France Pours €655M Into Sovereign AI Chatbot for Million-Strong Civil Service

France will spend an additional €655 million on artificial intelligence, Prime Minister Sebastien Lecornu announced on Monday, with the centerpiece being a single sovereign conversational assistant for roughly one million civil servants. The chatbot is designed to handle administrative tasks such as streamlining judicial procedures, supporting research grant applications, and managing routine document processing.

The funding is a top-up to the roughly €109 billion in private AI investment France committed at last year's Paris summit. The chatbot is the most visible element, but the package also includes a dedicated public-health assistant for the state health insurance agency Ameli, a new platform for public data access, and investments in computing capacity, research, and industrial AI adoption.

Digital Sovereignty vs. Microsoft Dependence

France's push for digital sovereignty faces a persistent challenge: deep integration with US cloud providers. The Register reports that while France has successfully deployed Nextcloud for locally controlled storage, getting users off Microsoft Office and other American platforms remains difficult. The French government's own systems still rely heavily on Microsoft 365 and Azure, creating a tension between the sovereignty rhetoric and operational reality.

• The sovereign chatbot is intended to run on French-controlled infrastructure, not rented from US providers.
• Mistral AI, valued at €20 billion in recent funding talks, is the most likely domestic supplier for the assistant, though no vendor has been named.
• The EU is also planning gigafactory sites for AI compute, with French companies bidding for a site.
• No timeline has been given for the chatbot's deployment, nor a detailed breakdown of how the €655 million will be split across projects.

Implications for European Cloud and AI

The French approach reflects a broader European strategy to build sovereign AI infrastructure rather than lease frontier capacity from US firms. A guaranteed buyer the size of the French state provides the demand signal needed for domestic AI companies like Mistral to scale. However, the gap between ambition and execution is wide: moving a million civil servants off entrenched US platforms is a multiyear effort that requires not just new software but retraining, data migration, and cultural change.

What comes next is a series of tenders for the individual projects. The government has described the capability it wants, not the vendor it will use. The competition is no longer only about who trains the best model, but who supplies the public sector. Several European governments are now writing AI procurement directly into their spending plans, and France's €655 million is a down payment on that principle.