Summary Under the proposed Cloud and AI Development Act (CADA), the European Commission holds the exclusive authority to designate projects as "frontier AI priority projects." This designation is not automatic; it is granted via a formal Commission decision following open calls for expressions of interest. To qualify, a project must be pioneering, undertaken by a European digital infrastructure consortium (EDIC) or eligible entity, involve at least three Member States, and require those states to pool computing resources. Once recognised, the Union commits to matching these resources from its European high-performance computing (EuroHPC) capacity.
Detail
The proposed Cloud and AI Development Act (CADA), COM(2026) 502 final, establishes a targeted mechanism to accelerate the development of "frontier AI"βdefined in Article 2(4) as AI models or systems that approach, reach, or exceed the current state of the art. Recognising that such technologies require unprecedented scale and cross-border collaboration, the proposal centralises the designation power in the European Commission to ensure strategic alignment and resource efficiency.
The Decision-Maker: The European Commission
The authority to recognise a project as a "frontier AI priority project" rests solely with the European Commission. Article 8 explicitly states: "The Commission may, by means of a decision, recognise as frontier AI priority projects, projects selected through open calls for expression of interest..."
This provision clarifies that:
- No Unilateral Designation: Individual Member States, private companies, or research institutes cannot self-declare a project as a priority.
- Formal Procedure: The recognition is a legal act ("by means of a decision"), creating a binding status that triggers specific obligations and support mechanisms under the Regulation.
- Strategic Focus: The decision is not discretionary in a vacuum; it is strictly bound by the criteria set out in Article 8 and the strategic objectives of Grand Challenge 3 (Frontier AI) outlined in Annex I.
The Selection Process: Open Calls for Expression of Interest
The proposal mandates a transparent and competitive selection process to avoid favouritism and ensure the best projects are chosen. Article 8 requires that projects be "selected through open calls for expression of interest."
This mechanism serves several purposes:
- Transparency: It ensures that all eligible entities across the Union have equal opportunity to apply.
- Strategic Alignment: The calls will be tailored to address specific gaps in frontier AI development, such as architectural design, multimodal capabilities, or agentic systems, as detailed in Annex I.
- Merit-Based: Selection is contingent on fulfilling the cumulative criteria of Article 8, rather than political negotiation alone.
Mandatory Criteria for Recognition (Article 8)
For the Commission to issue a positive decision, a project must satisfy three cumulative criteria laid out in Article 8. Failure to meet any single criterion disqualifies the project.
1. Pioneering Nature and Strategic Focus
The project must be a "pioneering project, focused on the support and scaling-up of frontier AI technologies."
- Scope: This targets the development of next-generation multimodal models and systems that push the boundaries of current algorithmic capabilities.
- Alignment: As noted in Annex I, this includes advancing capabilities in advanced reasoning, cross-modal understanding, and agentic capabilities, as well as investigating novel approaches to model efficiency and cognitive modelling.
2. Eligible Legal Structure and Multi-Member State Participation
The project must be undertaken by a specific type of legal entity and involve broad participation:
- Legal Entity: The project must be undertaken by a European digital infrastructure consortium (EDIC) established pursuant to Decision (EU) 2022/2481, or "another legal entity eligible for funding under Union law."
- The "Three-State" Rule: Crucially, the project must "involve the participation of at least three Member States." This requirement is designed to prevent national silos and ensure that frontier AI development is a genuinely European endeavour, fostering cross-border integration of research and infrastructure.
3. Pooling of Resources
Recognition is conditional on concrete resource commitments. The participating Member States must "pool computing time and other relevant resources to support the implementation of the designated project."
- Tangible Commitment: This is not merely a political declaration; it requires Member States to commit actual computing capacity and other resources to the project.
- Collaborative Implementation: The pooling mechanism ensures that the project benefits from a combined national effort, creating a critical mass of resources necessary for frontier AI training and deployment.
The Role of Member States and the Union in Resource Allocation
While the Commission decides which projects are priority projects, Article 9 dictates how resources are allocated once a project is recognised. This creates a "matching" incentive structure:
- Member State Obligation: Under Article 9(1), the Union and Member States must ensure that "sufficient AI computing resources from their compute capacities are allocated to support the development of frontier AI priority projects."
- Union Matching: 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."
- Capacity Limits: This matching is subject to the "limits of available capacity," acknowledging the current scarcity of high-performance computing resources in the EU.
This structure ensures that Member States are incentivised to pool resources, knowing that their contribution will be leveraged by the Union's EuroHPC capacity, thereby maximising the total compute available for strategic AI development.
What this means for you
For public authorities, research consortia, and industry stakeholders, the CADA proposal introduces a new pathway for accessing strategic EU computing resources.
1. Strategic Planning and Consortium Building
If your organisation aims to develop frontier AI, you cannot act alone. You must:
- Form a Consortium: Establish or join a European digital infrastructure consortium (EDIC) or an entity eligible for Union funding.
- Secure Partners: Identify and secure commitments from at least two other Member States to participate in the project. This requires early diplomatic engagement and alignment of national AI strategies.
- Prepare for Open Calls: Monitor the Commission's announcements for "open calls for expression of interest" and prepare proposals that explicitly demonstrate how the project addresses Grand Challenge 3.
2. Resource Commitment and Matching
Be prepared to commit tangible resources. The "pooling" requirement means Member States must pledge computing time.
- Leverage EuroHPC: Understand that your national contribution will trigger a matching commitment from the Union's EuroHPC capacity, effectively multiplying your available resources.
- Capacity Constraints: Be aware that the Union's matching is limited by "available capacity." Early participation may be advantageous to secure resources before capacity constraints tighten.
3. Alignment with Sovereignty and Procurement
Projects recognised under Article 8 will likely involve sensitive data and critical infrastructure.
- Sovereignty Compliance: Ensure that the cloud and AI services used in these projects comply with the Union assurance levels required under Articles 29 and 30.
- Procurement Strategy: Public bodies involved in these projects must align their procurement strategies with the CADA sovereignty framework to ensure the infrastructure supporting these priority projects is resilient and secure.
Common misconceptions
"Any large AI project can be designated a frontier AI priority project."
- Reality: No. The project must specifically support Grand Challenge 3 (Frontier AI) and be selected through an open call. It must also meet the strict participation (3+ Member States) and pooling criteria of Article 8.
"A single Member State can designate its own national project as a frontier AI priority project."
- Reality: No. Article 8 explicitly requires the participation of at least three Member States. The Commission makes the final decision, not individual national governments.
"Recognition guarantees unlimited computing power."
- Reality: Recognition triggers a commitment to allocate sufficient resources, but this is subject to the limits of available EuroHPC capacity. The Union matches resources contributed by Member States, but it does not create capacity out of thin air.
"Private companies can apply directly for this status."
- Reality: The project must be undertaken by an EDIC or an entity eligible for Union funding, and it must involve the participation of at least three Member States. Private entities must operate within this consortium framework.
Official sources
Related
- Who pays for computing resources in frontier AI projects under CADA?
- Who can apply for frontier AI priority project recognition under CADA?
- CADA Frontier AI Priority Projects: Targeted Strategic Sectors
- What public funding is linked to frontier AI priority projects under CADA?
- CADA Open Calls: How the Commission Selects Frontier AI Priority Projects
This is general information about a draft EU regulation, not legal advice.