Summary Under the proposed Cloud and AI Development Act (CADA), both AI factories and AI gigafactories are classified as high-intensity computing infrastructure essential for the EU's technological sovereignty. The proposal does not create separate legal definitions for the two; instead, it groups them together in Article 7(2)(e) as strategic assets. The distinction lies in scale and strategic designation: while "AI factories" generally support broad research and industrial deployment, "AI gigafactories" are implied to be larger-scale, cross-border strategic assets designed to meet the EU's most demanding computational needs. Both are central to national strategies and serve as the primary infrastructure for frontier AI priority projects, which require unprecedented computing power to train models that approach or exceed the state of the art.

Detail

The Cloud and AI Development Act (CADA), proposed by the European Commission on 3 June 2026 (COM(2026) 502 final), establishes a framework to strengthen Europe's cloud and AI ecosystem. A core objective of the proposal is to address the shortage of computing capacity in the EU, which currently constrains the ability of European industries to benefit from AI technologies, particularly those requiring low-latency and high-performance computing.

To achieve this, the proposal mandates that Member States integrate specific high-intensity infrastructure into their national planning. The text explicitly distinguishes between different types of facilities based on their scale and strategic role, even while grouping them under a single regulatory requirement.

High-Intensity Computing Infrastructure: Article 7(2)(e)

The primary legal anchor for these facilities is Article 7, which requires Member States to establish national cloud and AI strategies within one year of the regulation's entry into force. These strategies must include specific measures to invest in critical infrastructure.

Article 7(2)(e) explicitly lists the types of infrastructure that must be addressed:

"measures to invest in high-intensity computing infrastructure, including AI factories, AI gigafactories and quantum computers as strategic national and cross-border assets supporting research, development and industrial AI deployment across strategic sectors."

This provision places both AI factories and AI gigafactories on equal footing as high-intensity computing infrastructure. They are not merely standard data centres; they are specialized facilities designed to handle the immense computational loads required for training and deploying advanced AI models. The text further characterizes them as "strategic national and cross-border assets," indicating that their importance extends beyond local or national boundaries to the Union level.

Scale and Strategic Designation

While the proposal groups them together in Article 7(2)(e), the linguistic distinction between "factory" and "gigafactory" implies a difference in scale and strategic ambition, consistent with terminology used in other EU industrial policy contexts (such as the battery and semiconductor sectors).

  • AI Factories: These are large-scale data centres optimized for AI workloads. They serve as the backbone for regional and national AI adoption, supporting the "Cloud and AI Leadership Initiatives" outlined in Title II of the proposal. They are designed to support the broad deployment of AI across the public and private sectors, including industrial AI models for healthcare, automotive, and manufacturing.
  • AI Gigafactories: The term "gigafactory" suggests facilities of exceptional size and capacity. In the context of CADA, these are identified as the larger-scale strategic assets. The proposal's explanatory memorandum notes that the EU needs to "expand its cloud and data centre capacity to support the wider deployment and diffusion of AI" and aims to "triple EU capacity in the next five-to-seven years." AI gigafactories are likely the primary vehicles for achieving the upper end of this capacity expansion. They are intended to anchor the EU's most critical computational needs, potentially serving multiple Member States or supporting Union-level initiatives that require massive, coordinated compute clusters.

The proposal aims to reach the needed capacity by 2035. While AI factories contribute to this goal, gigafactories represent the "next-generation computational resources" necessary to ensure the EU can compete globally in frontier AI development.

Connection to Frontier AI Projects

The distinction between these facilities becomes particularly relevant when considering frontier AI priority projects. Frontier AI is defined in Article 2(4) 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."

Developing these models requires unprecedented amounts of computing power, far beyond what standard data centres can provide. Under Article 8 and Article 9 of CADA, the Commission can recognize certain projects as "frontier AI priority projects." These projects receive prioritized access to AI computing resources.

Article 9(2) states that the Union will match the AI computing resources contributed by Member States to these frontier AI priority projects, within the limits of available European High Performance Computing (EuroHPC) capacity.

AI gigafactories are the natural infrastructure for these strategic initiatives. Because frontier AI requires massive, coordinated compute clusters, the larger scale of gigafactories makes them the ideal fit for hosting these projects. AI factories, while still capable of supporting significant AI workloads, may focus more on the deployment, fine-tuning, and operational use of models rather than the initial, resource-intensive training of frontier models. The proposal explicitly links these assets to the "Cloud and AI Leadership Initiatives," which aim to support "pioneering projects in frontier AI that develop frontier AI models and systems as strategic assets" (Article 4(3)).

National Strategies and Coordination

Article 7 requires Member States to establish national cloud and AI strategies within one year of the regulation's entry into force. These strategies must align with the EU's digital targets, including the deployment of secure edge nodes and the adoption of cloud services.

When drafting these strategies, Member States must consider the geographic distribution of these assets. The proposal emphasizes "balanced geographic deployment" to avoid concentration risks. This means that while gigafactories might be fewer in number due to their size and cost, AI factories should be distributed more widely to ensure that all regions of the EU have access to high-performance computing. This distribution supports the goal of reducing dependencies on third-country providers by creating a resilient, decentralized network of sovereign compute resources.

What this means for you

For public-sector bodies, procurement officers, and industry stakeholders, understanding the distinction between AI factories and AI gigafactories is crucial for planning and procurement activities under the proposed CADA.

  1. Strategic Planning: When developing your national or organizational cloud and AI strategy (as required by Article 7), you must identify where these high-intensity computing assets will be located. You need to determine whether your jurisdiction will host an AI factory, contribute to a cross-border AI gigafactory, or rely on existing EuroHPC infrastructure. The proposal requires these strategies to include measures to invest in these specific assets as "strategic national and cross-border assets."
  2. Procurement Criteria: CADA introduces new procurement rules. Article 32 requires contracting authorities to include "Union added value" criteria in public procurement for cloud computing services and AI systems. When procuring compute capacity, you may need to prioritize providers that utilize EU-designed or manufactured hardware, which is a key focus of the infrastructure investments in AI factories and gigafactories.
  3. Frontier AI Access: If your public sector body is involved in or supports frontier AI research, you should coordinate with national competent authorities to ensure access to the computing resources allocated under Article 9. Understanding which facilities (factories vs. gigafactories) are designated for these priority projects will help you navigate the allocation process. Frontier AI projects will likely require access to the massive capacity of gigafactories.
  4. Sovereignty Levels: The proposal establishes a Union cloud computing sovereignty framework with four assurance levels. Procurement of cloud services from AI factories and gigafactories must align with the appropriate assurance level determined by risk assessments under Article 29. For critical public order activities, you may need to procure from services offering higher assurance levels (2, 3, or 4), which may influence which infrastructure providers you can engage.

Common misconceptions

  • Misconception: AI gigafactories are a separate legal category with different rules.
    • Reality: CADA does not create a separate legal definition for "AI gigafactory" distinct from "AI factory." Both are grouped together in Article 7(2)(e) as high-intensity computing infrastructure. The distinction is primarily one of scale and strategic designation, not a different regulatory regime.
  • Misconception: Only private companies can build these facilities.
    • Reality: The proposal encourages public-private partnerships and EU funding (e.g., through Horizon Europe or the Digital Europe Programme) to support the development of this infrastructure. Public authorities play a key role in designating acceleration zones (Article 10) and facilitating permits for these projects.
  • Misconception: Frontier AI projects can use any data center.
    • Reality: Frontier AI priority projects require specific, high-capacity compute resources. Under Article 9, these resources are allocated from available EuroHPC capacity and other high-intensity infrastructure. Standard data centres may not have the necessary power, cooling, or network connectivity to support these workloads efficiently.

Related

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