France bets €655 million more on AI sovereignty
France is doubling down on artificial intelligence. Prime Minister Sébastien Lecornu announced Tuesday an additional €655 million investment through the France 2030 public innovation program — on the very day Washington moved to restrict foreign access to Anthropic’s most powerful AI models. The announcement lands at a moment when France’s own reliance on U.S. tech firms is under active, if unfinished, scrutiny.
At a Glance
France is injecting an additional €655 million into AI through its France 2030 program, targeting computing infrastructure, research, companies, and industrial supply chains.
France’s domestic intelligence agency, the DGSI, is in an ongoing transition away from U.S. data analytics firm Palantir — whose co-founder Peter Thiel is a close ally of Donald Trump — toward French firm ChapsVision, though the shift remains incomplete and may not be fully operational until 2027.
French ministries will now be required to demonstrate their use of AI to receive favorable budget allocations — an unprecedented form of state conditionality.
This image is used for illustrative purposes only.
A political signal as much as an industrial one
On Tuesday, June 16, on the eve of VivaTech — Europe’s largest technology trade show, now in its tenth edition in Paris — Prime Minister Sébastien Lecornu announced an additional €655 million (approximately $710 million at current exchange rates) earmarked for artificial intelligence under France 2030, the French government’s flagship public investment program for innovation.
The funds are designated to strengthen computing infrastructure, research capacity, companies, and industrial supply chains.
The timing carries weight. The previous week, the Trump administration had ordered Anthropic — the U.S. AI startup behind the Claude family of models — to suspend access to its two most powerful systems, Claude Fable 5 and Mythos 5, for “any foreign national,” citing national security. The decision triggered an immediate backlash across France’s political spectrum: Jordan Bardella, president of France’s far-right Rassemblement National (RN) party; Jean-Luc Mélenchon, leader of the hard-left La France Insoumise (LFI); Gabriel Attal, former Prime Minister and centrist figure; Édouard Philippe, former Prime Minister and head of the center-right Horizons party; and Bruno Retailleau, Interior Minister, all warned of an emerging “AI war” and the urgent need for technological independence from the United States.
The Palantir question: transition, not rupture
Among the day’s announcements, one speaks directly to France’s unresolved dependency on U.S. technology: the ongoing effort to migrate the DGSI — France’s domestic intelligence agency (Direction générale de la sécurité intérieure) — away from Palantir, the American data analytics firm whose co-founder Peter Thiel is a close political ally of Donald Trump.
The picture, however, is more complex than a clean break. Palantir’s contract with the DGSI was renewed for three years as recently as December 2025, extending a partnership that dates back to around 2016 and centers on Palantir’s Gotham platform. In parallel, French firm ChapsVision has been developing an alternative sovereign solution — Argonos — through a competitive procurement process. That transition is underway but not yet complete; full operational replacement could take until 2027.
“We cannot accept new strategic dependencies in the digital domain.”
That was Lecornu’s framing Tuesday, as he expressed the desire to “build genuine autonomy” in AI rather than remain exposed to partners who could “turn off the tap” on access. The statement reflects France’s broader posture on digital sovereignty — though it was not accompanied, on June 16, by any announcement of a formal contractual severance with Palantir.
The situation is, in some ways, more telling than a clean rupture would be. France is simultaneously locked into a U.S. contract it is actively trying to exit and building the sovereign alternative it hopes will one day replace it. That is not a weakness so much as a map of the problem: sovereign technological capacity cannot be legislated overnight.
AI as a budget lever — an unprecedented conditionality
Beyond the investment envelope, Lecornu introduced a structural innovation with no real precedent in French public administration: the capacity of government ministries to use AI “will now be factored into budget allocations.” Every ministry will be required to demonstrate how it is deploying artificial intelligence to streamline administrative procedures, improve public services, and reduce unnecessary costs.
The Justice and Interior ministries will gain access “this year” to state-of-the-art AI capabilities through the GenIAl portal — already in use at the Ministry of Armed Forces — to handle sensitive data and, for example, accelerate visa processing. On the health side, Ameli, the website run by France’s national health insurance system (Assurance maladie, roughly equivalent to a national Medicare administration), will launch an AI-powered “public health assistant” allowing patients to “entrust their health data to an AI managed by the national insurer rather than a foreign company.”
Why this matters beyond France
France’s moves fit into a broader pattern, though it would be premature to read them as part of a coordinated European strategy. What is observable is that the U.S. restriction on Anthropic’s models has functioned as a wake-up call: it made concrete, in an operationally immediate way, what it means for state functions to depend on foreign-controlled digital infrastructure.
Lecornu is operating on two fronts simultaneously: domestically, he is imposing modernization through budgetary pressure; externally, he is positioning France as a sovereign actor in a sector where dependence on U.S. or Chinese technology giants has been the norm — including for security services.
The slow-motion Palantir transition is a case study in that tension. Replacing a deeply embedded operational platform requires years, not announcements. The political intent is clear; the operational reality is messier. That gap — between the sovereign ambition and the contractual reality — is arguably more instructive than any single press release.
One broader question remains. Are €655 million enough to matter at scale? U.S. public and private AI commitments — across various estimates — run into the hundreds of billions of dollars. The gap is enormous. France is betting on targeted precision — sovereign infrastructure, public data platforms, core state functions — rather than raw volume. Whether that bet pays off will depend less on the size of the check than on the coherence of the strategy behind it.
The bottom line
The real question this announcement raises is not whether €655 million is enough to make France an AI power. It is whether Europe is capable of turning the Anthropic episode — a unilateral American decision taken without consulting its allies — into the starting point for a common industrial policy. The DGSI’s slow exit from Palantir is a microcosm of the larger challenge: the distance between declaring sovereignty and achieving it is measured not in press releases, but in years of procurement, testing, and transition. So far, every country has been navigating that distance alone.
Sources: France 24 · AFP