Summary Under the proposed Cloud and AI Development Act (CADA), "pooling computing time and other relevant resources" is a mandatory precondition for any project to be recognized as a frontier AI priority project. Article 8(c) explicitly requires that participating Member States collectively contribute compute capacity and supporting resources to support the implementation of these designated projects. This pooling mechanism is not merely a suggestion; it is the trigger that enables the Union to "match" those contributions with additional European high-performance computing (EuroHPC) access time under Article 9(2). Without this pooled commitment from Member States, the Union matching mechanism cannot be activated, and the project would fail to meet the criteria for priority recognition.

Detail

The proposed Cloud and AI Development Act (CADA), COM(2026) 502 final, establishes a comprehensive framework to strengthen Europe's technological sovereignty in artificial intelligence. A central pillar of this framework is the support for "frontier AI priority projects"β€”pioneering initiatives designed to develop and scale up frontier AI technologies as strategic assets. To qualify for this special status and the associated support, projects must meet strict criteria outlined in Article 8 of the proposal.

The Legal Requirement: Article 8(c)

Article 8 sets out the specific criteria for the Commission to recognize a project as a frontier AI priority project. These projects are selected through open calls for expressions of interest and must support "grand challenge 3" as set out in Annex I of the regulation, which focuses on developing the next generation of multimodal frontier AI models and systems.

While the project must be pioneering and involve broad participation from entities across the Union, Article 8(c) imposes a specific collaborative obligation on the Member States involved. The text states that a project will only be recognized if:

"the participating Member States pool computing time and other relevant resources to support the implementation of the designated project."

This provision transforms frontier AI development from a purely national endeavor into a coordinated European effort. The term "pooling" implies that the contributing Member States must aggregate their respective computing capacities and resources, making them available collectively for the specific purpose of the designated project. It is not sufficient for Member States to merely promise future support or maintain separate, siloed contributions; the resources must be pooled to ensure the project has the scale and stability required for frontier AI development.

What Constitutes "Other Relevant Resources"?

While the text explicitly mentions "computing time," the phrase "and other relevant resources" allows for flexibility in what is contributed, provided it supports the project's implementation. Given the capital-intensive and technically complex nature of frontier AI, these resources likely include:

  • High-Performance Computing (HPC) Access: Dedicated time on national or regional supercomputers, specifically within the EuroHPC network.
  • Data Resources: Access to high-quality, compliant datasets necessary for training and validation, potentially linked to the data spaces mentioned in the preamble.
  • Expertise and Infrastructure: Technical support, energy infrastructure for data centers, or specialized hardware designed and manufactured in the Union.

The exact definition of what constitutes "relevant resources" may be further clarified in delegated acts or guidance from the Commission, but the core intent is to ensure that the project is backed by tangible, pooled commitments from the participating states.

The Link to Union Matching: Article 9

The obligation to pool resources under Article 8(c) is directly linked to the financial and computational support mechanism established in Article 9. Article 9 outlines how the Union and Member States will provide computing support for AI projects.

Specifically, Article 9(2) states:

"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 "match-and-scale" dynamic. The Union's commitment to match resources is conditional upon the Member States first pooling their contributions. If Member States do not pool their computing time and resources as required by Article 8(c), the precondition for Union matching under Article 9(2) is not met. Consequently, the project would not receive the additional boost from the Union's share of European high-performance computing (EuroHPC) capacity.

Why Pooling is a Precondition

The requirement to pool resources serves several strategic purposes within the CADA framework:

  1. Ensuring Critical Mass: Frontier AI models require massive amounts of computational power to train and refine. No single Member State may possess sufficient capacity to support a leading-edge project alone. Pooling ensures that the combined resources of participating states reach the threshold necessary for meaningful progress.
  2. Promoting Cross-Border Collaboration: By requiring pooling, CADA incentivizes Member States to collaborate and coordinate their national AI strategies. This aligns with the broader goal of creating a cohesive European AI ecosystem rather than fragmented national efforts.
  3. Leveraging Union Funds Efficiently: The Union's matching mechanism is designed to amplify national investments. By requiring pooling first, the EU ensures that its own resources (via EuroHPC) are used to supplement genuine, collective national efforts, maximizing the impact of public spending.
  4. Risk Mitigation: Shared resource commitments reduce the risk of project failure due to insufficient compute capacity. It also distributes the financial and operational burden across multiple states, making large-scale projects more viable.

Implementation and Monitoring

The Commission will monitor the implementation of these provisions. Member States must notify the Commission of their national strategies and their contributions to frontier AI projects. The Commission will assess whether the pooling requirements have been met before granting recognition under Article 8 and before triggering the matching mechanism under Article 9.

For cloud service providers and data centre operators, this means that projects seeking frontier AI priority status will need to demonstrate clear, verifiable commitments from multiple Member States. The pooling arrangement should be documented and transparent, showing how computing time and other resources are aggregated and allocated to the project.

What this means for you

As a cloud service provider, data centre operator, or project lead, understanding the pooling requirement is crucial for engaging with frontier AI projects that may seek priority status.

  • Project Viability: If you are developing or hosting a frontier AI model, ensure that your consortium includes participating Member States that are willing and able to pool their computing resources. Without this pooled commitment, your project cannot be recognized as a frontier AI priority project, and you will miss out on the Union-matching benefits.
  • Resource Aggregation: Be prepared to work with multiple national authorities to aggregate compute time. This may involve negotiating access to national HPC facilities, coordinating data access, and aligning technical infrastructure across borders.
  • Documentation: Maintain clear records of the pooled resources. The Commission will likely require evidence that the pooling has occurred and is sufficient to support the project's implementation. This includes documenting the amount of computing time, the nature of other resources, and the commitments from each participating Member State.
  • Strategic Partnerships: Consider forming partnerships with other EU-based providers and national research institutions to help meet the pooling requirement. Collaborative approaches can strengthen your application and demonstrate the cross-border nature of the project.

Common misconceptions

  • "Pooling means donating resources for free." Pooling does not necessarily mean that Member States or providers must donate resources without compensation. It refers to the aggregation and collective availability of resources for the project. Financial arrangements between states and providers can still exist, but the resources must be pooled for the project's benefit.
  • "Only computing time matters." While computing time is explicitly mentioned, "other relevant resources" can also be pooled. This includes data, expertise, and infrastructure. Ignoring these non-compute resources may weaken your project's overall resource base.
  • "Union matching is automatic." Union matching under Article 9(2) is not automatic. It is conditional on the Member States first pooling their resources as required by Article 8(c). If the pooling requirement is not met, the Union will not match the contributions.
  • "Any AI project can pool resources." Only projects that are recognized as "frontier AI priority projects" under Article 8 benefit from this specific framework. General AI projects do not have access to the same Union-matching mechanism unless they meet the strict criteria for frontier AI priority status.

Related

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