AI Gigafactories Initiative

AI Gigafactories Initiative is sponsored by European Commission. Supports large-scale AI computing infrastructure, relevant to AI governance research.

Public funding for AI gigafactories Public funding for AI gigafactories To enable European businesses to use artificial intelligence (AI) in a trustworthy manner and remain internationally competitive, the European Commission is working to secure a leading role for Europe in the field of artificial intelligence and aims to make Europe an “AI continent”.

To this end, the European Commission is pursuing a comprehensive strategy, which is primarily set out in its 2024 AI Innovation Package, the Competitiveness Compass 2025, the AI Continent Action Plan and the Apply AI Strategy.

The AI Continent Action Plan includes, among other objectives, the creation and expansion of (public) AI infrastructure, for which the European Commission is currently aiming to mobilise various investments totalling €200 billion through its InvestAI initiative. The investment programme also provides for €20 billion to be made available for an investment fund to support the establishment of four to five AI gigafactories.

AI gigafactories are large-scale computing and development centres, specifically designed for the training, operation and further development of AI. In Germany, the European Commission’s investment plans have attracted great interest both from businesses and various public institutions.

The federal government had already announced in its coalition agreement its intention to bring at least one of the European AI gigafactories to Germany (items 2193 ff.) In this article, we summarise what is already known about the selection process and funding modalities. We take a closer look at the possible funding condition that AI gigafactories must be implemented through public-private partnerships ("PPP").

The European Commission and the EuroHPC Joint Undertaking (EuroHPC JU), a public-private partnership founded in 2018 to significantly expand high-performance computing capacities in Europe and enhance Europe’s technological competitiveness, are currently preparing the funding of AI gigafactories.

Among other things, they published a call for expressions of interest on 9 April 2025, to which industry representatives, public and private investors and EU member states could respond until 20 June 2025. This non-binding consultation helps the European Commission and EuroHPC JU to identify interested consortia and collect initial contributions for possible implementation.

However, before the official call, further steps remain on the part of the European Commission and EuroHPC JU: securing the budget, amending the EuroHPC regulation, defining the InvestAI facility and ensuring compliance within the EU State aid legal framework. The required legal framework for establishing a fund to support AI gigafactories and for the design of funding conditions is not yet in place.

Nevertheless, previous communications from the European Commission allow initial conclusions about the selection of funding recipients and some funding modalities.

At the AI Gigafactories workshop on 22 May 2025, it was announced that there will be a three-stage selection procedure, overseen by the EuroHPC JU: initially, an independent expert panel will evaluate the applications received, followed by a due diligence review conducted by a financial institution. Subsequently, a selection decision will be made by EuroHPC JU with guidelines on contracts, financing and procurement rules and procedures.

The European Commission’s documents for the AI Gigafactories workshop from 22 May 2025 already provide an initial insight into possible selection criteria for funding. The selection will take into account: the technical part of the submitted proposal, the potential impacts and the financial feasibility.

The European Commission further breaks down the individual criteria, with the technical part of the proposal receiving quantitative emphasis.

Additional subcriteria include: “objectives and technical quality of the proposal,” “quality of the work plan,” “quality of physical, IT and network infrastructure,” “sustainability and energy efficiency,” and “experience and know-how of the consortium in setting up similar large-scale facilities”.

Regarding the criterion of potential impacts, the European Commission further distinguishes between “quality of services, including security and trustworthiness” and “EU added value and impact on the European AI ecosystem.

” For financial feasibility, subcriteria include “investment commitments of Member States / public sector and private partners” and “quality and financial viability of the proposed business model (due diligence review).

” Even if the European Commission already lists possible selection criteria, the published documents, subject to the final criteria, only allow limited conclusions to be drawn about the design of the applicant's own funding application. On the one hand, the weighting of the individual criteria and sub-criteria remains uncertain. Secondly, the sub-criteria in the technical section are formulated in a comparatively open manner.

Public-private partnership as a funding condition In various communications, in particular the AI Gigafactories Workshop of 22 May 2025 and the AI Continent Action Plan of 9 April 2025, the European Commission indicates that AI gigafactories are primarily intended to be implemented by PPPs. This requirement is justified by the fact that the level of investment required cannot be realised by private individuals alone.

However, this appears not to have been definitively decided yet. Indications of flexibility between PPPs and conventional funding can be found in the Call for Expression of Interest in AI Gigafactories dated 9 April 2025, which states that funding may be provided in the form of a PPP. PPPs can take very different forms.

One possibility is the establishment of a joint project company, but cooperation based on contractual arrangements is also feasible. The involvement and influence of the public partner tend to be relatively limited, for example, in the operator model, where the private partner operates the project alone and bears the economic risk.

More extensive opportunities for influence, particularly in the form of an obligation to provide, exist in models such as the concession model, where planning, construction, financing and operation are handled by the public sector, but the economic risk is predominantly borne by the private partner.

Another conceivable option is a company model, in which planning, construction, financing and operation are managed through the project company.

In an example cited by the European Commission in the AI Continent Action Plan, a public-private partnership (PPP) will exist even if the European Union and, where applicable, the Member States as public partners provide direct grants, while private applicants are responsible for financing the remaining amount as well as the construction, planning and operation of the relevant AI gigafactory.

However, it remains questionable whether concluding a PPP is actually necessary in this case, given that the PPP model outlined by the European Commission is, at least economically, fundamentally similar to conventional funding through a grant agreement or funding decision. By contrast, funding via grant agreement or decision is more established in practice and less complicated.

A PPP model would only be justified if the European Commission anticipates that the public sector can secure specific (particularly operational) rights of influence through the PPP that it would not be able to obtain under “traditional” funding via a grant agreement or funding decision.

The requirement that AI gigafactories be implemented through PPPs can, depending on the design of the PPP, also complicate matters for private companies because the public partner in a PPP is subject to special public-law constraints: for example, budgetary or public procurement rules, but also public-law disclosure obligations or fundamental rights restrictions, which may be relevant in relation to future users of the AI gigafactory.

In particular, where cooperation occurs through a joint project company to be established (a “joint undertaking”), these special constraints must be reflected contractually. In order to ensure that the project is implemented on time, the potential partners should endeavour to reach an agreement as early as possible.

At present, it is not yet possible to definitively say what the selection process and funding modalities for AI gigafactories will look like. The current status of the InvestAI initiative shows that a successful application will depend on farsighted planning and close coordination with public authorities.

It is likely to be in the interest of private investors that entering into a public-private partnership (PPP) is not made a condition for funding. Instead, established and well-tested funding mechanisms (funding decision, public-law contract) should be used.

The significant investment requirement alone is not in itself an argument for making a PPP a funding condition since substantial public funding through decisions is also granted in many other areas. We would like to thank Finn Knoblauch for his assistance in writing this article.

Max Helleberg +49 30 20942039 Regulatory and Governmental Affairs Labour law reform 2027: What employers can expect Digital Omnibus: AI is nearing the finish line – is EU data law heading down the wrong track? Europe on the way to technological sovereignty?

The EU’s Tech Sovereignty Package More EU-level steering, new selection criteria, new obligations: the EU proposal on the 2 GHz MSS band and its implications from 2027 onwards “What does the German Public Procurement Acceleration Act mean for planning and design consultancies?

” – interview in competitionline on 1 June 2026 Dusseldorf Higher Regional Court refers Armed Forces Planning and Procurement Acceleration Act to Federal Constitutional Court: is removing suspensive effect of appeals unconstitutional?

New rules for allocation of the scarce resource “grid connection capacity” German parliament passes law to speed up the awarding of public contracts – Bill to speed up public procurement – Noerr Insight No. 7 U.S. Enforcement: Uniform Policy on Corporate Voluntary Self-Disclosures and Mitigation of Penalties