterion

Whether or not one agrees with London’s decision, the new development is that a technology contract has been blocked — and contested in court — on the grounds of alignment of values and governance, not price and performance alone. It is the direction the whole of European public procurement is taking: the EU AI Act requires demonstrable human oversight, Italian Law 132/2025 places responsibility on the individual official, and tender specifications are beginning to demand algorithmic transparency and reversibility. Anyone selling technology to the public sector — in Italy as elsewhere — would do well to arrive at those questions with answers already in writing: where the data sits, who can see it, how you get out.

Lesson three: sovereignty is not a slogan, it is an exit clause that actually works

The British paradox is instructive: immediate efficiency has built, contract after contract, a dependency that Parliament now calls a strategic weakness. The alternative is not giving up the best tools — it is insisting on an architecture in which the data and the operating model remain with whoever owns them, and the supplier is a replaceable component: this is the definition of sovereignty we use, and it holds for a government ministry just as much as for a manufacturing company. Genuine break clauses, data exportable in open formats, complete documentation: these are negotiated at signature, or not at all.

What to do if you buy critical technology

  1. Map your current dependencies: which processes would grind to a halt if supplier X disappeared or tripled its prices?
  2. Test the exit clause on paper: does it exist? Does it cover data, formats, migration timescales and costs?
  3. Demand reversibility in the tender specification, before signing: full export, documentation, hand-over support on the way out.
  4. Diversify where the risk is systemic: a single supplier for critical infrastructure is a point of weakness, whatever logo it carries.

Want to measure your dependency on technology suppliers — and what it would really cost to change? Half an hour with one of our experts for a first map.

Sources