What is Grand Challenge 3 (frontier AI) under CADA?

Grand Challenge 3 under the proposed Cloud and AI Development Act (CADA) is dedicated to "Frontier AI," aiming to develop next-generation multimodal models

Summary Grand Challenge 3 under the proposed Cloud and AI Development Act (CADA) is dedicated to "Frontier AI," aiming to develop next-generation multimodal models and systems with advanced reasoning, cross-modal understanding, and agentic capabilities. As defined in Annex I, point 3, this challenge seeks to pioneer novel algorithmic capabilities and architectural designs to maintain the EU's competitive edge. It serves as the strategic foundation for operational objective 3 of the Cloud and AI Leadership Initiatives and establishes the criteria for designating frontier AI priority projects under Article 8, which unlock access to Union computing resources and funding.

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

Detail

The proposed Cloud and AI Development Act (CADA), COM(2026) 502 final, establishes a comprehensive framework to strengthen Europe's cloud and AI ecosystem. A central pillar of this framework is the "Cloud and AI Leadership Initiatives," which are structured around eight specific "grand challenges" detailed in Annex I. Grand Challenge 3 is exclusively focused on "Frontier AI," representing the EU's strategic response to the rapid evolution of artificial intelligence and the need to reduce dependencies on third-country technologies.

The Strategic Scope of Grand Challenge 3

According to Annex I, point 3, Grand Challenge 3 is titled "Frontier AI." Its primary mandate is the development of the next generation of multimodal frontier AI models and systems that pioneer novel capabilities. The proposal explicitly identifies three core technical domains that define this challenge:

Advanced Reasoning: The challenge focuses on architectural designs and development that push the boundaries of current algorithmic capabilities. This includes achieving superior performance in advanced reasoning, enabling models to perform complex logical deductions and problem-solving tasks that exceed existing state-of-the-art.
Cross-Modal Understanding: A key objective is the development of models capable of superior cross-modal understanding. This involves the ability to interpret and integrate information across different data types—such as text, images, audio, and video—simultaneously and coherently, moving beyond single-modality limitations.
Agentic Capabilities: The challenge targets the creation of systems with agentic capabilities. These are systems designed to operate autonomously, planning and executing actions to achieve specific goals, rather than merely responding to prompts. The proposal emphasizes the need for rigorous security standards and resilience in the orchestration of these autonomous agents.

Beyond these specific capabilities, the proposal notes that the focus includes investigating novel approaches to model efficiency, cognitive modelling, and alternative computational structures. The potential applications identified in Annex I extend to foundational science, such as scientific discovery and complex data interpretation, as well as the development of world models for improved reasoning, automated management simulation, and planning.

Link to Operational Objective 3

Grand Challenge 3 is not an isolated initiative; it is the practical implementation vehicle for operational objective 3 of the Cloud and AI Leadership Initiatives. Article 4(3) of the proposal explicitly states that under operational objective 3, the initiatives shall "support pioneering projects in frontier AI that develop frontier AI models and systems as strategic assets, including in key sectors such as cybersecurity."

This linkage ensures that the broad goals of the Cloud and AI Leadership Initiatives—outlined in Article 3 to promote research, innovation, and large-scale capacity—are concretely directed toward the most advanced AI developments. By anchoring Grand Challenge 3 to operational objective 3, CADA ensures that policy support, funding, and strategic attention are concentrated on developing frontier AI as a critical strategic asset for the Union.

Frontier AI Priority Projects and Article 8

To translate the ambition of Grand Challenge 3 into tangible outcomes, CADA introduces the mechanism of frontier AI priority projects. Article 8 sets out the specific criteria the Commission must use to recognize a project as a frontier AI priority project. For a project to qualify, it must meet the following cumulative conditions:

Alignment with Grand Challenge 3: The project must support "grand challenge 3 set out in Annex I," ensuring it focuses on the development and scaling-up of frontier AI technologies.
Pioneering Nature: It must be a pioneering project focused on the support and scaling-up of frontier AI technologies.
Consortium Structure: 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.
Cross-Border Participation: It must involve the participation of at least three Member States.
Resource Pooling: The participating Member States must pool computing time and other relevant resources to support the implementation of the designated project.

Once designated, these projects trigger specific support mechanisms. Article 9 mandates that the Union and Member States ensure sufficient AI computing resources are allocated to support these projects. Specifically, 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 provision is critical for overcoming the compute bottleneck that often hinders the training of large-scale frontier models.

Broader Context and Strategic Assets

The proposal frames frontier AI technologies as "critical strategic assets." Recital 16 of the proposal explains that strengthening the Union's capacity to develop and govern these technologies is essential to ensure the AI transition aligns with Union values, safety standards, and long-term economic interests. By supporting pioneering projects under Grand Challenge 3, the EU aims to scale up essential breakthroughs to maintain a competitive edge in the global digital economy.

Furthermore, the proposal highlights the importance of reducing dependencies on third-country technologies. Recital 16 notes that fostering the development of frontier AI technologies as strategic assets should "reduce current dependencies on third-country technologies and strengthen the Union's AI ecosystem." This strategic autonomy is a recurring theme, linking the technical goals of Grand Challenge 3 to the broader sovereignty objectives of the CADA proposal.

What this means for you

For technology leaders, researchers, and industry stakeholders, Grand Challenge 3 represents a significant shift in the EU's approach to AI innovation. It signals a move from general support to targeted investment in the most advanced, strategically critical AI capabilities.

1. Strategic Collaboration is Mandatory The requirement in Article 8 for projects to involve at least three Member States and be undertaken by an EDIC or eligible legal entity means that solo national efforts are unlikely to qualify as frontier AI priority projects. Organizations must proactively build cross-border consortia. If you are a research institution or a tech SME with expertise in multimodal AI or agentic systems, seeking partners in other Member States is a prerequisite for accessing the specific support mechanisms tied to Grand Challenge 3.

2. Access to High-Performance Compute One of the most valuable incentives for qualifying as a frontier AI priority project is access to computing resources. Article 9 guarantees that designated projects will receive matching AI computing resources from the Union's share of EuroHPC capacity. For organizations developing large-scale multimodal models, this access to high-performance computing is often the difference between a theoretical model and a deployed system. Aligning your R&D roadmap with the criteria of Grand Challenge 3 could unlock this critical infrastructure.

3. Focus on Novel Architectures and Efficiency Grand Challenge 3 is not merely about scaling existing models; it explicitly calls for "novel approaches to model efficiency, cognitive modelling, and alternative computational structures." Stakeholders should prioritize research into energy-efficient architectures and new computational paradigms. The proposal suggests that the EU is looking for breakthroughs that go beyond current state-of-the-art, rather than incremental improvements.

4. Alignment with Union Values and Safety The proposal emphasizes that the AI transition must align with "Union values, safety standards and long-term economic interests." As you develop frontier AI systems, particularly those with agentic capabilities, ensure that your safety protocols and governance frameworks are robust. The proposal highlights the need for rigorous testing and experimentation methodologies to minimize unintended autonomous behavior. Compliance with these emerging standards will be essential for project recognition and long-term market acceptance.

Common misconceptions

"Grand Challenge 3 is only for large hyperscalers." While frontier AI is resource-intensive, the proposal explicitly encourages broad participation through European digital infrastructure consortiums (EDICs). The requirement for at least three Member States suggests a collaborative model where SMEs, research institutes, and startups can contribute specialized expertise, novel algorithms, or sector-specific applications to a larger consortium. The focus is on the consortium's ability to pool resources, not on the size of individual participants.

"Designation as a priority project guarantees direct cash funding." Recognition as a frontier AI priority project under Article 8 primarily unlocks access to specific support mechanisms, most notably the allocation of AI computing resources from EuroHPC capacity under Article 9. While the proposal mentions that the initiatives may be supported by funding from Union programmes like Horizon Europe and the Digital Europe Programme, the designation itself is a gateway to compute and strategic alignment, not an automatic cash grant. Funding eligibility will still depend on separate selection processes.

"Frontier AI is just about bigger models." Grand Challenge 3 explicitly targets "novel approaches to model efficiency, cognitive modelling, and alternative computational structures." The focus is on architectural innovation and the development of capabilities like advanced reasoning and agentic behavior, not just increasing parameter counts. The proposal seeks to solve the fundamental challenges of how AI models think and act, rather than simply scaling up existing architectures.

"Agentic capabilities mean unrestricted autonomy." The proposal stresses the need for "robust technical frameworks to ensure the safety, accuracy and legal compliance of those systems." Recital 21 highlights the need for rigorous testing and experimentation methodologies to "minimise unintended autonomous behaviour." The development of agentic capabilities under Grand Challenge 3 is expected to be accompanied by strict safety and accountability measures, ensuring that autonomous systems operate within defined legal and ethical boundaries.

Official sources

Digital Decade Policy Programme (Decision (EU) 2022/2481)