Thales launches cortAIx in Germany to industrialize reliable AI in defense. The goal: to integrate AI into critical systems without losing control.
In summary
On January 20, Thales officially launched cortAIx in Germany, a new branch of a global network dedicated to the industrialization of AI in defense and sensitive infrastructure. The announcement may seem abstract, but it targets a very concrete reality: integrating AI into systems where error is not permitted. Thales presents cortAIx as a specialized accelerator focused on military and security applications, with one absolute priority: to produce trustworthy AI that can be used in the field. The challenge goes beyond simple algorithmic performance. It is about building a chain of trust around data, models, and decisions, so that AI remains understandable, controllable, and certifiable. Why Germany? Because the country has once again become a center of investment and demand for critical technologies, particularly in cybersecurity and command systems. For Thales, this launch is a strategic milestone: AI is no longer a laboratory, it is becoming a European industrial building block.
The launch of cortAIx in Germany, a strategic signal that is broader than it appears
The January 20 announcement is factual: Thales is opening a new cortAIx entity in Germany, bringing the number of sites in the network to five, after France, the United Kingdom, Canada, and Singapore. This detail is important. Thales is not talking about an isolated “AI project,” but rather an organized, replicated infrastructure designed to produce at scale.
The implicit message is clear. AI is no longer presented as a “plus” to improve a few functions. It is considered a structural element of operational superiority and system resilience. In a military environment where jamming, saturation, and cyber attacks are becoming permanent, AI is used to maintain decision-making quality and service continuity.
This strategy also responds to an industrial reality: it is becoming difficult to recruit and retain high-level AI talent without offering them an applied research framework, concrete projects, and deployment capacity. By opening in Germany, Thales is positioning itself close to a growing European defense market and a dense technical ecosystem.
What cortAIx is, beyond a brand name
The word cortAIx is presented as an accelerator. In industrial language, this means a structure that shortens the transition between research, prototypes, and deployable products. Here, the focus is explicit: AI applied to critical systems, i.e., environments where safety, security, and availability take precedence over the “wow” effect.
An internal accelerator that pools AI building blocks
Thales claims that cortAIx brings together AI expertise related to sensors, signal processing, embedded systems, and software architectures. The objective is twofold.
First, to avoid fragmentation. In a multi-domain group, AI can be developed in silos. This results in incompatible models, different evaluation tools, and duplication of costs.
Second, to industrialize. AI that is useful in defense is not just a high-performance model. It is a versioned, tested, monitored, secure, and maintainable model. This is often where AI projects fail: they work in demonstration, then come up against the reality of qualification, support, and integration.
A priority: trustworthy AI, not “black box” AI
Thales emphasizes the concepts of transparency and robustness. The company insists on one requirement: systems must remain explainable and governable. This brings us back to the idea of certifiability, which is a harsh but essential word: if AI cannot be qualified, it will not be used in critical functions.
In the military world, “opaque” AI can be tolerated in non-decisive recommendations.
It cannot be tolerated in a firing chain, deconfliction, navigation in contested space, or automatic cyber protection.
Why Germany, and why now
Setting up cortAIx in Germany is not a neutral choice. Germany combines three factors: a market, an urgency, and a technological fabric.
A market that has regained its central role in European defense
Germany is in the process of strengthening its capabilities, with a strong focus on infrastructure protection, technological sovereignty, and the modernization of defense systems. In this context, a player like Thales has an interest in being local, as close as possible to real needs and decision-makers.
This choice is also in line with a European approach: if defense AI is to be sovereign, it must exist in several countries, with robust industrial and academic collaborations, and not just in a single center.
An ideal terrain for cybersecurity and command architectures
The German launch stands out for its announced focus areas: cybersecurity, command systems, and sensor-centric applications. This triptych is consistent with current priorities.
A modern defense system does not win solely on the basis of its platform. It wins through sensor-effect integration, processing speed, and the ability to survive cyberattacks and jamming.
On these issues, Germany represents a highly credible industrial testing ground, as demand is strong and digital transformation programs have become a priority.
What is cortAIx used for in the field, in concrete terms?
Thales cites several areas of work. What is interesting is that we are not talking about “general” AI. We are talking about military use cases that are limited but powerful.
Better informed and faster command systems
In command and control systems, AI is used to merge massive data streams. A modern force processes radar tracks, tactical links, image intelligence, unit reports, and electromagnetic signals.
The challenge is not only to see, but to prioritize. In a saturated sky, AI can help classify threats, detect inconsistencies, and reduce the cognitive load on operators.
Be careful, this does not replace a leader. It improves the quality of the picture, and therefore the speed of action.
Autonomous cyber defense, useful but politically sensitive
Thales cites autonomous cyber defense and agent-based penetration testing. In practice, this can cover two very different functions.
The first is defensive: detecting abnormal behavior, isolating an area, reacting quickly, and proposing a corrective plan. This is very useful in a critical system, where a minute’s delay can be enough to lose control.
The second is offensive in simulation: testing your own systems to identify vulnerabilities before your adversary does. This is a logic of robustness, not a fantasy of “automatic cyberwarfare.”
The key point is governance. AI in cybersecurity must not act without safeguards. It must be supervised, audited, and limited in its actions.
More efficient sensors, without blowing up bandwidth
The sensor applications axis is central. Embedded AI can filter, compress, and select useful information directly at the sensor level. This changes the operational equation, as it avoids sending “everything” to the cloud or to a back-end center.
In a contested war, bandwidth and connectivity are not guaranteed. Useful AI is often frugal AI, capable of operating locally and degrading its performance cleanly without collapsing.
How to integrate AI into a critical system without deceiving ourselves
Defense systems cannot be satisfied with “benchmarks.”
They must function under degraded conditions.
Explainability as a condition for deployment
Explainability does not mean “understanding everything.” It means understanding enough to decide and justify. In a critical system, it must be possible to explain why a lead is classified as hostile, why network behavior is deemed suspicious, or why a route is proposed.
Without this ability, AI becomes a black box. And a black box becomes an institutional risk.
Robustness to attacks and deception
Military AI must withstand:
- jamming,
- deception,
- decoys,
- adversarial attacks,
- “poisoned” data.
These risks are not theoretical. They are already being observed in the field in conflicts where the enemy tests, adapts, and exploits the slightest weakness.
Qualification and maintenance
In the defense industry, an AI model is not “delivered once.” It must be maintained, corrected, and retrained, but without breaking certification.
It’s a paradox: AI learns, but a critical system must remain stable. The solution involves strict governance, frozen versions, and controlled update cycles.
This is typically the type of project where an accelerator like cortAIx can make a difference, as it combines AI engineering and industrialization requirements.
The strategic impact: Thales strengthens a European center of gravity
The creation of cortAIx in Germany has several strategic impacts, visible in the short and medium term.
A response to European demand for digital sovereignty
Defense AI cannot depend solely on uncontrolled building blocks. The issue is not anti-American or anti-technology. It is operational: who controls the data, models, and tool chains?
By deploying in several countries, Thales is strengthening a form of European digital sovereignty. This does not mean self-sufficiency. It means the ability to choose, audit, and remain in control of the life cycle.
A lever of credibility with the armed forces
For the military, AI is only acceptable if it is robust, explainable, and sustainable. Thales therefore seeks to position itself as an “AI manufacturer,” not a purveyor of promises.
In this context, Germany’s openness adds depth: greater customer proximity, greater co-development capacity, and a better understanding of employment constraints.
Consistency with next-generation aviation programs
One point is worth noting: Thales has already formalized a partnership with Dassault Aviation around “controlled and supervised” AI for defense aviation. This directly addresses the challenges of future collaborative combat: drones, sensors, data fusion, and assisted decision-making.
AI is not an option for these architectures. It is a prerequisite for operation. cortAIx therefore serves as a cross-functional building block, ready to be used in several programs, not just one product line.
What cortAIx will not do, and that’s healthy
Let’s be honest. cortAIx does not guarantee an immediate leap in superiority. It guarantees a framework for industrialization.
AI in defense is often the victim of two extremes:
- enthusiasm, which promises total autonomy,
- fear, which imagines a total loss of control.
The reality is more pragmatic. Systems advance incrementally: better fusion, better detection, better cyber-response, better prioritization.
And that is exactly what saves time, ensures survival, and improves accuracy.
The real test ahead: speed of integration, not volume of announcements
If cortAIx wants to make an impact, it will have to produce visible results:
- building blocks integrated into existing systems,
- demonstrations in contested conditions,
- credible qualification processes,
- and an ability to deploy without increasing maintenance costs.
The launch in Germany shows that Thales wants to be at the center of the European game. The question is no longer “Is AI useful?” The question is: who knows how to industrialize it properly, and who knows how to make it acceptable in a critical system?
This is where the accelerator will be judged. Not on a promise, but on its ability to deliver effective, controlled, and sustainable AI.
Sources
- Thales — “cortAIx, Thales’s AI accelerator, launched in Germany to Drive AI for Critical Systems”
- Thales — “Building Trustworthy AI for Critical Systems”
- Thales — “Thales speeds up its development of AI for defense”
- Thales — “Thales launches cortAIx in the UK with 200 experts in AI for critical systems”
- Thales — “PROTECT TO BUILD A FUTURE WE CAN ALL TRUST (CSR Integrated Report 2024)”
- Thales — “Dassault Aviation and cortAIx sign a strategic partnership for a sovereign AI serving the air combat of the future”
- EDB Singapore — “Thales accelerates AI Innovation with cortAIx SG”
- ASDNews — “Thales and Partners Selected for 1st European Project to Develop Sovereign AI Embedded Cyberdefence”
- EDR Magazine — “cortAIx, Thales AI accelerator, launched in Germany to Drive AI for Critical Systems”
- WebWire — “cortAIx, Thales’s AI accelerator, launched in Germany to Drive AI for Critical Systems”
War Wings Daily is an independant magazine.