The Competition over Palantir Alternatives for AI-C2 Systems and South Korea's Strategy during the OPCON Transition Period
Executive Summary
France's Thales HexaForce, Turkey's EVREN, and Australia's sandbox isolation approach represent distinct pathways of responding to the ontological dependence imposed by Palantir's Maven system. The common denominator among these three cases is the shared understanding that adopting AI-C2 standards is not merely about software procurement, but rather entails the transfer of operational grammar and overall data sovereignty. Japan's case—where the initial adoption of Maven shows signs of delaying the transition to domestic alternatives due to rising switching costs—supports the reality of this concern. As South Korea faces the intersection of its unique wartime operational control (OPCON) transition timeline and the global competition over AI-C2 standards, it must pursue a strategy that simultaneously secures ROK-US interoperability and operational autonomy through a dual architecture that separates the data layer from the application layer. Establishing a dedicated technical evaluation team led by the Ministry of National Defense (MND) and the Defense Acquisition Program Administration (DAPA), alongside securing dedicated defense computing infrastructure, are proposed as short- to medium-term priorities.
I. Issue Analysis
Issue Analysis: Europe and Australia's Search for Palantir Alternatives in AI-C2 Systems and South Korea's Response Strategy
1. Background and Progress
The unveiling of HexaForce by France's Thales was not an impromptu decision. Thales had already tested the system at NATO's Coalition Warrior Interoperability eXercise (CWIX) held in Poland last June [1][3]. The official launch took place approximately three months after the exercise, a decision made after substantial consultations with NATO member states had already progressed [3]. Patrick Moreau, Vice President of Multi-Domain Operations at Thales, stated that the company is engaged in "highly advanced" discussions with several NATO countries [3]. Thales is positioning HexaForce as a multi-domain overlay that can be integrated with legacy and domestic systems from the strategic to the tactical level [3].
This trend aligns with France's broader defense policy of "sovereign AI." The French Air and Space Force, in collaboration with the French Defense Procurement Agency (DGA) and the defense AI agency, plans to conduct a joint flight of a manned fighter jet and a combat drone in 2028 [4]. This project, named "Hyperion," aims to develop an open modular architecture and sovereign AI technologies to secure government control without relying on foreign companies for core systems [4]. This demonstrates that HexaForce is not an isolated product, but rather a result of the broader drive toward technological self-reliance within the French defense ecosystem.
The background behind the launch of Turkey's EVREN is of a different nature. Launched last weekend by Turkey's Presidency of Defense Industries (SSB), this platform targets a broad user base, including defense contractors, tech firms, academia, entrepreneurs, and students [1]. The key emphasis in the SSB's announcement was data sovereignty rather than technical performance. All prompts, responses, and operational data on the platform are processed entirely within Turkey's domestic high-performance GPU infrastructure [1]. The design principle explicitly states that the transfer of sensitive data to foreign clouds is strictly blocked [1]. This is an extension of the defense self-reliance policy that Turkey has pursued since its exclusion from the F-35 program [3].
The common background running through both cases is a wariness of the ontology standards and CJADC2 concepts led by the US firm Palantir [3]. The case that most clearly illustrates the counter-reaction to this wariness is Japan. While drafting a record-high budget of 8.9 trillion yen for fiscal year 2027, Japan's Ministry of Defense decided to prioritize the introduction of Palantir's Maven Smart System for the Japan Joint Operations Command (JJOC) [6]. Although Japan presented a dual roadmap to replace it once domestic systems are developed, analysts point to the risk of structural dependence due to rising switching costs [6]. The pursuit of independent systems by Europe and Turkey can be read as a preemptive response to such risks of dependence [3].
2. Current Situation
Australia has taken a different path from Europe and Turkey. The Australian Defence Force adopted a method of testing Palantir's Maven system in a sandbox environment isolated from the defense network, with its AI functions turned off. This is a compromise "brake" strategy aimed at maintaining control over the data layer without completely abandoning interoperability. The Australian Strategic Policy Institute (ASPI) held an AI masterclass in Canberra last month, gathering defense, foreign affairs, development, and industry experts to focus on this issue [9]. The local perspective defining Australia as an "AI middle power" leans toward seeking a third path—neither fully integrating into US big tech standards nor being completely excluded from them [9]. Indeed, OpenAI announced its intention to use Australia as a testbed for countering "rogue agents," a decision based on the assessment that the Australian government holds a leading position in technology policy [10].
In Europe, beyond France, Germany's Helsing announced that its CA-1 Europa unmanned combat aerial vehicle (UCAV), equipped with its proprietary AI pilot "Centaur," repeatedly defeated human pilots in simulated dogfights [12]. The French Army is experimenting with reorganizing tank units into smaller formations in a drone-dominated battlefield [13], while the French Future Combat Command is exploring ways to strengthen deception and electronic warfare capabilities to counter battlefield transparency [14]. At the EU level, the 27 member states approved a joint investment of 1.5 billion euros under the European Defence Industry Programme (EDIP) targeting five priority areas, including drones/counter-drones and air/missile defense [16]. However, this is the cumulative result of individual corporate and member-state initiatives, rather than a unified, official EU-wide decision on AI-C2 standards.
Cases highlighting the risks of Palantir's Maven system are also being raised by local media. In February, the Moroccan outlet TelQuel covered an incident in which an elementary school in Minab, Iran, was destroyed by a Tomahawk missile, killing over 150 people, including 123 children. Citing a Bloomberg investigation, the outlet reported that outdated data, diminished human safeguards, compressed timelines, and overreliance on the Maven system shaped this targeting process [11]. Adnane Kaab, a strategic analyst and former Moroccan Air Force officer, assessed that AI did not make the decision itself, but rather altered the speed and structure of the decision-making process [11]. This case is being cited as a background rationale for why Europe, Turkey, and Australia remain wary of fully integrating into Palantir's standards.
3. Key Actors and Positions
Thales aims to establish HexaForce in the market as an overlay compatible with legacy systems through individual consultations with NATO member states [3]. The French government, through its Air and Space Force, the DGA, and the defense AI agency, supports the sovereign AI policy to reduce dependence on foreign firms at the national strategic level [4]. Turkey's SSB prioritizes data sovereignty through EVREN, extending its post-F-35 defense self-reliance policy into the AI domain [1][3]. The Australian Department of Defence and ASPI seek to manage risks through a sandbox approach that neither fully rejects nor fully accepts Palantir's standards, thereby building a position as an "AI middle power" [9]. Japan's Ministry of Defense opted for a pragmatic path of initial integration into US standards, but this has invited concerns over structural dependence, contrasting sharply with the strategies of Europe, Turkey, and Australia [6]. The US firm Palantir continues to expand the military service capabilities of its Maven Smart System, seeking to preemptively establish its standards [7]. While NATO validates the interoperability of Thales and Turkish systems through its official CWIX exercise framework, it internally grapples with tensions between the alliance's command structure and individual member states' efforts to strengthen independent capabilities [3].
4. Key Issues
The core issue is whether the dual goals of avoiding standard dependence and maintaining interoperability can coexist. From the perspective of emerging technology developments, the approaches of Europe, Turkey, and Australia strike a balance at different points. While Thales and Turkey have brought their independent platforms close to the procurement stage by passing them through NATO's official validation framework, Australia opted for a compromise of adopting the US system while retaining control. In the domain of emerging and non-traditional security, the Moroccan case reconfirmed that malfunctions or overreliance on AI-C2 systems can lead to actual loss of human life [11]. At the alliance and multilateral security level, as warnings from top NATO commanders suggest, individual member states' moves to strengthen independent capabilities could conflict with the integration of the alliance's overall command structure [3]. With South Korea's OPCON transition roadmap underway, a dual-architecture strategy that separates the data layer from the application layer, along with securing dedicated defense AI computing infrastructure, will likely be key to balancing interoperability and autonomy in the future [3][6].
II. In-Depth Analysis
In-Depth Analysis: Europe and Australia's Search for Palantir Alternatives in AI-C2 Systems and South Korea's Response Strategy
1. Root Cause Analysis
The root cause of this phenomenon lies in the fact that an AI-C2 system is not merely a matter of software procurement. Palantir's Maven Smart System integrates data collection methods, target identification algorithms, and command decision flows into a single ontology [6]. The moment a nation's military adopts this ontology, it simultaneously adopts the US military's data structure and operational grammar. An op-ed in the Moroccan outlet TelQuel sharply identifies this issue. Analyzing the Tomahawk misstrike in Minab, Iran, last February, it points out that AI did not make the decision directly [11]. Instead, "AI changed the speed and structure of the decision" [11]. The reconstruction shows that a combination of outdated data, diminished human safeguards, compressed decision-making time, and overreliance on the Maven system led to the tragedy [11]. This demonstrates that a country adopting a standard is not simply borrowing technology, but is also importing the risk structures embedded within it.
The Turkish case reveals a cause on a different level. What the SSB prioritized in the design of EVREN was not algorithmic performance, but the physical location of data [1]. The principle that all prompt and response data must be processed entirely within domestic GPU infrastructure to prevent it from leaving the country is a lesson learned from its past exclusion from the F-35 program. Turkey's assessment that US export controls or program exclusions could be repeated at any time is reflected in the platform's design itself. In France's case, the cause is more rooted in industrial policy. The Hyperion project pursued by the French Air and Space Force explicitly states its goal as "securing government control without relying on foreign companies for core systems" [4]. This is the result of overlapping factors: the industrial interests of its domestic defense contractor Thales, the procurement sovereignty logic of the DGA, and the Air Force's demand for operational autonomy.
2. Structural Context
In terms of security architecture, this competition reveals tensions within the NATO command structure. The statement by the Thales Vice President that discussions are "highly advanced with several NATO countries" [3] proves that CJADC2 has not yet been established as a single NATO-wide standard. Passing through NATO's official CWIX exercise framework means that HexaForce and EVREN have acquired official status within the alliance as alternatives to the US standard. However, the fact that warnings are being issued by top NATO commanders regarding the tension between strengthening independent capabilities and maintaining the integration of the alliance's command structure [3] suggests that this competition is ongoing and unresolved. This is where the pursuit of sovereignty by individual European nations structurally clashes with NATO's interoperability requirements.
In terms of industrial structure, the EU's move toward joint defense procurement serves as a background. Last month, the 27 EU member states designated five priority projects, including drone/counter-drone systems and air/missile defense, for joint investment under the European Defence Industry Programme (EDIP) [16]. While this is an attempt to avoid redundant investment within the region and increase joint spending, it also carries industrial policy implications aimed at reducing procurement dependence on non-European firms. Cases like Thales's HexaForce and German firm Helsing's AI fighter pilot "Centaur" [12] should be viewed as products of this pressure for self-reliance within the European defense ecosystem. Helsing's announcement that its AI piloting system repeatedly defeated experienced pilots in simulated dogfights [12] demonstrates that European defense firms are not mere followers, but are determined to compete with independent technological capabilities.
In terms of political structure, each country's choice is proportional to its level of trust in its relationship with the US. Australia maintains a relatively high level of trust within the AUKUS framework with the US. Consequently, the Australian military's choice is compromise rather than rejection. Adopting the Maven system itself but testing it with AI functions turned off in a sandbox isolated from the defense network is a middle path aimed at controlling data sovereignty risks without sacrificing interoperability. Conversely, Turkey, having experienced actual damage through its exclusion from the F-35 program, chose a more radical path of self-reliance. France, given its strategic status as a permanent member of the UN Security Council and an independent nuclear power, possessed political legitimacy for its sovereign AI policy from the outset. The differing choices of these three nations stem not from technological gaps, but from differences in their trust structures with the US and their past experiences.
3. Historical Precedents and Comparison of Similar Cases
This issue is an extension of past standard competitions in weapon systems. During the Cold War, conflicts repeatedly arose during NATO's introduction of common communication protocols between nations that accepted US standards and those that adhered to independent systems. The difference today is that, unlike past communication or ammunition standards, AI-C2 standards subject the data itself to the algorithmic logic of another country. The switching costs are far higher than replacing communication equipment, and once a data ontology is established, it is difficult to decouple.
A more direct precedent is Japan's choice. In its fiscal year 2027 budget, Japan's Ministry of Defense decided to prioritize the introduction of Palantir's Maven for the JJOC, while presenting a dual roadmap to replace it once domestic systems are developed [6]. However, previous EAI analyses have pointed out the possibility that structural dependence could become entrenched due to rising switching costs [6]. The choices of Europe and Turkey can be interpreted as a reaction based on the observation that Japan's "adopt first, replace later" path is difficult to transition into an actual replacement phase [3]. In other words, after observing the path Japan took, Europe and Turkey chose to preemptively avoid the risks of that path. Australia's sandbox strategy represents a third way in the middle—adopting the system but fundamentally blocking the path to dependence.
The Minab misstrike in Iran also serves as a precedent [11]. This incident led to analyses suggesting that the issue was not a defect in the Maven system itself, but rather an overreliance on the command decision-making structure that Maven generates [11]. The message this interpretation sends to policymakers in Europe and Turkey is clear: adopting a standard is not merely choosing a tool, but also accepting the decision-making speed and accountability structures that the tool creates.
4. Key Variables Shaping Future Developments
The first variable is whether a standard is finalized at the NATO level. While passing the CWIX granted official status, it does not equate to a single standard status. As warnings from top NATO commanders suggest, if the coexistence of multiple standards within Europe is prolonged, the risk of fragmentation in alliance command and control will increase [3]. The way this fragmentation is resolved will determine the de facto status of CJADC2.
The second variable is the validation of real-world performance. Helsing's announcement that its AI fighter pilot defeated experienced pilots in simulations [12] and the French Air and Space Force's planned 2028 manned-unmanned joint flight experiment [4] have not yet been proven with real-world combat data. If European and Turkish alternative systems fail to demonstrate reliability comparable to Palantir's system in actual operational environments, the political justification for their self-reliance policies will inevitably weaken.
The third variable is the competition to secure computing infrastructure. An analysis by the Carnegie Endowment for International Peace points out that the speed of building AI chip clusters determines the decisive advantage between nations [8]. Just as Turkey faces the challenge of securing high-performance GPU infrastructure domestically while championing data sovereignty [1], the sustainability of an independent AI-C2 policy ultimately depends on the ability to secure semiconductor and computing infrastructure. This is directly linked to the key monitoring point of emerging technology developments, raising the possibility that the outcome of the standard competition will be decided by hardware supply chains rather than algorithms.
The fourth variable is the US policy response. Whether the experiments of Europe, Turkey, and Australia expand or become isolated depends on the incentives or pressures that Palantir and the US Department of Defense employ against allies' moves to diverge. The existence of allies like Japan that adopted the system early provides the US with grounds to claim the "normality of the standard," while simultaneously serving as a cautionary tale for other allies [6].
The aforementioned structure is intertwined with South Korea's discussions on the OPCON transition. As the combined defense posture and OPCON transition are addressed as core agendas in the alliance and multilateral security domains, whichever path South Korea chooses is highly likely to be a variation of one of three precedents: Australia's sandbox, Turkey's complete self-reliance, or Japan's initial adoption. A dual-architecture strategy that separates the data layer from the application layer needs to be reviewed as a compromise among these three precedents [6].
III. Recommended Response and Action Plan
Comprehensive Recommended Response and Action Plan
1. Comprehensive Assessment and Recommended Response
The cases of Europe, Turkey, and Australia demonstrate three distinct solutions. France chose to establish an independent standard led by Thales, securing legitimacy by passing NATO's official exercise framework [3]. Turkey prioritized data sovereignty, pursuing domestic infrastructure self-sufficiency [1]. Australia opted for a compromise, not rejecting the standard but retaining control. The common denominator among these three pathways is wariness of the structural dependence that arises when uncritically adopting Palantir's ontology [6][3]. Japan's case of prioritizing Maven's adoption shows that this wariness is not unfounded. Despite its dual roadmap, analysts have already pointed out the possibility that the replacement with domestic technology could be delayed due to rising switching costs [6].
The implications for South Korea are clear. Its unique institutional timeline of the OPCON transition and the international trend of AI-C2 standard competition are unfolding simultaneously. The OPCON transition is premised on expanding operational autonomy. However, if the command and control system that forms the foundation of that autonomy is built in a way that relies entirely on the ontology of a specific US corporation, a direct clash between institutional transition and technological dependence will occur. This means that the demand for interoperability in the ROK-US combined defense posture within the alliance and multilateral security domain will stand in tension with the demand to secure autonomy in the military use of AI within the emerging and non-traditional security domain.
The core of the recommended response is to adapt Australia's "brake" strategy to the South Korean context. South Korea needs to adopt a dual architecture that separates the data layer from the application layer [6]. While maintaining the level of interoperability required for ROK-US combined operations at the application layer, the data layer must be kept on domestic infrastructure physically isolated from the defense network. This represents a combination of the data sovereignty principle demonstrated by Turkey's EVREN and Australia's sandbox isolation method [1].
2. Short-, Medium-, and Long-Term Action Plans
In the short term, the Ministry of National Defense (MND) and the Defense Acquisition Program Administration (DAPA) should jointly establish a dedicated technical evaluation team for AI-C2 adoption. This team's mission should be to design an Australian-style sandbox testing method tailored to the domestic environment before deciding whether to adopt Palantir's Maven system. A phase to validate only data flows and interface structures without activating AI functions must be explicitly included in the OPCON transition roadmap. Simultaneously, the team needs to conduct a comparative review of the technical specifications of Thales's HexaForce and Turkey's EVREN to explore the applicability of an overlay approach that avoids dependence on a specific vendor [3][1].
In the medium term, the task is to secure dedicated computing infrastructure for defense AI. Turkey's establishment of domestic GPU infrastructure to fundamentally block dependence on foreign clouds [1] serves as a direct reference for South Korea's defense data center investment direction. The Ministry of Science and ICT (MSIT) and the MND should cooperate to secure separate high-performance computing resources dedicated to defense, while reviewing measures to minimize linkage with commercial clouds. Furthermore, within the framework of ROK-US-Japan trilateral security cooperation, South Korea needs to continuously track the progress of Japan's Maven adoption to observe how switching costs and dependence risks actually materialize [6].
In the long term, the government should support the development of independent AI-C2 systems by domestic defense contractors from an industrial policy perspective. The state-led pathways of technological self-reliance pursued by France through Thales and Turkey through the SSB can align with South Korea's defense export strategy. However, a weakness for South Korea in this process is the lack of a multilateral validation framework like NATO's CWIX. Diplomatic efforts must be pursued in parallel to establish separate interoperability validation tracks between the ROK and the US, or among the ROK, US, and Japan, ensuring that the independent system is not an isolated technology but is recognized as an accepted standard within the alliance.
3. Monitoring Indicators and Trigger Points
The first indicator is the operational progress of the Maven system by the Japan Self-Defense Forces. Whether the actual switching costs in the JJOC and the timeline for replacement with domestic technology proceed as originally planned will serve as a benchmark for South Korea's decision-making [6]. The second is how the tension between independent capabilities and the alliance command structure, raised by top NATO commanders, is resolved [3]. If this tension intensifies and NATO moves to restrict alternative systems like HexaForce or EVREN, South Korea's strategy for building an independent system will also require review. The third is the point at which Australia's sandbox testing results actually lead to integration with its defense network. At this juncture, the level to which Australia activates AI functions will serve as a direct reference point for South Korea's dual-architecture design.
Three trigger points can be established. First, when the US officially requests the adoption of a specific AI-C2 standard for reasons of ROK-US combined operational interoperability. Second, when the standardization of command and control systems is officially placed on the agenda during the OPCON transition negotiation process. Third, when domestic defense contractors declare the development of an independent AI-C2 system, or when a state-led platform initiative similar to those of Thales or the SSB is announced. If any of these three points materialize, an emergency review system involving the MND, the Ministry of Foreign Affairs (MOFA), and DAPA must be activated.
4. Summary and Conclusion
The search for alternatives by Europe, Turkey, and Australia collectively demonstrates that passive acceptance of the US Palantir standard is not the only viable option [3][1]. Each country is finding a compromise that avoids standard dependence while preserving alliance interoperability in accordance with its own institutional conditions. Given that South Korea faces the unique variable of the OPCON transition, it cannot replicate these cases exactly. A dual architecture separating the data layer from the application layer, securing dedicated defense computing infrastructure, and continuously tracking the Japanese case represent the most realistic pillars of response at present. Explicitly linking these three pillars with the OPCON transition roadmap is the key task for the next phase [6].
References
[1] [Daily Sabah] Türkiye launches defense-focused AI platform EVREN
[2] [Council on Foreign Relations (CFR)] Why AI’s Biggest Rivals Are Suddenly Calling for Restraint
[4] [Defense News] French Air Force plans loyal wingman flight in 2028 in sovereign AI push
[5] [Brookings - TechStream] The PLA’s Strategic Support Force and AI Innovation
[7] [Palantir] Palantir Expands Maven Smart System AI/ML Capabilities to Military Services
[8] [Carnegie Endowment] The Compute Coalition: How to Build the Future of AI in the Free World
[9] [The Diplomat] Australia as an AI Middle Power
[11] [TelQuel] [Tribune] Guerre et IA : quand la machine change la chaîne de décision
[12] [Daily Sabah] German defense firm says AI jet can beat human pilots
[13] [Defense News] France experiments with smaller tank units to evade drone attacks
[14] [Defense News] French Army wants to keep adversaries guessing on see-through battlefield
[16] [Defense News] EU nails down five defense priority areas to channel common spending
[17] [Wired] These AI Experts Want to Do High-Stakes Research Out in the Open
*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?