Competition over Independent AI Command and Control Systems in Europe and Non-Western Nations: Implications for South Korea's Response
Executive Summary
Thales's HexaForce (France) and SSB's EVREN (Turkey) represent proactive efforts by European and non-Western actors to seek alternatives to the US Palantir-led CJADC2 standard and ontology system. These platforms are already close to the procurement stage, having successfully passed NATO's official Coalition Warrior Interoperability Exercise (CWIX). Both cases stem from a shared concern: how to maintain interoperability while avoiding dependency on foreign standards. This can be interpreted as a reaction to the structural dependency risks exposed by Japan's adoption of Palantir Maven. As warned by the Supreme Allied Commander Europe, tensions persist within Europe between strengthening independent capabilities and maintaining the alliance's command structure, leaving the future of this standard competition highly fluid. Against the backdrop of its ongoing wartime operational control (OPCON) transition roadmap, South Korea should consider a dual-architecture strategy that separates the data layer from the application layer, while simultaneously securing dedicated computing infrastructure for defense AI.
I. Issue Analysis
Competition over Independent AI-C2 Systems in Europe and Non-Western Nations: Issue Analysis
1. Background and Progress
The unveiling of Thales's HexaForce was not a matter of chance. Thales tested the system during NATO's Coalition Warrior Interoperability Exercise (CWIX) held in Poland last June [1]. The official unveiling took place approximately three months after the exercise, a decision made when consultations with NATO member states were already well underway [1]. Patrick Moreau, Thales's Vice President for Multi-Domain Operations, stated that the company is engaged in "highly advanced" discussions with several NATO countries [1]. The company is positioning HexaForce as a multi-domain "overlay" that can integrate legacy and national systems from the strategic to the tactical level [1].
The background behind Turkey's launch of EVREN differs in character from that of France. Led by Turkey's Presidency of Defense Industries (SSB), this platform targets a broad user base, including defense companies, technology firms, academia, entrepreneurs, and students [4]. The SSB's announcement emphasized data sovereignty over technical performance. It explicitly stated that all prompts, responses, and utilized data are processed within Turkey's domestic high-performance GPU infrastructure, fundamentally preventing the transfer of sensitive data to foreign clouds [4]. This is an extension of the defense self-reliance policy that Turkey has pursued since its exclusion from the F-35 program.
The common thread running through both cases is a wariness of the ontology standards and CJADC2 concepts led by the US firm Palantir. A previous EAI analysis examined Japan's Ministry of Defense allocating funds in its FY2027 budget of 8.9 trillion yen to procure the Palantir Maven smart system for the Self-Defense Forces' Joint Operations Command [3][6]. The core issue highlighted in that case was "the potential for structural dependency to develop due to rising transition costs, despite a dual roadmap aimed at eventually replacing it with domestic systems" [3]. The moves by France and Turkey can be read as proactive responses by European and non-Western actors to mitigate this risk of dependency.
2. Current Status
Statements from Thales indicate that the system has not yet reached the stage of formal contract signing. While the phrase "highly advanced" suggests mature discussions, it does not constitute a binding agreement [1]. However, passing the CWIX test means the system has cleared a baseline hurdle in NATO's interoperability certification process [1].
The activities of the German defense tech company Helsing follow a similar trajectory. Helsing announced that Centaur, the AI piloting system for its CA-1 Europa unmanned combat aerial vehicle (UCAV), defeated human pilots in simulated dogfights [7]. During a demonstration conducted in front of German Digital Minister Carsten Wildberger, a former Mexican Air Force pilot and a German military test pilot were repeatedly defeated [7]. With its first test flight scheduled for next year, this project also illustrates the competitive landscape in which European defense firms are rushing to demonstrate their proprietary AI combat capabilities.
Parallel efforts are also underway regarding data sharing from the Ukrainian frontlines. The UK Ministry of Defence has partnered with Ukraine's Avengers AI Lab [13]. This is significant as it marks the first time Kyiv has opened up data from its battlefield drone AI training platform to a foreign government [13]. This represents another pathway for Europe to accumulate real-world combat data independently, without relying on US standards.
Meanwhile, signals of caution regarding Europe's independent moves are emerging from NATO command. General Alexus Grynkewich of NATO warned that EU defense efforts should not compete with the alliance [10]. This remark, while not limited to AI-C2 systems, reflects the discomfort of Washington and NATO leadership regarding Europe's broader self-reliance efforts.
3. Key Actors and Positions
Thales (France) adopts an "overlay" strategy, placing its system on top of existing legacy and national systems rather than replacing NATO standards entirely [1]. This approach eats into the market while causing less friction than Palantir's wholesale replacement. For Thales, as a European defense company, attracting a majority of NATO member states into its ecosystem ahead of US standards is a matter of business survival.
Turkey's Presidency of Defense Industries (SSB) prioritizes keeping data, models, and computing resources entirely within the country [4]. Despite being a NATO member, Turkey has a history of turbulent defense relations with the United States. Its exclusion from the F-35 program led Turkey to view technological self-reliance not merely as industrial policy, but as a means of managing security risks. In this context, EVREN is an attempt to anchor the entire defense industry ecosystem to a domestic platform.
Helsing (Germany) competes by directly embedding AI piloting capabilities into complete weapon systems [7]. Its approach of conducting public demonstrations in front of high-ranking government officials is clearly intended to influence the German government's procurement decisions.
NATO Command exhibits a dual attitude: it seeks to manage individual national systems through interoperability verification frameworks like CWIX, while remaining wary of European independent initiatives clashing with the alliance's systems [1][10].
The United States (Palantir and the Department of Defense) does not appear as a direct spokesperson in this matter, but serves as the de facto background and reference point for seeking alternatives. As seen in the case of Japan, the adoption of Palantir Maven integrates an ally's data systems into US ontology standards under the guise of interoperability [3][6]. The independent paths taken by Europe and Turkey are strongly characterized as reactions to avoid being integrated into this structure.
4. Key Issues
The first issue is the tension between interoperability and technological sovereignty. While Thales's overlay strategy is a compromise aimed at securing domestic industrial interests without compromising interoperability, it has the potential to clash with NATO's standardization logic in the long run.
The second issue is the institutional tension between NATO and the EU. General Grynkewich's remarks suggest that discussions on European defense self-reliance are not only confined to political integration discourse, but could also cause friction with NATO's command structure in concrete technological areas like AI-C2 [10]. This aligns with the structural constraints pointed out in a previous EAI analysis, namely "the differing threat perceptions between France and Germany and the institutional coexistence of the EU and NATO" [12].
The third issue is the national differences in how data sovereignty is secured. While Turkey focuses on securing sovereignty through the localization of physical infrastructure, France places more weight on standard competition at the software layer. This divergence suggests that the ways in which European and non-Western nations respond to US standards will not be uniform in the future.
II. In-Depth Analysis
Competition over Independent AI-C2 Systems in Europe and Non-Western Nations: In-Depth Analysis
1. Root Cause Analysis
The root cause of this competition is not a technological issue, but a question of sovereignty. Palantir's Maven smart system is not just a software package. It is a system designed on top of the CJADC2 concept and ontology standards accumulated by the US Department of Defense over nearly two decades [3]. The moment an ally adopts this system, its military data is integrated into the data classification system and tagging structure defined by the United States. While this process is carried out under the banner of interoperability, it practically results in a single nation monopolizing the standards of the data layer.
The threat perceptions of France and Turkey stem from precisely this point. It is no coincidence that Thales designed HexaForce as an "overlay that can integrate legacy and national systems" [1]. It is a compromise aimed at meeting NATO interoperability requirements without becoming dependent on Palantir standards. Similarly, what the Turkish SSB emphasized in EVREN was not technical performance, but data sovereignty that "fundamentally prevents the transfer of sensitive data to foreign clouds" [4]. Neither country rejects the security partnership itself; rather, they are wary of the partnership solidifying into a one-sided dependency on technological standards.
Underlying this wariness is the path-dependency issue demonstrated by the Japanese case. Although Japan's Ministry of Defense put forward a dual roadmap to eventually replace the system with domestic alternatives, analyses suggest that "because the data system is designed on top of US military ontology standards from the outset, a structural dependency may form where transition costs grow over time" [6]. The moves by France and Turkey should be viewed as proactive efforts to avoid this trap.
2. Structural Context
In terms of security structure, this competition reflects a double bind within NATO. General Alexus Grynkewich of NATO warned that EU-level independent defense efforts must not compete with the alliance itself [10]. Paradoxically, this remark suggests that the movement to secure independent capabilities within Europe has already materialized to the point where NATO leadership is taking notice. The testing of Thales's HexaForce within the official NATO CWIX framework can also be read as a calculated move to strike this balance. It is a dual strategy: integrating into NATO standards while ensuring that the entity defining those standards is not narrowed down solely to the United States.
In terms of economic structure, the computing infrastructure gap acts as a practical constraint on this competition. Anders Dam Jensen, Executive Director of the European High Performance Computing Joint Undertaking (EuroHPC), points out that Europe lags significantly behind the US and China in AI model development, nurturing AI companies, and building the computing power to support them [15]. This is the context behind EuroHPC's investment of 75 billion kroner. In other words, the attempts by France and Turkey to build independent AI-C2 systems cannot be achieved through political will alone; they must be accompanied by competition in the underlying infrastructure to secure domestic GPU infrastructure and computing resources. Turkey's decision to anchor EVREN to its domestic high-performance GPU infrastructure [4] can be seen as a design choice made with this constraint in mind.
In terms of industrial structure, the competition among defense firms overlaps with interstate competition. Helsing's demonstration of its AI unmanned combat aircraft, Centaur [7], is a separate project from Thales and EVREN, but it shows that European defense firms are under pressure to quickly demonstrate their respective AI capabilities. This contrasts with political integration discussions at the EU level, which remain largely declarative, and aligns with the European defense industry landscape where technological competition at the individual firm and project level is already proceeding with concrete schedules [12].
3. Historical Precedents and Comparative Cases
The closest point of comparison is Japan's adoption of Palantir Maven. Japan allocated funds in its FY2027 budget of 8.9 trillion yen to procure the Maven smart system for the Self-Defense Forces' Joint Operations Command [3][6]. The war in Ukraine and the US-Israel-Iran conflicts served as catalysts that accelerated this decision [6]. The moves by France and Turkey share the same starting point, having realized the military necessity of AI command and control systems through these same conflicts. However, their destinations differ. Japan prioritized adopting US standards and deferred replacing them with domestic systems to a later stage. Conversely, France and Turkey are attempting to secure domestic or proprietary standards first. This difference stems from the nature of their alliance structures. Japan is in a position where it directly feels burden-sharing pressures within the bilateral US-Japan alliance [6]. In contrast, France possesses relatively more negotiating leverage within the multilateral NATO framework, and Turkey already has the momentum of an independent path that has reduced its reliance on the US since its exclusion from the F-35 program.
The case of AI training data sharing between Ukraine and the UK [13] is another type of precedent. This is not a competition over standards, but rather a matter of transferring real-world combat data assets between nations. As the first instance of Kyiv opening up battlefield drone AI training data to a foreign government, it demonstrates a third model of selective cooperation while maintaining data sovereignty. It shares a similar logic of compromise with Turkey's EVREN, which keeps data processing domestic as a rule while opening the platform to the broader defense industry ecosystem.
China's path toward military AI integration also serves as a valid point of comparison. According to a Brookings Institution analysis, China has pursued an innovation-driven military AI strategy centered on the People's Liberation Army Strategic Support Force, with the goal of becoming the world's leading AI power by 2030 [11]. This represents a fully self-reliant path that does not rely on external standards from the outset, making its starting conditions fundamentally different from the "partial self-reliance within the NATO framework" pursued by France and Turkey. While China faces no constraints from alliance structures, France and Turkey must seek self-reliance within the institutional constraints of NATO's interoperability requirements.
4. Key Variables Shaping Future Developments
The first variable is whether HexaForce and EVREN can demonstrate actual interoperability in official NATO certification procedures such as CWIX. Although Thales passed the test in June [1], this is only a baseline hurdle and is far from official adoption. The timing and scale of when "highly advanced" discussions [1] translate into actual contracts will serve as the first indicator of success or failure.
The second variable is the speed at which the European computing infrastructure gap is closed. If EuroHPC's 75-billion-kroner investment [15] fails to narrow the actual computing power gap, the independent AI-C2 initiatives of France and Turkey may remain confined to software designs, falling into the paradox of relying once again on US cloud and computing resources.
The third variable is how tensions between NATO leadership and individual member states are managed. As shown by General Grynkewich's warning [10], if Europe's efforts to strengthen independent capabilities run directly counter to NATO's integrated command structure, political friction could delay technological cooperation itself. Conversely, if the French and Turkish "overlay" models prove compatible with NATO standards, a third path—one that Japan did not choose—could spread to other allies as a reference model. The success or failure of this path holds implications that extend beyond Europe, as it provides a direct reference point for South Korea's command and control architecture design following the OPCON transition.
III. Recommended Policy Responses
Competition over Independent AI-C2 Systems in Europe and Non-Western Nations: Recommended Policy Responses
1. Comprehensive Assessment
The moves by Thales and the Turkish SSB are not merely product launches by individual entities. They signal that the search for alternatives to US standards within NATO member states has matured to the point of passing official training frameworks like CWIX [1]. This indicates that national policy shifts in emerging technology areas are transitioning into actual procurement stages. At the same time, this trend shows that the axis of interstate competition over AI is diversifying beyond the US-China bilateral dynamic to include European and non-Western actors.
While Japan's adoption of Palantir Maven and the independent paths of France and Turkey appear to be opposite choices, they stem from the same underlying concern: how to secure interoperability while avoiding dependency on foreign standards [3][6]. The risk of "structural dependency due to rising transition costs" [3] highlighted in the Japanese case is precisely the trap that France and Turkey are seeking to avoid. South Korea has yet to define a clear position on this spectrum. With the OPCON transition roadmap underway, choosing a standard for its command and control system is not an issue that can be deferred.
However, it must be clearly recognized that this competition also harbors potential for conflict within Europe. General Grynkewich's warning [10] demonstrates that strengthening independent capabilities at the EU level can create friction with the alliance's command structure. This is why South Korea should approach the issue under the premise that the competition over standards remains fluid, rather than hastily benchmarking the standards of a specific camp.
2. Short-, Medium-, and Long-Term Action Plans
Short-term (within 6 months): The Ministry of National Defense and the Defense Acquisition Program Administration (DAPA) should establish working-level channels to monitor the CWIX test results and NATO certification procedures of Thales's HexaForce. The priority is to document the technical differences between Palantir's standards and the HexaForce-style overlay approach. This will serve as foundational data for AI-C2 clauses to be incorporated into the OPCON transition roadmap. Simultaneously, South Korea needs to review whether the closed design of Turkey's EVREN, which relies on domestic GPU infrastructure [4], can serve as a reference for building a Korean defense AI platform.
Medium-term (1–2 years): South Korea should flesh out a dual-architecture strategy that separates the data layer from the application layer. This aligns with the direction proposed in the previous EAI analysis of the Japanese case [3]. The approach involves maintaining interoperability with US standards at the application layer while securing domestic defense data sovereignty at the data layer. In this process, it is also worth considering linking the overlay technologies of European defense firms like Thales and Helsing with domestic defense companies through cooperative channels. Helsing's demonstration of its AI piloting system, Centaur, defeating human pilots in simulated dogfights [7] shows that the demonstration speed of European companies is faster than expected.
Long-term (3 years and beyond): Any choice of standard will lose its efficacy unless it is accompanied by securing domestic computing infrastructure. The case of EuroHPC investing 75 billion kroner to narrow the computing power gap between Europe and the US/China [15] demonstrates that standard sovereignty cannot be separated from infrastructure investment. South Korea must also treat the acquisition of dedicated defense AI GPUs and computing resources as a separate line item in its defense science and technology budget.
3. Monitoring Indicators and Trigger Points
The first indicator is the point at which consultations between Thales and NATO member states transition into formal contracts. Moving from the "quite advanced" stage of talks to binding procurement contracts will begin to reveal the actual winner of the standards competition in Europe [1].
The second indicator is the actual operational performance and scope of adoption of Türkiye’s EVREN. If the data sovereignty model championed by the SSB is assessed to perform on par with Palantir’s standards, it is highly likely to serve as a benchmark for non-Western countries.
The third indicator is a shift in the attitude of the NATO leadership. Whether discussions on an independent EU-level command and control system continue to expand even after the warning by Commander Greenkeywich [10] serves as a signal to gauge the direction of standards fragmentation within the alliance.
The fourth indicator is the trend in actual transition costs following Japan's adoption of Palantir Maven. Whether the dual roadmap actually leads to the replacement of domestic corporate systems or entrenches structural dependency [3][6] will serve as an empirical case study for South Korea to reference when making similar choices.
4. Summary and Conclusion
The attempts by France and Türkiye to build independent AI-C2 systems are not a direct rejection of US standards, but rather a compromise response aimed at managing dependency risks. In both cases, there is a clear calculation to meet NATO interoperability requirements while avoiding ceding control over data standards [1][4]. South Korea does not need to approach this competitive landscape through a dichotomy of choosing one standard over another. A dual architecture that separates the data layer from the application layer offers a realistic path to simultaneously manage the dual goals of interoperability and technological independence. Codifying this provision into the OPCON transition roadmap has now become an urgent task that can no longer be delayed.
References
[2] [Carnegie Endowment] The Compute Coalition: How to Build the Future of AI in the Free World
[4] [Daily Sabah] Türkiye launches defense-focused AI platform EVREN
[5] [Bruegel] Session 1a: Europe's digital sovereignty: lessons from China's AI quest
[7] [Daily Sabah] German defense firm says AI jet can beat human pilots
[8] [Council on Foreign Relations (CFR)] Why AI’s Biggest Rivals Are Suddenly Calling for Restraint
[10] [Hürriyet Daily News] NATO commander warns EU defense efforts must not rival alliance
[11] [Brookings - TechStream] The PLA’s Strategic Support Force and AI Innovation
[13] [Defense News] Ukraine’s deadly AI training platform for drone tech opens to UK companies
[15] [Børsen] Dane was a top figure in NATO - now he has received 75 billion for a crucial task
[17] [Hürriyet Daily News] US, China agree to open AI incident reporting channel
[18] [Nikkei Asia] NTT, SoftBank eye AI cybersecurity systems that keep data in Japan
[19] [Defense News] EU nails down five defense priority areas to channel common spending
[20] [Le Monde] AI leaders Sam Altman and Dario Amodei acknowledge need for cooperation at UN
[21] [Defense News] French Army wants to keep adversaries guessing on see-through battlefield
[22] [Cyprus Mail] Cyprus rejects Turkish criticism of defence cooperation
[24] [Daily Sabah] EU security protocol must complement NATO, not duplicate it: Kallas
[25] [DW (Deutsche Welle)] Backlash over data centers tests Europe's AI ambitions
*This text is an AI translation of an original written in Korean. Some translations or nuances may be inaccurate.
This report is an in-depth analysis planned by an EAI researcher, grounded in sophisticated AI-assisted research, and finalized by the EAI researcher.
How can I help you?