Summary Yes, a frontier AI priority project can absolutely span more than three Member States. Under Article 8(b) of the proposed Cloud and AI Development Act (CADA), the requirement for participation from "at least three Member States" is a strict minimum threshold, not a maximum cap. The proposal explicitly encourages "broad participation" to pool greater computing resources and strengthen the Union's strategic autonomy. There is no upper limit on the number of participating Member States; a project involving four, ten, or all 27 Member States is fully compliant and strategically advantageous.

Detail

The Cloud and AI Development Act (CADA), as proposed in COM(2026) 502 final, establishes a comprehensive framework to bolster Europe's technological sovereignty. A central pillar of this framework is the support for "frontier AI priority projects." These are defined as pioneering initiatives designed to scale up frontier AI technologiesβ€”AI models or systems that "approach, reach or exceed the current state of the art" as defined in Article 2(4).

A frequent point of confusion for consortium leaders and public-sector bodies concerns the geographic scope of these projects. Specifically, does the "three Member State" rule act as a ceiling? The answer is a definitive no. The legal text and the accompanying explanatory memorandum clarify that the threshold is a floor designed to ensure cross-border relevance, not a restriction on ambition.

The "At Least Three" Rule: A Minimum, Not a Cap

The core legal requirement is found in Article 8(b) of the proposal. This article sets out the criteria the Commission must apply when recognizing a project as a frontier AI priority project. It states that the project must be:

"undertaken by a European digital infrastructure consortium established pursuant Decision (EU) 2022/2481 or another legal entity eligible for funding under Union law and it involves the participation of at least three Member States."

The phrasing "at least three" is legally significant. In EU legislative drafting, this terminology establishes a lower bound. It mandates that a project cannot be recognized if it involves only one or two Member States, as such a scope would fail to demonstrate the necessary Union-wide strategic relevance. However, the text contains no language suggesting a maximum limit. There is no provision in Article 8, nor in the surrounding articles, that prohibits a project from involving four, five, or all 27 Member States.

Therefore, a consortium comprising ten Member States is not only compliant but fully aligned with the literal text of the regulation. The requirement is satisfied once the threshold of three is met; any participation beyond that remains permissible and is not subject to additional legal hurdles regarding the number of states.

Policy Intent: Encouraging Broad Participation

While the legal text sets the minimum, the policy intent behind CADA strongly favors broader collaboration. The Explanatory Memorandum, specifically Recital 34, provides crucial context for interpreting Article 8.

Recital 34 notes that "given the unprecedented scale of resources required for frontier AI development, it is necessary to set criteria for the designation of a project as a frontier AI priority project." It acknowledges the "technical complexity and capital-intensive nature" of these projects, stating that they "require a collaborative approach at Union level."

Crucially, the recital continues: "It is therefore appropriate to require them to involve broad participation from entities across the Union." This language signals that while three Member States meet the legal baseline for eligibility, the Commission's strategic preference is for projects that maximize cross-border integration. A project spanning a wider geographic area is better positioned to address the "market failures of imperfect information" and the "limited and geographically concentrated availability of computing capacity" identified as key problems in the proposal.

The Strategic Advantage: Pooling Compute Resources

The primary practical benefit of expanding the number of participating Member States beyond the minimum is the ability to pool and match computing resources. This mechanism is detailed in Article 9 of the proposal.

Article 9(1) mandates that "The Union and the Member States shall ensure that sufficient AI computing resources from their compute capacities are allocated to support the development of frontier AI priority projects."

More importantly, Article 9(2) establishes a matching mechanism: "The Union shall at least match the AI computing resources contributed by Member States to frontier AI priority projects to the extent that sufficient AI computing capacity is available within the Union's share of European high performance computing access time."

This creates a direct incentive for broader participation. If a project involves only three Member States, the Union matches the compute contributed by those three. If a project involves ten Member States, the Union matches the compute contributed by all ten. Consequently, a consortium with more participating Member States can potentially access a significantly larger pool of matched compute capacity from the European High Performance Computing (EuroHPC) infrastructure.

For frontier AI, which requires massive computational power for training and inference, this scaling effect is critical. By encouraging more Member States to join, the proposal effectively multiplies the available resources for the project, directly supporting the objective of "increasing the Union's resilience and strategic autonomy" outlined in Article 1(3).

Eligibility and Consortium Structure

To qualify for recognition, the project must be undertaken by a "European digital infrastructure consortium" (EDIC) established under Decision (EU) 2022/2481, or "another legal entity eligible for funding under Union law." This flexibility allows for various consortium structures, provided they can demonstrate the participation of the requisite Member States.

The Commission retains the discretion to recognize projects via a formal decision, as outlined in Article 8. While the legal criteria are binary (met or not met), the evaluation of the project's strategic value may favor those with deeper integration across a larger number of Member States. A project demonstrating broad participation directly addresses the proposal's goal of reducing dependencies on third-country technologies by leveraging diverse European capabilities and creating a "robust financial and talent flywheel."

What this means for you

For public-sector procurement officers, research consortium leaders, and potential applicants, understanding this distinction is vital for strategic planning and maximizing project impact.

  1. Design for Scale, Not Minimums: When drafting calls for expressions of interest or consortium agreements, do not artificially limit the number of participating countries to three. Instead, structure your governance, financial contribution models, and data-sharing agreements to accommodate a larger number of partners. A broader consortium increases the project's resilience, resource base, and political weight.
  2. Leverage the Compute Matching Mechanism: Remember that under Article 9, the EU matches compute contributions. By onboarding more Member States, you increase the total volume of committed national compute, which in turn triggers a larger matching contribution from the Union. This is a powerful financial and technical incentive for expanding your consortium beyond the minimum requirement.
  3. Highlight Cross-Border Impact in Applications: In your application to the Commission, explicitly emphasize the breadth of your consortium. Reference Recital 34's call for "broad participation" to demonstrate that your project aligns with the EU's strategic goals for technological sovereignty and resilience. Show how your project leverages the "unprecedented scale of resources" mentioned in the recitals.
  4. Ensure Eligibility of All Partners: Ensure that all participating entities from each Member State are eligible for funding under Union law. The requirement is for the participation of Member States, which is typically fulfilled through their designated research organizations, universities, or public infrastructure bodies. Verify that your legal entity structure (EDIC or other) can validly represent this multi-state participation.

Common misconceptions

Misconception 1: "Three is the maximum number of partners." Some applicants mistakenly interpret "at least three" as a strict limit, fearing that adding more partners might complicate the project or exceed a perceived cap. This is incorrect. The regulation imposes no upper limit on the number of participating Member States. The phrase "at least" defines the floor, not the ceiling.

Misconception 2: "Only EDICs can apply." While Article 8(b) mentions European digital infrastructure consortiums (EDICs), it explicitly allows for "another legal entity eligible for funding under Union law." This means that other consortia structures, such as joint undertakings or ad-hoc alliances, may also qualify if they meet the eligibility criteria for EU funding and involve the required Member States.

Misconception 3: "Frontier AI projects are only for tech companies." The proposal emphasizes the role of public sector bodies and Union entities. Article 8 requires participation from Member States, and Article 9 involves the matching of public compute resources. Therefore, these projects are inherently public-private partnerships or inter-governmental collaborations, not solely commercial ventures. The "broad participation" encouraged in Recital 34 often involves national research infrastructures and public bodies.

Misconception 4: "More states mean more bureaucracy." While managing a larger consortium requires robust governance, the proposal's matching mechanism in Article 9 is designed to reward scale. The administrative effort of coordinating more Member States is offset by the significant increase in available compute resources and the strategic alignment with EU sovereignty goals, which can facilitate smoother access to Union-level support.

Official sources

Related

This is general information about a draft EU regulation, not legal advice.