Inter-NATO-inal: Centres of Excellence in Contestation

[Madara Meļņika is a PhD Researcher in Law at the European University Institute, where her research examines how international security organisations respond to environmental realities.]

In May 2026, the media reported plans for NATO’s AI Centre of Excellence to be established in Rennes, a city already famous for France’s armed forces’ cyber-defence command and other AI institutions. Allegedly, France was the sole candidate to host this Centre of Excellence (hereinafter – the COE), and the path to accreditation seemed clear. However, it did not go as neatly as planned. In June, the United States decided to block the establishment of this COE. This hindrance coincided with a broader US proposal to transform the current COE framework by significantly reducing the number of centres from 30 to 5 – one for each military domain (land, air, sea, cyber, and space).

At this potentially pivotal time for the COE network, this contribution reflects on how the COEs have worked so far, why they matter and what these debates might mean for NATO’s future.

COEs as International Institutions

Notwithstanding the occasional media and academic slips and imprecisions that suggest otherwise, COEs are not NATO ‘think tanks’, simple NATO bodies, or part of the NATO Command Structure. Rather, the Centres, currently close to thirty, are semi-independent military organisations with their own budgets and management structures. Each financed by a narrow list of states, the Centres focus on specific topics, ranging from strategic communication to mountain warfare. Several also work on environmental science and policy. While relatively autonomous, they are affiliated with NATO and thus belong to its ecosystem, which also grants them the right to use its brand. 

The creation of the COE involves two distinct steps. At first, it must be established in accordance with the North Atlantic Treaty, and then acknowledged by designated NATO bodies. The prerequisites for treaty-compliant establishment include sponsorship by at least two NATO nations (one of which is the Framework Nation, hosting the facility, but the rest – Sponsoring Nations, providing the funds and personnel), and at least some military organisation and manning (pp. 10-11). The institution must also have specific objectives, including focus on concept and doctrine development, education and training, and analysis and lessons learned. Once accredited by the Military Committee and endorsed by the NAC, a Centre may describe itself specifically as a NATO COE. Furthermore, for it to become a NATO Military body, the Framework Nation must request activation, which the NAC decides by the silence procedure (i.e., the lack of objections during the prescribed period). Once the Centre is activated as a NATO Military Body, it can receive international status under Article 14 of the Paris Protocol (p. 9). 

Within NATO, the COEs are coordinated by the Allied Command Transformation (ACT) alongside the Supreme Allied Commander Europe (SACEUR). The relationship between participating countries, as well as between the COEs and the ACT, is governed by two separate Memoranda of Understanding (p. 59) – an Operational MOU concluded among the Framework and Sponsoring Nations, and a Functional MOU agreed between those nations and ACT.  HQ SACT Directive 80-03 (HQ SACT Coordination of NATO COEs) elaborates on the daily coordination of NATO COEs within HQ SACT by pillar and cluster. 

Due to their financial and institutional independence, the HQ SACT cannot obligate or compel COEs to act in any specific way. It can only propose specific research streams or expected outcomes from COEs and assist with coordinating collaboration. Overall, requests for support to the COEs may be submitted by ACT, ACO, the Sponsoring Nations, and the wider NATO community. The final decision on the task allocation rests with the COE’s Steering Committee, composed of representatives of the Framework and Sponsoring Nations.

An External Push to Change States’ Actions Within an IO

From the abovementioned, it follows that blocking the accreditation of the AI COE, alongside the request to transform the COE ecosystem, has a limited, yet potentially long-lasting effect on the ecosystem of international organisations within NATO. 

On the one hand, a veto of the accreditation of a COE undercuts the potential cooperation among allies who would have wanted to benefit from the specific institution. By choosing this path instead of contributing to the COE as a Sponsoring Nation, the US expands its influence over the status of an IO of which it is not a member.

Conceivably, such a decision was made due to the US’s own interest in exerting greater control over the AI COE’s actions. Indeed,  while the initial reasons for the US withdrawal of support towards AI COE include the claim that COEs tie up military personnel who, in the US’s view, would contribute more efficiently to NATO’s command structure, the media also reports on the US’s interest in retaining control over AI development. Blocking the accreditation of the current institution might free up time for further debate on who the Framework Nation should be and how the AI COE ought to be governed. An example of a calibrated approach in this direction is NATO’s Counter Intelligence COE, which has two Framework Nations – Poland and Slovakia.

On the other hand, the withholding of the accreditation is lawful under NATO’s core principle – the rule of consensus. The accreditation and other forms of acknowledgement surrounding it elevate the institution to the status of a NATO-associated body, and all allies must align with such a NATO-related decision. The US, like any other ally, has the right to request reconsideration of accreditation regulations as part of the broader NATO strategy, including the role of COEs. Also, the proposal to rethink the COE model is in line with NATO’s deliberations – after all, the current system was only established following the adoption of the 2002 Prague Capability Commitment. 

Finally, the withholding of accreditation does not constitute a complete veto on cooperation among the states devoted to establishing the new club. The impact of such a veto is limited to the second step of the establishment process – the added value of NATO recognition. Accordingly, France has declared its ambition to continue establishing the AI COE even if the accreditation is deemed impossible. 

Practical as it is, this workaround is a painful one, and examining it highlights the role COEs play in NATO. 

The Crux of Approval 

Under NATO’s own process, accreditation opens the door to the North Atlantic Council’s grant of status under the Paris Protocol on the Status of International Military Headquarters, which confers privileges and immunities. An unaccredited COE would be an entity without those specific immunities and would instead require a specific host-state agreement. That, per se, should not cause any issues apart from administrative burden, as the host state already undertakes significant responsibilities under the Operational MOU. Yet, without NATO’s recognition, the overall contribution of the institutions would be significantly weakened.

Without the specific accreditation and with an individual solution in place, the COEs – whether the upcoming AI COE or the existing COEs, which might be reformed if their accreditation is stripped – would be similar to any other international organisation. For their member states, they would remain both a proclamation of values and a means of conveying shared behavioural expectations. Likewise, they would still continue to provide training, education, and other forms of social interaction, which help lay the foundation for harmonising norms and identities within the COE’s member states. The role for the broader NATO network, however, would be reduced. While it cannot be ruled out that the independent COEs might also extend training to the broader community, the complete overlap of competencies and rights between accredited and non-accredited institutions is unlikely.  So, if only five accredited COES remain as per the US’s current plan, the current scale of learning, which includes 150 NATO-approved courses and 20,000 personnel trained annually, would certainly need to be toned down.  

The same applies to the development of standards and concepts in the specific fields. The strategic losses caused by the displacement of the current COEs would include advancements in research on the military impacts of climate change, strategic communications, civil-military cooperation, mountain warfare, stability policing, energy security, and other areas. While some of these might be incorporated into the “five-domain-COEs” proposal, more sensitive or cross-cutting topics would slowly dissipate. 

The broader cost for NATO, however, would be the loss of ubiquitous visibility and soft power.

What Stands to Be Lost

In the limited literature, COEs have been described as forums for information exchange, matchmakers between private institutions and state militaries, and outreach tools for NATO in developing various relationships. In their bridging role, the COEs cooperate with governmental and non-governmental organisations, think tanks, academia, and industry. They have a political role at both the national and personal levels. 

For new NATO member states, the COEs serve as a pathway to share their expertise and have their voices and input in shaping NATO’s doctrine and protocols (p. 569). In this way, the epistemic equality of member states is further strengthened by elevating each one’s specific skill set and specialisation. Simultaneously, the establishment and presence of the COE in the member states also facilitates the visibility of NATO’s brand beyond military conceptions. Through their use of flags and logos, the organisation of conferences, the establishment of academic and NGO cooperation, and media presence, the COEs strengthen the territorial connectivity of each corner of the alliance to the broader political identity. 

The political ripples stretch far beyond the COE’s members’ circle as well as NATO’s allies. COEs are a tool for connecting the alliance with the external partners, both materially and emotionally.  Joining NATO’s COE has been seen as evidence of the deepening of relations between South Korea and NATO. The military-neutral Ireland and Austria are Contributing Participants in COEs. Facilitation of outreach also occurs inter-organisationally. For instance, the Civil-Military Cooperation COE serves as the European Union Discipline Leader for CIMIC. 

Finally, the COEs are often shaped by national experts who are deployed to the institution for a limited time. These might be both military and civilian personnel. Due to their agility, national specialisations, and in some cases a stronger focus specifically on civilian expertise, the COEs contribute to NATO discussions with insights that are external to the institutional bureaucracy. And this is also an aspect in which the US’s criticism of the COEs ‘tying up military personnel’ falls short of being comprehensive. COEs such as CCASCOE, CCDCOE, and StratCom COE are currently led by civilian directors and mostly employ civilian researchers. In fact, the civilian nature of viewpoints also adds to the general reminder of NATO’s political facet.

So, COEs are relatively small, yet noteworthy tools for cooperation, research, strengthening inter-state relations, signalling specific priorities, and developing terminology, concepts and doctrine that keep pace with operational reality. On top of that, these institutions “humanise” NATO, bringing it closer to the inhabitants of allies’ cities and showcasing the alliance’s non-military face. And all of this happens at no extra cost to NATO as an institution or the allies which do not belong to the specific COEs. Deleting the “NATO” name before the COE title and stripping them of the rights to use NATO’s visuals and brand would cost the alliance not only the knowledge, training and agility, but also low-cost visibility, a civilian face, and its connection to the local societies of the member states, which have been proudly hosting the COEs so far. Now these centres would disappear in the sea of other organisations. 

Conclusion

To summarise, from a legal perspective, no external country can force another country to create or dissolve an international organisation that the latter would be a member of. That also applies to Centres of Excellence, as these research and training institutions rely on sovereign decisions regarding funding and priorities. However, consensus and agreement remain at the core of NATO, and without them, accreditation and acknowledgement cannot be granted. Accordingly, the possible new centres remain in a political limbo. 

 If the accreditation is vetoed or if the COE system is fully reconsidered, any state can continue to create its own club of cooperating partners. Such cooperation, without knowledge transfer to the broader NATO ecosystem, would cement ties among the member states of the specific international organisation, but would distort the goals of broader harmonisation. 

Reducing the number of specialised, decentralised COEs would also remove NATO’s name and logo from the buildings across the alliance’s territory, which currently proudly host the centres at no cost to NATO’s budget. This emotional presence, reiterated through media appearances, specific collaborations with academia and partner states, and, at times, state-focused research, reduces the distance between the general population, especially the one of the smaller or younger member states, and the alliance. Thus, the deconstruction of the COEs could ultimately weaken NATO by eroding both its real-world epistemic authority and its societal embeddedness. It can only be hoped that the current misunderstandings stem from the sensitivity of the AI topic and will be resolved in a manner that strengthens the alliance in the long term.