Dassault Aviation and Thales are joining forces around cortAIx to develop sovereign artificial intelligence dedicated to the fighter jets and drones of tomorrow.
In summary
The partnership between Dassault Aviation and Thales via its AI accelerator cortAIx marks a milestone for French and European defense aeronautics. Signed on November 18, 2025, and presented at the Adopt AI Summit in Paris, the agreement aims to develop controlled and supervised embedded AI for piloted aircraft and drones, capable of analyzing tactical situations, assisting in decision-making, planning missions, and coordinating swarms of platforms. The ambition is clear: to have sovereign AI, designed and hosted in Europe, adapted to security, certification, and defense secrecy constraints. Backed by more than 600 AI experts at Thales and more than €1 billion in annual R&D at Dassault, this alliance is set to inform future Rafale standards, combat drones, and programs such as the Future Combat Air System. It also responds to a political challenge: reducing dependence on large non-European digital platforms.
The strategic partnership between Dassault Aviation and Thales
The agreement announced at the end of November 2025 between Dassault Aviation and Thales via cortAIx concerns the development of “controlled and supervised AI” for defense aeronautics. It covers key combat aviation functions: observation, situation analysis, decision support, planning, and operations control for both piloted aircraft and autonomous systems.
In concrete terms, this involves embedding AI components in future Rafale standards (notably Rafale F5) and combat drones, while connecting them to ground-based and “combat cloud” processing capabilities. These algorithms will have to operate in harsh environments: energy, latency and cybersecurity constraints, as well as certification requirements specific to critical aeronautical systems.
This partnership capitalizes on trajectories already underway. Dassault is investing more than €1,108 million in global R&D (including €539 million in financing and €483 million in self-financing) for the Rafale F4/F5 standards, UCAS drones, and the NGF demonstrator. For its part, Thales achieved a turnover of €20.6 billion in 2024, including nearly €11 billion in defense, and is positioning AI as a central focus for its growth.
A response to the demands of collaborative combat
This move is in line with the Future Combat Air System, where “combat cloud” and collaborative systems must connect fighter jets, drones, land and sea sensors, and command centers via distributed and highly digitized architectures. Thales is already a leader in the “Combat Cloud” component of FCAS, while Dassault remains the architect of fighter jets.
AI is no longer just a tool for marginal optimization. It is becoming the core of operational value:
- Automated prioritization of threats in congested airspace;
- Real-time management of mixed aircraft/drone formations;
- Dynamic reconfiguration of missions in the face of mobile and jamming adversary systems.
The partnership therefore aims to move from peripheral decision-making tools to AI integrated into the very architecture of weapons systems, while keeping the pilot at the center of the decision-making loop.
The cortAIx platform, the backbone of sovereign AI
The agreement is based on cortAIx, Thales‘ AI accelerator, presented as the first European platform dedicated to critical AI applications for defense, aeronautics, space, and cybersecurity.
Thales claims to have more than 600 AI and data experts, around 100 AI doctoral students each year, and more than 200 patents filed in AI for critical systems. The manufacturer already has more than a hundred products incorporating AI components, from air defense radars to cybersecurity systems.
A Lab, Factory, and Sensors architecture
cortAIx is based on a three-pillar architecture:
- cortAIx Lab: an integrated laboratory based in Saclay, presented as the most powerful in Europe for critical AI. It brings together fundamental research, academic collaborations, and rapid algorithm prototyping, with direct access to high-performance computing infrastructure.
- cortAIx Factory: an AI industrialization “factory” responsible for transforming demonstrators into certifiable products. It standardizes development methods, datasets, MLOps tools, and manages the entire model lifecycle (training, validation, deployment, updates).
- cortAIx Sensors: the component dedicated to sensors (radar, optronics, electronic warfare). The aim is to integrate AI as close as possible to the data, directly into the embedded processing chains, to reduce latency and limit the data flows to be transmitted.
For Dassault, this architecture provides a framework for integrating AI into aircraft design from the earliest stages: scenario simulation, algorithm training on digital twins of sensors or cockpits, and then transfer of these models to aircraft computers.
Integrating AI into aircraft and drones
One of the major challenges will be integrating these cortAIx building blocks into existing and future avionics architectures. This requires hardened computers capable of executing AI models with strict real-time, temperature, and vibration constraints.
In a Rafale or a future NGF, AI will be involved at several levels:
- Fusion of multi-sensor data (AESA radar, IRST, electronic warfare, data link) to produce a synthetic tactical image that is intelligible at a glance for the pilot;
- Filtering and prioritization of alarms to avoid cognitive overload;
- Recommendations for avoidance trajectories or attack profiles while respecting flight envelopes and rules of engagement;
- Management of remote carrier drones escorting the aircraft, with real-time task allocation (reconnaissance, decoy, saturation).
Use cases for crews and autonomous systems
For crews, the challenge is not to replace the pilot, but to transform the aircraft into a digital teammate. For example, AI will be able to:
- Automatically analyze high-definition video feeds to detect objects of interest on the ground several dozen kilometers away;
- Propose attack or penetration plans based on enemy radar coverage, weather, and available fuel;
- Optimize navigation in real time to limit radar or infrared signatures.
For more autonomous systems, such as combat drones, AI will have to manage tasks such as detecting cyberattacks on board, reconfiguring the mission in the event of a loss of connection, or executing group behaviors in a swarm. Thales is already working on deployable AI agents for onboard cyberdefense in aircraft, through the AIDA project funded by the European Defense Fund.
The question of budgets and investments
This type of program cannot be conceived outside of budgetary constraints. The development of certifiable onboard AI for defense aeronautics will cost hundreds of millions of euros over the decade, divided between upstream studies, software development, flight testing, and data infrastructure.
The scale of the investment is already clear:
- Thales is forecasting defense revenues of nearly €11 billion in 2024, up nearly 14%, with an order book of €47 billion in mid-2024.
- Dassault Aviation is spending more than €1 billion per year on R&D, with a growing share dedicated to integrating new digital capabilities into the Rafale and future programs.
- France plans to increase its defense spending from around €50 billion in 2025 to more than €60 billion by 2030, creating a favorable environment for structural investments in weapons systems and AI.
In this context, the Dassault–Thales alliance allows for the pooling of investments, teams, and infrastructure. Instead of multiplying isolated demonstrators, the idea is to industrialize an “AI backbone” that can be reused across several programs: Rafale, drones, FCAS, ground-to-air systems, and even naval and space applications.
Data security and the advantage of sovereign AI
One of the central arguments for the agreement is the promise of sovereign artificial intelligence. Behind the phrase lies a very concrete issue: the operational data of the armed forces—flight paths, radar modes, rules of engagement, mission profiles—cannot be transferred to foreign commercial clouds subject to extraterritorial legislation.
cortAIx is designed to work with:
- Controlled computing infrastructures, hosted on industrial or government sites;
- Encrypted and segmented processing chains, limiting the risk of data leaks;
- Explainable AI models, designed to avoid the “black box” effect in high-stakes decisions.
Thales emphasizes that its AI solutions for defense must be “robust, cyber-secure, and transparent” in order to withstand adversaries capable of launching attacks through corrupted data, sensor disruption, or model exploitation. Work on the AIDA AI agent for aircraft cyber protection illustrates this trend toward integrating cybersecurity from the design stage.
For client states, the benefit is twofold: limiting dependence on non-European big tech and retaining control over algorithms, training datasets, and software updates throughout the lifecycle of next-generation combat aircraft.
Competitive advantages over other AI approaches
Compared to more generalist AI approaches, which focus on the commercial cloud and very broad models, the Dassault–Thales alliance is banking on “mission-specific” AI:
- Adapted to limited embedded computing resources (“frugal AI”);
- Designed to coexist with certified avionics software;
- Designed from the outset for export control and cybersecurity.
This focus on controlled and supervised embedded AI can be a competitive advantage in the international fighter jet market. Many countries are now seeking advanced AI capabilities without accepting complete dependence on the software and cloud ecosystem of a single country. An offering combining Rafale F5, Thales mission systems, and cortAIx building blocks could meet this demand.
For both the French armed forces and export customers, the challenge will be to achieve measurable operational gains: reduced decision-making time, improved survivability in contested environments, and optimized maintenance through predictive fault detection. At this level, AI will not be judged on its algorithmic elegance, but on its concrete impact in operations.
A milestone for French and European combat aviation
This partnership alone does not resolve the tensions surrounding major programs such as FCAS, but it sends a signal: France and its manufacturers are determined to secure autonomous expertise in aeronautical defense AI, even if it means moving faster on certain pillars such as collaborative air combat.
The coming years will allow several key points to be verified:
- The actual ability to integrate these cortAIx components into existing aircraft without exploding costs or delays;
- The maturity of adaptive algorithm certification processes in an environment where aviation authorities remain, rightly, very cautious;
- The acceptance by crews of a digital teammate that offers tactical options without ever replacing human decision-making.
If these obstacles are overcome, the Dassault-Thales agreement could become a textbook case of convergence between critical AI, sensitive operational data, and weapons system architectures. If not, it will at least serve as a large-scale laboratory for testing, in real-world conditions, what AI can—and cannot—do in the cockpit and at the heart of combat systems.
Sources
- Thales press release, “Dassault Aviation and cortAIx sign a strategic partnership for a sovereign AI serving the air combat of the future,” November 2025.
- Thales, press releases “Thales speeds up its development of AI for defense” and presentation of cortAIx (Lab, Factory, Sensors), 2024.
- Thales, 2024 annual results and 2024 half-year results, indications on defense revenue and order book.
- Dassault Aviation, institutional documents and 2024 annual report, R&D data and Rafale F4/F5 and NGF programs.
- Thales, AIDA projects and publications on AI for embedded cyberdefense and critical systems.
- Specialized summary articles (Airforce Technology, Defence Industry Europe, Meta-Defense analysis) on the Dassault–Thales partnership and the challenges of embedded AI for combat aviation.
War Wings Daily is an independant magazine.