Palantir in Europe: trapped by dependency
European governments want to break free from Palantir. But the U.S. firm's "land and expand" strategy has turned a commercial contract into a structural dependency.
A portrait of a $1.2 billion technological lock-in.
At a glance
France renewed its Palantir contract with its domestic intelligence service (DGSI) in late 2025 — because no sovereign alternative was ready to replace it.
Switzerland’s military concluded in December 2024 that data entrusted to Palantir could be accessed by U.S. intelligence services in ways that “cannot technically be prevented” — and recommended alternatives it has yet to identify.
The UK’s National Health Service has a break clause on its $440 million (£330 million) Palantir contract in March 2027 — but no competitor has demonstrated equivalent data-federation capability at that scale.
This image is used for illustrative purposes only.
The facts: a footprint that goes beyond the controversy
Founded in 2003 by venture capitalist Peter Thiel and CEO Alex Karp, Palantir Technologies has spent two decades building a unique position at the intersection of intelligence, defense and public health data. In 2025, the firm reported revenues of over $4.6 billion — a 56% increase year-on-year. Most of that growth is driven by government contracts.
In Europe, the map of public-sector engagement is denser than the headlines suggest. The United Kingdom holds the largest share: according to an investigation published in January 2026 by The Nerve, a UK investigative outlet led by journalist Carole Cadwalladr, identified contracts with the British state exceed $900 million (approximately £670 million at current exchange rates). These span the National Health Service ($440 million, or £330 million, over seven years), the Ministry of Defence — including a $320 million (£240 million) contract extension awarded in December 2025 without competitive tender, building on an existing strategic partnership — police forces, intelligence agencies, and a $20 million contract with AWE Nuclear Security Technologies (formerly the Atomic Weapons Establishment), the agency responsible for the UK’s nuclear warheads.
In France, the DGSI — France’s domestic intelligence service, roughly equivalent to the FBI — has used Palantir tools for several years. A written question submitted to the French Senate in May 2026 reveals that this contract was renewed for three years in late 2025, with the government acknowledging that “the replacement tool was not yet complete.” The Netherlands, Denmark, Germany, Poland, Norway and Spain are among the identified government clients. In March 2025, NATO acquired Palantir’s Maven Smart System for use within Allied Command Operations. The European Commission has not responded to a parliamentary question from MEP Tineke Strik (GroenLinks–PvdA, Netherlands), a member of the Committee on Civil Liberties, Justice and Home Affairs, on its own use of Palantir technology.
How the trap works: “land and expand”
Palantir’s commercial strategy is documented under the term “land and expand.” Entry often comes at marginal cost — an initial contract priced at a nominal £1 during the COVID-19 pandemic, or an emergency offer to Swiss health authorities citing the NHS as a reference. Once data is integrated into Palantir’s Foundry platform, migrating to an alternative requires a full IT overhaul that few public administrations have the technical or budgetary capacity to undertake.
A risk assessment completed by the Swiss Army in December 2024 documents this dynamic with unusual candor for an official document. Palantir relies on a proprietary system requiring the permanent presence of the firm’s own specialists to function — which, the report warns, could “limit the army’s ability to act in crisis situations.” The assessment also flags risks of wrongful targeting through statistical correlations, and concludes that data leaks “cannot technically be prevented.” Its recommendation: “The Swiss Army should consider alternatives to Palantir.” It does not identify which ones.
In the UK, a parliamentary report described Palantir’s programs as giving the national government an “unacceptable point of weakness.” The decision to extend the Ministry of Defence’s contract by $320 million without competitive tender — formalized alongside a strategic partnership announced during President Trump’s state visit in September 2025 — has raised questions about where commercial relationships end and diplomatic dependency begins.
What the data reveals: the asymmetry of exit
The difficulty of “exiting” Palantir is not rhetorical — it is architectural. Foundry functions as an integration layer that absorbs data flows from dozens, sometimes hundreds of organizations (240 NHS trusts in the British case). Replacing it requires not only finding equivalent software, but re-federating data sources that were never connected to each other except through Foundry.
The operational stakes cut both ways. The UK government and Palantir have cited benefits from the NHS deployment, including an estimated 110,000 additional operations unlocked by more efficient scheduling and bed management. NHS England’s own programme review rated the platform “green” — on track — as of July 2025. These reported gains are part of why the dependency is so difficult to unwind: the argument for staying is not merely commercial, but clinical.
France’s DGSI case illustrates this asymmetry: the domestic intelligence service has an operational tool that the government admits it cannot do without in the short term, while the sovereign replacement tool remains “still in development.” The three-year renewal in late 2025 pushes the question to 2028 at the earliest — by which point the dependency will be deeper still.
European financial institutions increased their Palantir holdings by around 70% in 2026 compared to 2024, according to data compiled in MEP Strik’s parliamentary inquiry. The total value of European institutional stakes in the firm is estimated at approximately $27 billion. Banks including Deutsche Bank, BNP Paribas and Barclays are among those with reported positions, according to investigative analysis by Follow the Money. This creates a structural European financial interest in maintaining the contracts — which complicates any political decision to exit.
The dimension no one is discussing: the CLOUD Act
The European controversy over Palantir has a legal dimension that public debate tends to understate: the CLOUD Act (Clarifying Lawful Overseas Use of Data Act), passed by the U.S. Congress in 2018, allows the American government to compel U.S. companies to produce data stored outside the United States — under certain legal conditions, without necessarily notifying the foreign government concerned.
This legal framework applies to Palantir. The Swiss Army identified it as the primary risk in its assessment. British legal experts have called for NHS contracts to include specific “Transfer Risk Assessments” addressing CLOUD Act exposure. The European Commission has not publicly stated whether any potential use of Palantir technology has been assessed for GDPR compliance in light of this law.
The analogy for a North American reader is direct: the CLOUD Act functions like a federal warrant that can follow data wherever it lives — even in a foreign country’s defense systems. For a European government, entrusting intelligence or public health data to a U.S. firm means accepting, at least in principle, that an American court could theoretically order its disclosure. This risk is not hypothetical — it is written into U.S. law.
Analysis
Europe’s public debate on Palantir has focused on the right subjects but asked the wrong questions. The controversy centers on the firm’s values — its contracts with Israel’s military, its role in ICE immigration enforcement, Alex Karp’s April 2026 manifesto declaring that Silicon Valley engineers have an “obligation to participate in the defense of the American nation.” These facts are real and documented.
But they obscure the structural question: can a democratic state reasonably decide to exit Palantir once it has embedded the firm’s services into core sovereign functions? The answer, in light of the French, British and Swiss cases, seems to be: not without considerable cost, not in the short term, and not without a sovereign alternative that does not yet exist.
Dependency on Palantir is not accidental — it is the product of a deliberate commercial strategy, a technological offering with no public-sector equivalent, and a sovereign alternatives development timeline that has consistently lagged. In France, Artemis.IA, GenIAl.intradef and Comand AI are in development — but their resources (approximately $10 million for Comand AI, a team of around 20 people) are not comparable to those of a firm employing thousands of developers and holding more than $970 million in annual U.S. federal contracts alone, as of 2025.
The real question is not whether European democracies want to exit Palantir. They have said they do. The question is whether political will can precede technical capacity.
The bottom line
European digital sovereignty is often presented as a project — a horizon to be built. The Palantir case reveals it is first a debt: a set of contracts signed without full strategic assessment, whose exit costs now exceed the political and technical capacity to bear them. The Netherlands voted a motion to “reduce dependency” in 2025 — but its minister was forced to concede in 2026 that no alternative was yet available. Switzerland formally recommended substitutions — without identifying any. The UK set a break clause for 2027 — without a qualified competitor in sight.
The open question is not whether European governments want to leave. It is whether the political decision to leave can come before the technical means to do so — or whether states will continue renewing contracts they have declared unacceptable, because they cannot yet do otherwise.
Sources: French Senate (written question Q260508686, May 2026) · NHS England (contract explainer, April 2026) · Medact (briefing, March 2026) · Euronews Next (June 5, 2026) · EU Perspectives (May 2026) · Le Grand Continent (a Paris-based European geopolitics review, April 2026) · The Nerve / Carole Cadwalladr (January 2026) · Swiss Army risk assessment / Republik-WAV (December 2024) · Follow the Money (April 2026)