Paris turns digital sovereignty from political rhetoric into security policy
PARIS, FRANCE | JUNE 2026. France has chosen domestic technology company ChapsVision to replace Palantir in the large-scale data-analysis operations of the General Directorate for Internal Security, known as the DGSI. Prime Minister Sébastien Lecornu presented the decision as a strategic step toward genuine French autonomy in digital infrastructure and artificial intelligence, arguing that critical state capabilities cannot remain dependent on foreign providers whose access, policies or technologies could be restricted by political decisions made abroad.
The announcement marks a significant change in France’s intelligence architecture. Palantir has worked with the DGSI since the aftermath of the November 2015 terrorist attacks in Paris, when French security services urgently needed a platform capable of integrating and processing enormous volumes of fragmented information. The American company’s Gotham system offered advanced tools for connecting databases, identifying patterns and supporting intelligence investigations at a scale that few European alternatives could match at the time.
France initially portrayed its dependence on Palantir as temporary. Contracts were signed and renewed while the government waited for a credible national solution capable of meeting stringent operational and security requirements. The relationship nevertheless continued for approximately a decade, with renewals in 2019, 2022 and again in December 2025. The most recent extension reportedly covered three additional years, making the sudden selection of a French replacement especially notable.
Palantir responded by stating that its existing contract remains fully in force. This creates uncertainty regarding the timing and structure of the transition. French authorities have not yet explained whether ChapsVision will immediately assume all functions, operate alongside Palantir during a phased migration or gradually replace specific analytical capabilities. Moving highly sensitive intelligence systems is technically complex and cannot be treated as a routine software substitution.
The challenge is not limited to transferring databases. Intelligence platforms combine classified information, investigative workflows, analytical models, access controls and institutional knowledge developed over years of use. Any transition must preserve continuity while preventing security gaps, data loss or reduced analytical capacity. France must demonstrate that strategic autonomy can be achieved without weakening the DGSI’s ability to detect terrorism, espionage, cyberattacks and foreign interference.
ChapsVision already possesses experience within the French intelligence environment. The company secured an initial DGSI contract in 2024 for processing heterogeneous data and now aims to take responsibility for the exploitation of much larger information volumes. Its selection represents a major commercial and institutional victory for a company seeking to become a European leader in data intelligence and agent-based artificial intelligence.
The broader message extends beyond a single contract. France increasingly views digital dependence as a national-security vulnerability. Governments that rely on foreign cloud systems, artificial-intelligence models and intelligence platforms risk losing operational freedom when political relations deteriorate. A supplier headquartered under another jurisdiction may be subject to export controls, security directives or legal demands that conflict with the interests of the client state.
Lecornu’s warning focused on the possibility that strategic partners could effectively turn off access to essential artificial-intelligence capabilities. That concern has intensified as Washington adopts more restrictive policies around advanced models and technologies considered sensitive. For France, sovereignty now means more than keeping data physically within national borders. It also requires control over software, analytical infrastructure, updates, licensing and the conditions under which critical systems remain available.
The decision reflects a wider European reassessment of Palantir. Germany’s military has reportedly moved away from the company, while political and parliamentary pressure has intensified in the United Kingdom over contracts involving the National Health Service and law enforcement. These debates combine concerns about procurement, privacy, foreign influence, data governance and dependence on a company closely associated with the American defense and intelligence establishment.
Palantir’s technological performance is not necessarily the central issue. Its platforms have gained influence precisely because they can integrate complex information and generate operationally useful intelligence. The French decision instead reflects a strategic calculation: even an effective foreign technology can become problematic when it occupies a position that a government considers indispensable to national security.
France is supporting this shift with broader investment. Lecornu announced plans to allocate €655 million to artificial intelligence and create a common chatbot for government services. Additional projects include a health-focused chatbot for the national insurance system and a digital platform intended to simplify access to public data. Together, these initiatives suggest that Paris wants to build an integrated national AI ecosystem rather than merely replace one American contractor.
Success, however, will depend on capability rather than nationality alone. A domestic platform must meet the same standards of reliability, scalability, cybersecurity and analytical precision expected from established global providers. Sovereignty cannot be achieved by purchasing inferior technology under a national label. It requires sustained investment, access to specialized talent and a procurement system capable of supporting innovation without sacrificing accountability.
The transition also raises questions about oversight. Intelligence applications of artificial intelligence can strengthen national security, but they can also expand surveillance capacity and concentrate power inside institutions that operate largely outside public view. France will need strong safeguards governing data access, algorithmic accountability and the proportionality of analytical tools. National ownership does not automatically eliminate the risks associated with mass data processing.
France’s decision therefore represents more than a change of supplier. It is a declaration that control over data and artificial intelligence has become inseparable from political sovereignty. The central test will be whether ChapsVision can provide operational independence while maintaining the security performance France obtained from Palantir. Should the transition succeed, it could become a model for European governments seeking technological autonomy. Should it fail, it may expose how difficult it is to escape systems on which states have already become strategically dependent.
Digital sovereignty begins where technological dependence ends.