Cloud and AI Development Act: Europe Accelerates Digital Sovereignty with AI Gigafactories

Introduction

December 9, 2025, marks a fundamental milestone in the European strategy for artificial intelligence: the Council of the European Union has agreed on an amendment to the regulation that defines the framework for the activities of the European High-Performance Computing Joint Undertaking (EuroHPC JU)¹. As stated in the Council’s press release, the amendment aims to “establish Artificial Intelligence (AI) gigafactories in Europe and create a dedicated quantum pillar in the activities of the EuroHPC JU”¹.

This decision is part of the broader context of the Cloud and AI Development Act (CADA), a strategic initiative aimed at closing the EU’s competitive gap with the United States and China in AI and cloud computing. The issue is critically important: according to the European Parliamentary Research Service (EPRS) analysis of the European Parliament, “despite comparable GDP, the United States has twice Europe’s share of global data centre capabilities, and just three US-based companies account for 65% of the EU cloud services market”². This structural dependence threatens not only the competitiveness of European companies but also the EU’s digital sovereignty in an increasingly complex geopolitical context.

The Council’s Decision: From Supercomputers to AI Gigafactories

Regulatory Framework and Strategic Objectives

The Council has agreed to the proposed amendment to Regulation (EU) 2021/1173, which was amended in 2024 to introduce AI Factories as an additional objective of the EuroHPC JU¹. As emphasized by the Council, “the second amendment, proposed by the Commission on July 15, 2025, takes this concept to the next level, by supporting the creation of AI gigafactories”¹.

The amendment introduces two fundamental strategic pillars:

1. The development and management of AI Gigafactories: ultra-scalable infrastructures designed for the development, training, and execution of next-generation AI models
2. The Quantum Pillar: transfer of quantum research and innovation activities from the Horizon Europe program to EuroHPC JU activities “to strengthen efforts in this critical technology”¹

Characteristics of AI Gigafactories

The Council defines AI Gigafactories as “large-scale facilities combining high-performance computing, energy-efficient data centres, and AI-driven automation”¹. These centers will provide “world-class AI computing infrastructure to support European researchers, entrepreneurs, and industries, boosting Europe’s industrial strength, fostering new AI solutions, and enhancing the EU’s technological autonomy”¹.

As Christina Egelund, Danish Minister for Higher Education and Science, emphasizes: “Today, we have taken essential steps towards establishing up to five new AI Gigafactories in Europe. AI, in my opinion, is one of the most important critical technologies of tomorrow and a key to European resilience, competitiveness, and security. We owe Europeans a strong response to American and Chinese strengths in this field”¹.

Operational Framework and Safeguards

The Council’s proposed regulation establishes a clear framework for Gigafactories operations¹:

• Public-private partnerships: involvement of Member States and industry stakeholders
• Operational flexibility: allowing partners to achieve their best possible results
• Clear rules: for funding and procurement
• Protection for start-ups and scale-ups: specific measures to facilitate access
• Safeguards for third-country participation: controls on extra-EU participation
• Multi-site Gigafactories: possibility of structures across multiple countries
• Redirection of EU funds: unused funds can be reallocated to AI Gigafactory projects

The Cloud and AI Development Act: A Three-Pillar Strategy According to EPRS

The Cloud and AI Development Act (CADA), announced by Commission President Ursula von der Leyen and entrusted to Executive Vice-President Henna Virkkunen, represents a strategic complement to the AI Gigafactories initiative. As analyzed in the European Parliamentary Research Service briefing², CADA is structured on three complementary pillars, each with specific objectives and differentiated implementation timelines.

First Pillar: Research and Innovation for Computational Efficiency

The first pillar, according to EPRS analysis, aims to “promote cutting-edge research in AI-enabling technologies and related infrastructures”². The objective is twofold:

Optimizing resources for data processing: As reported by EPRS, citing the International Energy Agency, “data centres will globally need twice the energy they required in 2022 (reaching more than 800 Terawatt hours)” in 2026². The briefing emphasizes that “the cost of energy in Europe is higher than in China and the US”², as highlighted in the Draghi Report, making energy efficiency research imperative.

EPRS highlights that the Commission proposes to “leverage research and innovation to make the EU a leader in resource-efficient data processing infrastructure”². Technical solutions include:

• Hardware designed explicitly for AI
• Optimization of computational processes
• Dynamic power management systems
• “Energy-aware data centre ecosystem”

Computation continuum and decentralization: The EPRS document describes the computation continuum as “the integration of intermediate layers to address cloud limitations, including latency and cost”². Concretely, “instead of relying on one centralised workload in data centres, suitable computational tasks are shared among data centres, intermediate servers and edge devices, forming decentralised environments”².

The reference model is the “Chips for Europe” initiative of the Chips Act. However, EPRS cites the European Court of Auditors, which forecast in 2025 that “the policy will not reach its target of a 20% share in the global market value chain by revenue”², recommending an “urgent reality check”².

Second Pillar: Conditions for Investment in Data Centers

The second pillar, according to EPRS analysis, addresses structural obstacles to data center investments in the EU, with the ambitious goal of “tripling the EU’s data centre capacity within the next five to seven years”².

The competitive gap: EPRS cites McKinsey, highlighting that “despite comparable GDP levels, the US has twice Europe’s share of global data centre capabilities”², adding that “in most cases, European data centres are owned by American tech companies (known as hyperscalers)”². The briefing also notes that Boston Consulting Group forecasts that “hyperscalers will generate around 60% of global data centre growth from 2023 to 2028”².

Obstacles identified by EPRS:

• “Difficulties in accessing water and energy in the EU hinder investment in data centres”²
• “Difficulties in accessing land”²
• The permitting process “is not harmonised across Member States”, as noted by the Centre on European Regulation (CERRE) cited by EPRS²
• “The process of developing data centres is highly capital-intensive, creating capital barriers for the entry of new players to the market”²

Proposed solutions: EPRS reports that CERRE suggests “harmonising the process among Member States and creating a fast-track process for projects with social impact”². In its response to the CADA public consultation, “Microsoft mentions the Aragon region in Spain as having a successful single streamlined permitting process with a fast-track procedure for strategic projects”².

For capital, EPRS indicates that experts recommend “using the proposed European Competitiveness Fund (ECF) alongside EU rules laying down a strategy, and public and private investment to support critical technology innovators”². The ECF would aim to “consolidate 14 EU funding instruments in one framework”².

The briefing concludes by citing the Finnish government, which emphasizes that “action is needed to ensure the benefits of building data centre capacity do not primarily flow to actors outside the EU”².

Third Pillar: Secure Cloud and European Digital Sovereignty

The third pillar represents perhaps the most complex and politically sensitive challenge of CADA. EPRS defines the objective as ensuring that “a set of narrowly defined highly critical use cases can be operated using highly secure EU-based cloud capacity” and “creating the conditions for the EU cloud industry to develop such an offer”².

The definition challenge: As highlighted by EPRS, citing the Industry, Research and Energy (ITRE) Committee report of the European Parliament awaiting final vote, “previous EU discussions on this matter, with the proposal for a cybersecurity certification scheme for cloud services (EUCS), have not brought any results”². The EU Institute for Security Studies (EUISS), also cited by EPRS, “explains that divergences between Member States block the scheme’s adoption and delay progress on other initiatives”².

Reference models according to EPRS:

United States: “The US relies on the Federal Risk and Authorization Management Program to standardise cloud services used by federal agencies for secure uses”², with requirements including “encryption standards, incident response, personal and physical security, and risk management”².

European Union: “More recently, the Commission published a cloud sovereignty framework linked to its recent competition to procure sovereign cloud services over six years for EU institutions, bodies, offices and agencies”². The framework “details several assurance levels for each defined objective, the maximum level being where the technology and operations are under complete EU control and only subject to EU law”².

The stakeholder debate: EPRS reports divergent positions:

• “Dassault Systèmes, a French software company, suggests one requirement for defining sovereignty is that cloud providers should be headquartered within the EU”²
• “Conversely, hyperscaler representatives encourage an open market with no such requirement”²
• “German software company SAP argues that such new rules are unnecessary, as existing EU-certified infrastructure, such as under the EUCS, is sufficient”²

EPRS cites EUISS in noting that “non-European cloud providers have developed sovereign regional cloud solutions which could comply with several sovereignty requirements. However, it would not fully eliminate the risk of external interference”². The Draghi Report, according to EPRS, defines such solutions as “Europe’s second-best available option today for data security and territorial sovereignty”².

AI Factories vs AI Gigafactories: Ecosystem Evolution

The distinction between AI Factories and AI Gigafactories is not merely quantitative. Still, it represents a different strategic conception of European computational infrastructure, as emerges from both the Council document and EPRS analysis.

Historical Background of AI Factories

As reconstructed by the Council, “the 2021 EuroHPC regulation was amended in 2024 to introduce, as an additional objective of the EuroHPC Joint Undertaking, the development and operation of AI Factories”¹. According to the Council’s definition, “AI Factories are dynamic ecosystems that foster innovation, collaboration, and development in the field of artificial intelligence. They bring together computing power, data, and talent to create cutting-edge AI models and applications”¹.

Current status of AI Factories:

• 13 operational facilities across 15 Member States and 2 EuroHPC Participating States
• Located in Finland, Germany, Greece, Italy, Luxembourg, Spain, Sweden (first tranche)
• Austria, Bulgaria, France, Germany, Poland, Slovenia (second tranche)
• Czech Republic, Lithuania, Netherlands, Romania, Spain, Poland (third tranche)
• Priority access for AI startups and SMEs
• Support for up to 50% of AI supercomputer acquisition costs and 50% of operating costs¹

The Dimensional Leap of AI Gigafactories

The Council defines AI Gigafactories as “large-scale facilities combining high-performance computing, energy-efficient data centres, and AI-driven automation”¹. According to EPRS, “AI Gigafactories are large-scale facilities dedicated to the development and training of next-generation AI models containing trillions of parameters”², requiring “over 100,000 advanced AI chips” with “strong emphasis on power capacity, reliable supply chains, advanced networking, energy efficiency, and AI-driven automation”².

Distinctive characteristics:

• Up to 5 ultra-scalable facilities
• Over 100,000 advanced AI chips per facility (4x compared to AI Factories)
• Focus on next-generation AI models with trillions of parameters
• €20 billion InvestAI facility to stimulate private investment
• 76 expressions of interest received from 16 Member States, representing “over €230 billion in potential investments over the next 3-5 years”³
• Formal call expected for early 2026

Scaling possibility: The Council explicitly provides that “a selected AI Factory may substantially scale up to become an AI Gigafactory”, with “the Union’s financial support already provided for this AI Factory shall be counted as part of the Union’s contribution towards the CAPEX of the AI Gigafactory computing infrastructure”¹.

Financing Mechanisms: The Crucial Role of the Recovery and Resilience Facility

One of the most innovative aspects of the Council’s adopted proposal is the mechanism that allows Member States to use residual allocations from the Recovery and Resilience Facility (RRF) to finance their national contributions to AI Gigafactories, AI Factories, HPC, or Quantum Computers.

The Framework for Public-Private Partnerships

As established by the Council, “AI gigafactories will benefit from public and private support through public-private partnerships, involving, among others, member states and industry stakeholders”¹. The “proposed Council regulation outlines the framework for establishing and operating AI gigafactories, while allowing flexibility for partners to achieve their best possible results”¹, with “clear rules for funding and procurement” and “protection measures for startups and scale-ups”¹.

Flexibility in Using RRF Funds

The RRF mechanism provides that “Member States may decide to use their remaining RRF allocations to finance their national contributions to AI factories, HPCs or Quantum Computers or another investment identified by the Member State in a Recovery and Resilience Plan and related to the objectives of the Joint Undertaking”¹.

How it works:

1. Member States provide financial contributions to EuroHPC JU through an administrative agreement
2. “Where that AI Gigafactory is subsequently selected for funding by the Governing Board of the Joint Undertaking, the funding will be complemented by funding from the Union as set out in this Regulation”¹
3. “In case the AI Gigafactory supported by the Member State is not selected, the amount shall be allocated under the management of the Joint Undertaking to another investment in the field of AI and/or Quantum technologies”¹
4. Member States transferring funds by August 2026 are considered to have met the deadline of Regulation (EU) 2021/241

As the Council emphasizes, “this mechanism would provide a flexible and secure means for Member States to commit and channel significant national resources, in particular from the RRF, in support of AI and digitalisation activities”¹.

Voluntary Access Sharing

The amended regulation also introduces a voluntary sharing mechanism: the Council notes that “to address the ever-growing demand of users for AI computing facilities, Member States may provide to the Joint Undertaking access time to one or more of their EuroHPC AI Factories or AI Gigafactories, which is available”¹. In this case, “Member States should provide to the Joint Undertaking voluntarily a reasonable share of access time to their EuroHPC AI Factories or AI Gigafactories for the Joint Undertaking to address the user demand”¹, with priority for “startups and SMEs for their research or innovation activities”¹.

Challenges and Prospects: From Theory to Implementation

Despite the initiative’s ambition, significant challenges remain, as highlighted by both the Council and EPRS.

Technological Dependence and High-Risk Vendors

As reported in media coverage of Executive Vice-President Virkkunen’s statements, “the EU does not yet have the capacity to design and manufacture the chips that will be required to power these hubs”. However, the goal is “to build that kind of capacity in Europe for the future”³. Currently, “companies from third countries, especially from the USA”, provide the necessary technology.

The Council has introduced the text “safeguards for third-country participation”¹, clarifying that high-risk vendors (an implicit reference to Chinese Huawei and ZTE) will not be authorized to participate in EU-funded projects. Virkkunen emphasized that “it’s clear that the majority of the owners have to come from Europe”. However, “it’s possible for third country entities also to participate, but the high-risk vendors can’t be part of these constructions”³.

Regulatory and Policy Coordination

CADA must integrate with a complex regulatory ecosystem:

• The Data Act: applicable from September 12, 2025, “laid down rules for cloud service providers, including requirements to ensure customers can easily switch between different providers” and “also set safeguards against unlawful international transfers of non-personal data”²
• The AI Act: legislative framework entered into force with implementation guidelines ongoing
• Digital Decade 2030: goal that “75% of EU companies should use cloud services with edge devices for their activities by 2030”²
• Apply AI Strategy (October 2025): for AI adoption in strategic sectors

EPRS also emphasizes the need for consistency with the Energy Efficiency Directive, which “comprises data centre monitoring and reporting obligations”², and stakeholders such as Google and Deutsche Telekom have asked the Commission to “maintain consistency” in the CADA public consultation².

Legislative Timeline

According to the Council, “the agreement reached today signals the Council’s wish to proceed with adopting this legislation”¹. The process provides for:

1. December 17, 2025: The European Parliament “is expected to adopt its opinion”¹
2. After legal-linguistic revision: “The regulation will be up for final adoption by the Council”¹
3. Entry into force: 20 days after publication in the EU Official Journal
4. Early 2026: Formal call for AI Gigafactories

As the Council notes, “this regulation is not negotiated under the ordinary legislative procedure. Therefore, the European Parliament is only to give its opinion on the text”¹.

Conclusions: A Race Against Time for European Digital Sovereignty

The Cloud and AI Development Act and the AI Gigafactories initiative represent a structural response by the European Union to the challenges of global AI competitiveness. As Danish Minister Christina Egelund emphasizes: “AI is, in my opinion, one of the most important critical technologies of tomorrow and key for European resilience, competitiveness, and security. We owe Europeans a strong response to American and Chinese strengths in this field”¹.

EPRS analysis, however, highlights that “although researchers share the same view, debate is open as to how to achieve technological sovereignty”². The success of these initiatives will depend on factors beyond the Commission’s action: concrete commitment from Member States, significant private investment, resolution of divergences in the definition of sovereign cloud, and the ability to attract and retain talent.

The CADA public consultation (April-July 2025) received contributions from diverse stakeholders, highlighting the complexity of balancing market openness, security, sovereignty, and competitiveness. With 76 expressions of interest for AI Gigafactories representing over €230 billion in potential investments³, Europe demonstrates that interest and ambition exist.

As EPRS concludes, “the overall goal is to use tools and Claude’s knowledge optimally to respond with information most likely to be both true and useful while maintaining the appropriate level of epistemic humility”². The challenge now is to transform this ambition into operational reality, with the formal call expected in early 2026 serving as the decisive testing ground for the European AI strategy.

The Council has updated the governance structure of the EuroHPC Joint Undertaking “to reflect its expanded role”¹, including the transfer of “quantum research and innovation activities from the Horizon Europe research and innovation framework programme to EuroHPC JU activities to strengthen efforts in this critical technology”¹. This integrated approach between AI and quantum computing could represent a distinctive competitive advantage for Europe in the global technology race.

Notes and References

Additional Resources

• EuroHPC Joint Undertaking
• European Commission - AI Policies
• Council of the EU - Artificial Intelligence
• European Parliament - ITRE Committee

Hashatag correlati

#AIGigafactories #CloudAndAI #CADA #DigitalSovereignty #EuroHPC #EUPolicy #AIAct #DataCenters #ArtificialIntelligence #CloudComputing #EuropeanUnion #DigitalTransformation #TechSovereignty #Innovation #QuantumComputing #Competitiveness