The European Union has moved decisively in late 2025 to back a new generation of "AI gigafactories" , massive, chip‑dense computing hubs designed to train frontier artificial intelligence models at continental scale. Through a combination of fresh regulation, large public funding commitments and a new partnership with the European Investment Bank Group, Brussels aims to anchor next‑generation AI infrastructure within the bloc rather than relying on foreign cloud and chip providers. This marks one of the most ambitious industrial technology bets the EU has ever made.
These gigafactories will build on an existing network of EU‑funded AI factories and supercomputers, but at a far greater scale and with a more explicit focus on industrial competitiveness and “tech sovereignty”. They are expected to host around 100,000 state‑of‑the‑art AI chips each, four times the capacity of today’s European AI factories, and to be financed through public‑private partnerships under the InvestAI initiative, which aims to mobilise €200 billion of AI investment, including a dedicated €20 billion facility for up to five gigafactories.
What the EU Means by an AI Gigafactory
In EU policy language, an AI gigafactory is not a physical factory for making chips, but a vast data and compute hub optimised for training and serving very large AI models. Each planned site will interconnect high‑performance computing clusters, AI accelerators, massive storage, and specialised networking with advanced software orchestration. The facilities are envisaged as open, shared infrastructure where startups, research labs, corporates and public institutions can access compute capacity that would otherwise be out of reach.
The gigafactories will be layered on top of Europe’s existing EuroHPC supercomputing network and the AI Factories initiative. Thirteen AI factories are already being deployed around leading European supercomputers; the gigafactories represent the next step, concentrating around 100,000 top‑end AI chips per site , roughly four times the density of those current factories. They are intended to power frontier‑scale models for language, vision, multimodal learning and scientific simulation, with an emphasis on European languages, datasets and public‑interest use cases.
Unlike traditional cloud regions that are owned and operated by a single corporate provider, the EU’s AI gigafactories are being framed as multi‑stakeholder, public‑private partnerships. Member states, industry consortia, research organisations and cloud operators will all play roles in design, funding and governance. The regulation under discussion in the Council explicitly ties gigafactories to the legal framework of the EuroHPC Joint Undertaking, giving them a dedicated pillar alongside classical supercomputing and quantum initiatives.
InvestAI and the €20 Billion Gigafactory Fund
The political and financial backbone of the plan is InvestAI, announced by Commission President Ursula von der Leyen in February 2025 at the AI Action Summit in Paris. InvestAI is designed to mobilise €200 billion in AI‑related investment across Europe, using the EU budget as a de‑risking layer to attract member‑state contributions, institutional investors and private capital. A ring‑fenced €20 billion component is reserved specifically for AI gigafactories, which will be funded via public‑private partnerships loosely inspired by collaborative science ventures like CERN.
The Commission’s vision is to finance four to five such gigafactories across the Union, with early scenarios pointing to locations in major digital and industrial hubs such as Germany, France, Spain and Italy, including the possibility of two sites in Germany. Public money is expected to cover roughly 30% of overall costs, with the remaining 70% provided by private operators and investors. This leverage model mirrors the AI Factories programme, where around €10 billion of mixed public funding is projected to unlock roughly €100 billion of private investment in AI‑related infrastructure and services.
To turn this political commitment into bankable projects, the Commission has signed a memorandum of understanding with the European Investment Bank Group. Under this MoU, the EIB and the European Investment Fund will provide advisory services to consortia that responded to an informal call for expressions of interest, helping them refine technical designs, governance models and financing structures a of a formal call planned for early 2026. The EIB Group will then be able to complement EU grants with loans and equity‑type instruments, further lowering risk for private backers.
A New Regulatory Framework for AI Gigafactories
Political support for AI gigafactories is not limited to funding announcements. On 9 December 2025, the Council of the EU adopted its position on an amendment to the regulation that governs the EuroHPC Joint Undertaking, specifically to give it a mandate to establish and operate AI gigafactories and to create a dedicated quantum pillar. This legislative move provides the legal scaffolding for how gigafactories will be procured, owned, operated and co‑financed across participating states and partners.
The Council text stresses flexibility for member states and industrial partners while setting core rules around funding, intellectual‑property rights, access conditions and cybersecurity. It also introduces specific protections for startups and scale‑ups, which have often struggled to secure access to high‑end compute controlled by a handful of global cloud vendors. The idea is that gigafactory resources should not be captured by a small number of incumbents but remain accessible to a broad spectrum of European actors building AI systems under the bloc’s AI Act.
In parallel, the Commission has trailed a forthcoming Cloud and AI Development Act aimed at tripling EU data‑centre capacity over the next five to seven years, prioritising highly sustainable infrastructure. This law is meant to ensure that the broader ecosystem of storage, networking and cloud‑native services can keep up with gigafactory‑scale compute. The package sits alongside other deregulatory proposals , including controversial plans to ease environmental assessments for strategic data‑centre and AI projects , which have sparked a heated debate about how far Europe should go in cutting red tape to compete in the global AI race.
Strategic Goals: Tech Sovereignty, Competitiveness and Public Interest
For Brussels, AI gigafactories are not just another digital investment programme; they sit at the intersection of industrial strategy, technological sovereignty and geopolitical positioning. EU leaders repeatedly argue that without native capacity to train and deploy frontier‑scale models, Europe risks becoming dependent on foreign platforms for critical services in healthcare, energy, manufacturing, defense and public administration. The gigafactories, in combination with the AI Act’s regulatory framework, are meant to give Europe both the tools and the rules for home‑grown, trustworthy AI.
Another central objective is to lower the barrier to entry for European SMEs, startups and research teams. Today, only a handful of well‑capitalised firms can afford to reserve thousands of state‑of‑the‑art AI accelerators for months at a time from commercial clouds. By subsidising shared, open infrastructure, the EU hopes to democratise access to compute, allowing smaller actors to train domain‑specific large language models, scientific models or foundation models tuned to European values and legal norms. This fits into a broader narrative of ensuring that the benefits of AI diffusion are more widely spread across the single market.
Finally, the gigafactories are expected to drive advances in key public‑interest domains. Commission documents and EuroHPC consultations highlight applications in personalised medicine, climate and energy optimisation, advanced materials, agriculture, mobility and space. Many of these areas require multi‑modal, high‑resolution modelling that is beyond the capacity of current EU supercomputers when run at large scale. By consolidating compute and data under shared governance, the EU believes it can accelerate progress on its Green Deal, health‑union and security agendas while also reinforcing its industrial base.
Financing Architecture and EIB’s Role
The memorandum of understanding between the European Commission and the EIB Group, published in early December 2025, is a critical enabling step for the AI gigafactory vision. Under this agreement, the EIB will use its InvestEU Advisory Hub and other instruments to help candidate projects reach bankability: refining cost estimates, environmental and social impact assessments, revenue models and risk‑sharing arrangements. The EIF will explore guarantee and equity structures suitable for consortia that include both infrastructure operators and AI‑native startups.
The EIB’s involvement also signals to markets that gigafactory projects are aligned with the bank’s climate and innovation mandates. While each facility will be energy‑intensive by design, the bank will push for high efficiency, renewables‑heavy power sourcing and advanced cooling solutions, reflecting its broader sustainable‑infrastructure criteria. This is particularly relevant as AI’s carbon and water footprint has become a growing political issue: the bank must reconcile its support for energy‑hungry compute with the EU’s legally binding emissions‑reduction trajectory.
On the financing side, the expectation is that EIB lending will sit alongside direct EU grants under InvestAI, national co‑funding from host states, and substantial private capital. Early Commission estimates suggest that up to five gigafactories could be supported by the €20 billion InvestAI facility, implying multi‑billion‑euro capex per site. The EIB’s participation is designed to crowd in institutional investors by offering them de‑risked tranches of debt and by signalling long‑term policy stability around the gigafactory programme.
Chip Supply, Industrial Ecosystems and Global Competition
Behind the ambitious plans lies a stubborn structural reality: Europe does not currently manufacture the volume or class of cutting‑edge AI chips needed to populate its gigafactories. The Commission openly acknowledges that, at least in the medium term, most of the roughly 100,000 accelerators per site will be imported, largely from US‑designed and Asian‑manufactured supply chains. This dependency sits uneasily alongside the bloc’s rhetoric on strategic autonomy, and it increases the urgency of implementing the EU Chips Act, which aims to double Europe’s overall chip‑production capacity by 2030.
At the same time, the gigafactories are expected to act as anchors for broader AI and semiconductor ecosystems. Co‑located with the compute hubs could be specialised data‑centre operators, model‑development firms, applied‑AI startups, and research institutes. Meanwhile, recent Commission approvals of substantial German state aid for new semiconductor fabs operated by companies like GlobalFoundries and X‑FAB underscore the push to expand Europe’s role in parts of the chip value chain, even if genuine frontier‑node AI chips remain largely imported for now.
Internationally, the gigafactory initiative positions the EU as a more assertive player in the AI infrastructure race, where the United States and a handful of large cloud hyperscalers have so far dominated. Brussels frames its approach as an alternative, more open model: publicly backed, multi‑tenant, and governed under EU data‑protection and AI‑governance rules. Success is far from guaranteed, but if the planned facilities come online on schedule by the late 2020s, they could give European actors a credible home base for training models that compete with , or at least complement , those produced in US and Chinese labs.
Environmental and Governance Controversies
Despite broad political backing, the AI gigafactory push has already triggered controversy, especially on environmental grounds. A recent Commission proposal to streamline permitting for strategic projects , including data centres and AI gigafactories , would allow member states to exempt such developments from mandatory environmental impact assessments. The package also includes other deregulatory steps on pollution reporting and chemicals databases, prompting environmental NGOs to accuse Brussels of a "green rollback" that risks undermining health and biodiversity protections in the name of competitiveness.
There are also governance questions around who will effectively control gigafactory resources. While official documents emphasise openness and fair access, there is an ongoing debate about how to prevent dominant cloud providers or large industrial champions from capturing the lion’s share of capacity via long‑term contracts. The Council’s draft regulation hints at safeguards for startups and scale‑ups, but the details of allocation mechanisms, pricing models and priority rules will be crucial in determining whether the infrastructure genuinely broadens access or simply subsidises a new club of incumbents.
Finally, civil‑society groups warn that the combination of massive computing facilities and looser environmental rules could exacerbate local conflicts over land use, water consumption and energy infrastructure. Host regions will expect jobs and innovation spill‑overs, but they may also face higher pressure on grids, heat‑management challenges, and complex trade‑offs between industrial development and environmental stewardship. How the EU and member states design community‑benefit schemes, transparency requirements and independent oversight could prove decisive for public acceptance of AI gigafactories.
As 2025 draws to a close, the EU’s backing for AI gigafactories is no longer just a rhetorical flourish; it rests on concrete legal proposals, a defined €20 billion funding window, and a partnership with Europe’s public development bank. If the planned calls for projects in early 2026 attract strong consortia and the first facilities move into construction later in the decade, Europe could significantly shift the geography of AI training infrastructure and offer its researchers and companies an unprecedented scale of compute close to home.
Yet the initiative also crystallises some of Europe’s hardest dilemmas: balancing climate ambition with energy‑hungry data centres, reconciling strategic autonomy with deep dependencies on foreign chips, and designing a governance model that truly widens access rather than reinforcing existing power imbalances. Whether AI gigafactories become a cornerstone of a more competitive, values‑driven "AI continent" or an expensive symbol of over‑centralised industrial policy will depend on the implementation choices the EU makes over the next few years.
