Summary AI gigafactories are large-scale, high-intensity computing facilities aimed at training the most advanced AI models — the top tier of the EU's compute ambitions and a pillar of its technological sovereignty. As proposed, the Cloud and AI Development Act (CADA) does not define them in its definitions article and does not fund them directly. Its enacting link is Article 7(2)(e), which requires Member States to include measures to invest in high-intensity computing infrastructure — naming AI gigafactories alongside AI factories and quantum computers — in their national cloud and AI strategies. Funding would come from existing or planned Union instruments (Horizon Europe, the Digital Europe Programme, InvestEU and the future European Competitiveness Fund), national support under State aid rules, and private investment.

Detail

The EU's AI strategy depends on access to sovereign, high-capacity compute. As the CADA explanatory memorandum sets out, the lack of domestic data centre capacity forces European enterprises to route critical workloads through foreign hyperscaler infrastructure, and computing infrastructures have become "strategic resources critical to the Union's economic security, sovereignty, resilience, and competitiveness." Within this picture, the proposal distinguishes two scales of facility: AI factories and AI gigafactories.

What are AI gigafactories?

CADA does not provide a standalone technical definition of "AI gigafactory" in Article 2. The term appears in the explanatory memorandum and once in the enacting text. In common usage within EU policy, gigafactories are the largest facilities, aimed at training the most advanced (frontier) AI models, sitting above general-purpose data centres and the broader-access "AI factories." Where AI factories aim to "provide broad access to high-capacity, next-generation computational resources for European businesses and researchers" (explanatory memorandum), gigafactories are oriented to the most demanding frontier-AI training workloads.

The role of Article 7: national strategies

The legal anchor in CADA's enacting text is Article 7, which requires Member States to adopt national cloud and AI strategies within one year of entry into force, consistent with the Regulation's objectives (Article 7(1) and (3)). Article 7(2)(e) specifically requires these strategies to include:

"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 treats AI gigafactories as strategic national and cross-border assets, embedding their planning in a coordinated national framework rather than leaving it to ad hoc commercial decisions.

Who funds them?

CADA provides the framework, not the money. As proposed, financing would draw on a mix of EU programmes, national support and private capital — the same architecture CADA uses for its Cloud and AI Leadership Initiatives:

  1. Union programmes. Article 6(3) allows the Leadership Initiatives to be supported by funding from Union programmes, "including Horizon Europe and the Digital Europe Programme." Recital 28 adds InvestEU. The explanatory memorandum expects the European Competitiveness Fund (ECF) — still a proposal — to serve as the main deployment instrument in the post-2027 multiannual financial framework, with FP10 (the Horizon Europe successor) supporting upstream research.
  2. National support and State aid. Recital 29 permits Member States to support the initiatives through research, development and innovation measures "in line with the applicable State aid rules." Large cross-border facilities are also candidates for IPCEIs, which the explanatory memorandum expects to "continue to support large-scale, cross-border projects" in cloud, edge, chips, cybersecurity or AI infrastructure.
  3. Private investment. Recital 29 encourages private-sector stakeholders to align their cloud and AI investment strategies with the Leadership Initiatives, helping create "a coherent and coordinated investment pathway."

A note on figures and named initiatives: the CADA proposal itself does not mention a specific "InvestAI" instrument or a headline figure such as EUR 20 billion. (InvestAI is a separate Commission initiative announced in 2025; it is not part of the CADA text.) Any such figures should not be attributed to CADA. The proposal also notes synergies with the EuroHPC Joint Undertaking: Article 9(2) provides that, for recognised frontier AI priority projects, the Union shall at least match Member States' contributed AI computing resources "to the extent that sufficient AI computing capacity is available within the Union's share of European high performance computing access time."

The frontier-AI link: Articles 8 and 9

Although gigafactories are not themselves a defined or separately funded object in CADA, the proposal's frontier-AI provisions are the closest thing to a dedicated compute mechanism for the most advanced workloads. Article 8 lets the Commission recognise, by decision, "frontier AI priority projects" selected through open calls — projects that are pioneering in scaling up frontier AI, undertaken by an EDIC or another entity eligible for Union funding involving at least three Member States, and where those States pool computing time and resources. Article 9 then provides computing support: the Union and Member States are to allocate sufficient AI computing resources to such projects within available capacity (Article 9(1)), and the Union shall at least match Member States' contributions from its share of European high-performance computing access time, subject to availability (Article 9(2)). For organisations whose ambitions match gigafactory-scale training, this recognition-plus-pooled-compute route is the most concrete pathway CADA sets out — distinct from, but complementary to, the national-strategy investment measures in Article 7(2)(e).

Strategic objectives and sovereignty

The drive for gigafactories is about sovereignty as well as raw capacity. The explanatory memorandum records that the EU providers' share of the cloud market fell from 29% in 2017 to 15% in 2022, and that three non-EU hyperscalers control over 70% of the European cloud market — exposing users to operational discontinuity, reduced control over data, and the extraterritorial reach of third-country laws. By building gigafactories, the EU aims to reduce these dependencies, keep advanced AI training under Union and national jurisdiction, and give European businesses (including SMEs and start-ups) access to high-performance compute. Sustainability is built in: the Leadership Initiatives' first operational objective (Article 4(1)) targets energy- and water-efficiency technologies for data centres, including innovative cooling, waste-heat utilisation and integration with energy grids.

What this means for you

For CTOs, architects and SMEs:

  • Sovereign compute access. As these facilities come online, European businesses should gain access to high-capacity compute under EU jurisdiction, reducing lock-in to non-EU providers for frontier-scale training.
  • Align with national strategies. Track your Member State's Article 7 strategy to understand where high-intensity computing investment is planned; let that inform your technology roadmap.
  • SME and start-up access is an explicit aim. The proposal seeks broad access, and Article 33 pushes Member States to improve SME participation in cloud and AI procurement of innovation.
  • Efficiency matters. Given Article 4(1)'s emphasis, energy-efficient design is likely to be both a competitive advantage and an expectation for supported infrastructure.

Common misconceptions

  • "CADA directly funds AI gigafactories." No. CADA mandates national strategies and sets up the Leadership Initiatives; financing comes from EU programmes, national budgets and private investment.
  • "CADA earmarks EUR 20 billion through InvestAI for gigafactories." The CADA proposal text contains no such figure and does not reference an "InvestAI" instrument; do not attribute that to CADA.
  • "Gigafactories are only for large corporations." The proposal aims to broaden access, including for SMEs, researchers and public bodies.
  • "AI gigafactories and AI factories are the same." They differ in scale and focus: AI factories provide broad access to next-generation compute, while gigafactories target the most advanced frontier-AI training workloads.

Related

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