Summary Under the proposed EU Cloud and AI Development Act (CADA) — a draft regulation that is not yet in force — a "frontier AI priority project" would be a pioneering AI project that the European Commission formally recognises by decision. As proposed in Article 8, the Commission could recognise such projects, selected through open calls for expression of interest, where they support "grand challenge 3" (Frontier AI) in Annex I, are led by a European digital infrastructure consortium or another legal entity eligible for funding under Union law, involve at least three Member States, and pool computing time and other resources. Recognition would unlock strategic support under Article 9, including a Union pledge to match the AI computing resources that Member States contribute.

Detail

The Cloud and AI Development Act (CADA) is a Commission proposal — COM(2026) 502 final — aimed at strengthening Europe's capacity in cloud computing and artificial intelligence. As a proposal, none of its provisions are yet law; the text below describes what it would do if adopted as drafted.

One strand of the proposal concerns "frontier AI." Article 2(4) would define frontier AI as "AI models or AI systems built upon such models that can perform a wide variety of tasks and that approach, reach or exceed the current state of the art." Article 3 sets advancing the Union's capabilities in frontier AI as operational objective 3 of the Cloud and AI Leadership Initiatives, and Article 4(3) says that under that objective the initiatives would "support pioneering projects in frontier AI that develop frontier AI models and systems as strategic assets, including in key sectors such as cybersecurity." To channel investment toward such work, CADA introduces the concept of frontier AI priority projects.

What would a frontier AI priority project be?

A frontier AI priority project would not be just any advanced-AI initiative. It would be a specific status that the Commission grants by formal decision to projects meeting strict criteria. As proposed in Article 8, "the Commission may, by means of a decision, recognise as frontier AI priority projects, projects selected through open calls for expression of interest that support grand challenge 3 set out in Annex I," provided certain criteria are met.

Two points follow. First, recognition would not be automatic: projects would compete through open calls for expression of interest. Second, the project must support grand challenge 3 (Frontier AI) in Annex I. That grand challenge centres on "developing the next generation of multimodal frontier AI models and systems and pioneering novel capabilities" — pushing the boundaries of current algorithmic capabilities in advanced reasoning, cross-modal understanding and agentic capabilities, and investigating novel approaches to model efficiency and alternative computational structures. Annex I lists potential applications including scientific discovery and the development of "world models" for improved reasoning, simulation and planning. Under Article 6(2), the operational objectives would be implemented through the grand challenges set out in Annex I.

The criteria in Article 8

Article 8 lists three criteria that a project would have to satisfy:

  1. A pioneering project. It must be "a pioneering project, focused on the support and scaling-up of frontier AI technologies" (Article 8(a)). This points to work that advances the state of the art rather than incremental application development.
  2. An eligible lead entity. 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" (Article 8(b)). Decision (EU) 2022/2481 is the Digital Decade Policy Programme; a European Digital Infrastructure Consortium (EDIC) is a legal vehicle that lets Member States jointly run multi-country digital infrastructure projects. Importantly, the lead does not have to be an EDIC — it can be any legal entity eligible for funding under Union law.
  3. Multi-country participation and shared resources. The project must involve "the participation of at least three Member States" (Article 8(b)), and "the participating Member States pool computing time and other relevant resources to support the implementation of the designated project" (Article 8(c)).

Recital 34 explains the rationale: frontier AI development requires resources at an unprecedented scale, so its technical complexity and capital-intensive nature call for a collaborative Union-level approach, with broad participation from entities across the Union — in particular through EDICs or any other legal structure capable of representing a meaningful share of the Union's interest.

What recognition would unlock

While Article 8 sets the criteria, Article 9 sets out the support. Article 9(1) provides 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 that fulfil the criteria set out in Article 8, within the limits of available capacity."

The key incentive is in Article 9(2): "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." In other words, the Union would aim to at least double the compute that participating Member States bring — but only as far as capacity allows. Recital 35 frames this matching as being on a proportional basis and within the limits of available EuroHPC capacity, and says it is without prejudice to the existing rules in Council Regulation (EU) 2021/1173 and to projects already benefiting from allocated EuroHPC resources.

Article 9(3) adds a broader, softer commitment: the Union and Member States "shall endeavour to provide sufficient computing resource for AI industrial innovation, physical AI and public sector AI projects."

Why frontier AI sits in a separate framework

The AI Act (Regulation (EU) 2024/1689), which entered into force on 1 August 2024, is a product-safety and fundamental-rights regulation for AI systems, and it includes rules on general-purpose AI models (Articles 51-56). It does not, however, regulate cloud infrastructure or sovereignty — the gap CADA is designed to address. CADA is about building and resourcing European AI capacity, not about regulating the safety of AI products; the two would operate side by side.

What this means for you

For public-sector bodies, research institutions and infrastructure operators following the proposal, frontier AI priority projects would matter for partnership and planning — if and when CADA is adopted.

  1. Cross-border partnership would be essential. As proposed, a project must involve at least three Member States. A purely national or single-organisation effort would not qualify, so building or joining a multi-country consortium would be a prerequisite.
  2. Alignment with grand challenge 3 would be tested. Projects would need to support the Annex I goals — next-generation multimodal models, advanced reasoning, cross-modal understanding and agentic capabilities. Narrow, non-frontier applications may not meet the "pioneering" bar in Article 8(a).
  3. Compute matching is the headline benefit. Recognition would open the door to the Article 9(2) matching mechanism within the Union's EuroHPC access time. Note the qualifications: matching is "at least" the Member State contribution but is bounded by available capacity.
  4. The lead vehicle matters. The lead must be an EDIC or another legal entity eligible for funding under Union law. Organisations considering a bid would want to confirm early whether their structure fits, or whether an EDIC is the right vehicle.
  5. Watch for open calls. Recognition would flow from open calls for expression of interest. Because CADA is still a proposal, no such calls exist yet; the process would only begin if and after the regulation is adopted.

Common misconceptions

Misconception 1: Any advanced AI project is a frontier AI priority project. Reality: Only projects the Commission recognises by decision under Article 8 would qualify, and only after selection through open calls. The criteria — pioneering work, an eligible lead entity, at least three Member States, and pooled resources — are cumulative.

Misconception 2: Recognition would guarantee unlimited compute. Reality: Article 9 ties allocation to capacity. Article 9(1) applies "within the limits of available capacity," and the Article 9(2) matching applies only "to the extent that sufficient AI computing capacity is available" within the Union's share of EuroHPC access time.

Misconception 3: Only EDICs or public bodies could lead, and private companies are excluded. Reality: Article 8(b) allows the lead to be an EDIC or "another legal entity eligible for funding under Union law." The lead need not be public or an EDIC; the firm requirements are an eligible legal entity, the participation of at least three Member States, and the pooling of computing time and other resources.

Misconception 4: It is only about training large language models. Reality: Grand challenge 3 covers multimodal models and systems, novel capabilities including agentic capabilities, model efficiency and alternative computational structures, with applications such as scientific discovery and world models. The scope is broader than training a single type of model.

Official sources

Related

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