Summary Under the proposed Cloud and AI Development Act (CADA), startups and SMEs can access subsidised AI compute through a two-tiered approach: first, by engaging with the national Experience and Acceleration Centres for AI (Centres for AI) as the primary entry point for support and infrastructure access; and second, by applying for direct allocation of computing resources from European high-performance computing (EuroHPC) infrastructure. As proposed, Member States must establish these Centres to connect organisations with European cloud providers and facilitate access to compute. Furthermore, the Union and Member States are obligated to allocate sufficient AI computing resources to support "AI industrial innovation" and "physical AI" projects, matching contributions to frontier AI priority projects within the limits of available EuroHPC capacity.

Detail

The Cloud and AI Development Act (CADA), as proposed in COM(2026) 502 final, establishes a comprehensive framework to address the computational capacity deficit in the EU. For startups and small and medium-sized enterprises (SMEs), the legislation outlines specific mechanisms to access compute power, primarily through the Cloud and AI Leadership Initiatives and the associated infrastructure networks. The Act does not create a single "cash grant" for compute but rather a legal obligation to prioritise access to existing and future European infrastructure for strategic projects.

Step 1: The Entry Point – Experience and Acceleration Centres for AI

The most immediate and practical route for a startup to access support, guidance, and compute resources is through the Experience and Acceleration Centres for AI (referred to as "Centres for AI"). As proposed in Article 5, each Member State is required to establish these Centres, building upon the existing network of European Digital Innovation Hubs (EDIHs).

Article 5(1) mandates that Member States establish these Centres to support the integration and scaling-up of AI use cases. Article 5(2) explicitly lists as an objective the acceleration of the broad adoption of cloud and AI technologies at regional and local levels, "notably for SMEs, SMCs [Small Mid-Caps] and public sector bodies."

Crucially for technical teams seeking infrastructure, Article 5(3)(a) tasks these Centres with "helping organisations accelerate their digital transformation through access to and use of AI technologies, including by connecting organisations with European providers of cloud and AI technologies." Furthermore, Article 5(2)(c) states that the objectives of the Centres include leveraging relevant infrastructure to "accelerate the development and fine-tuning of AI models and systems."

For a startup CTO, this means the Centre for AI in your jurisdiction serves as the first port of call. They are designed to provide not just advisory support, but also access to the infrastructure necessary for model development. This may include facilitating access to testing environments, providing technical guidance on applying for larger compute grants, or connecting the startup with European cloud providers offering sovereign compute solutions.

Step 2: Direct Allocation of Compute Resources and EuroHPC

Beyond local support, CADA proposes a mechanism for the direct allocation of AI computing resources, linking national strategies to the Union's high-performance computing capabilities. Article 9, titled "Computing support for AI projects," outlines how the Union and Member States will manage compute capacity.

Article 9(3) states that the Union and Member States "shall endeavour to provide sufficient computing resource for AI industrial innovation, physical AI and public sector AI projects." This is a key provision for startups working on industrial applications, robotics, autonomous systems, or other "physical AI" use cases.

For more advanced, large-scale projects, Article 9(2) specifies that 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 links CADA directly to the EuroHPC initiative. While EuroHPC access is currently managed through separate programmes, CADA formalises the obligation to allocate these resources strategically to support frontier AI.

Step 3: Funding, Open Calls, and Eligibility

The operational objectives of the Cloud and AI Leadership Initiatives, set out in Article 4, include supporting the development of advanced data centre technologies and cloud computing stacks. Article 6(3) clarifies that these initiatives may be supported by funding from Union programmes, including Horizon Europe and the Digital Europe Programme.

Startups and SMEs should note that while CADA establishes the obligation to provide access and support, the actual funding and access procedures are often implemented through these existing Union programmes. The proposal indicates that the Commission and Member States will implement these operational objectives through large-scale, cross-sectoral initiatives addressing "grand challenges" (see Article 6(2) and Annex I). These grand challenges include areas such as "Frontier AI," "Physical AI," and "Industrial AI," which are likely to be the focus of compute allocation.

Eligibility and Open Calls: While CADA itself sets the strategic framework, the specific eligibility criteria for accessing subsidised compute will largely be defined in the calls for proposals issued under the Digital Europe Programme and Horizon Europe, aligned with the national strategies required by Article 7. Article 7(2)(b) requires Member States to include measures in their national cloud and AI strategies to "accelerate the development and adoption of cloud and AI at national, regional and local level, particularly among public sector bodies, SMEs and SMCs."

Therefore, startups should monitor:

  1. National Cloud and AI Strategies: Required by Article 7(1) to be adopted by Member States. These will detail national support measures and how local Centres for AI will operate.
  2. Digital Europe Programme Calls: For access to AI factories and high-performance computing. These calls will likely prioritise projects aligned with the "Grand Challenges" in Annex I.
  3. Local Centres for AI: For immediate, smaller-scale compute needs, advisory support, and guidance on applying for larger grants.

The Act also emphasises the role of open source as a lever for sovereignty. Article 41 encourages the use of open standards and components, and Article 42 requires public sector bodies to share software under open-source licences. Startups that contribute to or utilise the EU Open Source Solutions Catalogue (Article 43) may find their projects more aligned with the strategic objectives of the Act, potentially improving their eligibility for support.

What this means for you

For CTOs and architects at startups or SMEs, CADA represents a formalised pathway to European compute resources, reducing reliance on non-EU hyperscalers. Here is how to act on this proposal:

  1. Identify Your Local Centre for AI: Locate the European Digital Innovation Hub (EDIH) in your region that has been refocused as a Centre for AI. Engage with them early. As per Article 5, their mandate includes connecting you with European providers and facilitating access to infrastructure for model fine-tuning. They can help navigate the application processes for larger compute grants.
  2. Align with "Grand Challenges": Review the "Grand Challenges" listed in Annex I of the CADA proposal (e.g., Industrial AI, Physical AI, Frontier AI). Tailor your project proposals to align with these strategic priorities, as compute resources under Article 9 will be prioritised for these areas.
  3. Prepare for EuroHPC Applications: If your startup is working on frontier or industrial AI, prepare to apply for compute time on EuroHPC systems. CADA's Article 9 reinforces the political commitment to allocating this capacity, but you must still compete for it through the established EuroHPC access policies, which will be adapted to reflect these new priorities.
  4. Monitor National Strategies: Wait for your Member State to publish its national cloud and AI strategy (required by Article 7). This document will contain specific details on national support measures, including potential subsidies or direct compute credits for SMEs.
  5. Leverage Open Source: Article 41 encourages the use of open-source solutions. Demonstrating that your stack relies on open-source components or contributes to the EU Open Source Solutions Catalogue (Article 43) may strengthen your position in procurement or funding applications, as the Act promotes technological sovereignty.

Common misconceptions

  • "CADA creates a new, standalone compute grant scheme." CADA does not create a new, separate funding programme. Instead, it establishes a legal framework that obliges Member States and the Commission to prioritise compute access for strategic AI projects. The actual funding and compute allocation are delivered through existing instruments like the Digital Europe Programme, Horizon Europe, and EuroHPC, guided by CADA's strategic objectives.

  • "Any startup can automatically get free compute." Access is not automatic or universal. Article 9 specifies that resources are allocated to "frontier AI priority projects," "AI industrial innovation," and "physical AI" projects. Startups must demonstrate alignment with these strategic goals and compete for resources, often through open calls managed by the Commission or Member States.

  • "The Centres for AI are just advisory bodies." While they provide advice, Article 5 explicitly tasks Centres for AI with facilitating "access to and use of AI technologies" and leveraging infrastructure for "development and fine-tuning of AI models." They are operational hubs designed to provide tangible access to resources, not just consultancy.

  • "CADA overrides the AI Act's compute requirements." CADA and the AI Act are complementary. The AI Act (Regulation (EU) 2024/1689) imposes obligations on providers of general-purpose AI models, including those with systemic risks. CADA facilitates the infrastructure and ecosystem to help European providers, including startups, meet these obligations and compete globally. CADA does not exempt any entity from AI Act compliance.

Official sources

Related

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