The EU spends billions on AI, but can anyone track the money?

Billions pledged, but operationalisation unclear

At the Paris AI Action Summit in February 2025, European Commission President Ursula von der Leyen announced the InvestAI Initiative, setting a political target of €200 billion for AI investment across the Union. The announcement received wide media coverage. What received far less attention was how little of that sum is currently verifiable. Another hundreds of billions were announced a couple of weeks ago in the Tech Sovereignty Package.

Open Future’s report ‘From Frameworks to Factories: Mapping EU AI Investment Architecture’ points out that the headline figure conflates several very different types of financial commitment: legally binding budget ceilings, EU budget guarantees channelled through implementing partners such as the European Investment Bank (EIB) Group, and political mobilisation targets that presuppose large volumes of private capital that have not yet materialised. For example, the €50 billion InvestAI Facility’s structure is still being finalised, with no private capital raised.

This is not an isolated problem. As the Organisation for Economic Co-operation and Development (OECD) noted in a 2025 working paper, AI investment is “clouded by fragmented data, inconsistent methodologies, and a lack of standardised measurement frameworks.” News headlines and corporate press releases frequently cite ambitious figures that are difficult to verify or compare.

A fragmented funding architecture that resists scrutiny

The EU funds AI through a layered system spanning multiple budget lines, joint undertakings, and financial instruments. Each of them combines different reporting rules and levels of transparency. The report maps these layers, covering programmes such as Horizon Europe, the Digital Europe Programme, the InvestEU Fund, and the European Defence Fund, as well as the EIB Group’s role as the EU’s main implementing partner for blended finance.

Several structural obstacles prevent systematic tracking. AI spending rarely appears as a standalone budget line: it is folded into broader digitalisation or innovation categories. There is no consistent AI-tagging system across EU programmes. This gap was flagged by the European Court of Auditors (ECA) in its 2024 special report, only for the European Commission to reject the recommendation. Data on public money flowing through blended finance vehicles, such as the InvestEU Fund or the TechEU Platform, often lacks end-beneficiary details, citing commercial confidentiality or security concerns.

On the private investment side, transparency is even thinner. Commercial databases such as Crunchbase, PitchBook, and Dealroom provide the most granular available data, but they rely on self-reporting, apply differing methodologies, and charge enterprise-level subscription fees that place them out of reach for most civil society organisations. For the EU AI Champions Initiative — a consortium of over 110 organisations behind a €150 billion private investment pledge — scrutiny requires piecing together press releases, as investor-by-investor allocations are not publicly disclosed.

Why this matters for digital rights

The EU spent a decade building what was widely regarded as the world’s most ambitious regulatory framework for digital technologies, culminating in the AI Act. The report documents how, by 2025, the emphasis had shifted: from AI frameworks to AI factories, and from governing AI to building it, with a vocabulary of regulatory ‘simplification’, competitiveness, and sovereignty replacing the earlier focus on ethics, fundamental rights, and transparency.

This shift is consequential. As an increasing share of EU AI policy is conducted through funding decisions rather than legislation, those decisions become harder to scrutinise. Funding choices determine which types of AI get built, by whom, for whose benefit, and under what conditions. Without robust transparency and accountability mechanisms, it is impossible to assess whether EU AI investment aligns with stated policy priorities, whether benefits are concentrated among a small number of powerful actors, or whether public money translates into public value.

Open Future, a European think tank and EDRi member, argues that the opacity surrounding AI investment must be overcome as the EU’s ambitions increase exponentially.

Following the money

The report identifies several steps that would enable meaningful public scrutiny of EU AI spending. At the EU institutional level, the European Commission should revisit the ECA’s recommendation to introduce a standardised AI-tagging system deployable across diverse funding instruments. Greater transparency is also needed regarding end beneficiaries of blended finance vehicles, with fewer carve-outs for commercial confidentiality.

For major public-private instruments such as the AI Champions Initiative, accessible, auditable ledgers of commitments and disbursements would allow civil society and journalists to distinguish genuine new investment from repackaged prior commitments. More standardised disclosure requirements for private capital flows to AI markets would complement these public-sector measures.

In parallel, Open Future’s Steering AI Investment activity will continue to develop the analytical tools and datasets needed to make EU AI funding flows more legible. This includes building baseline estimates of EU spending across strategic priority areas, mapping the concentration of public grant beneficiaries, and cross-referencing private capital data to identify who stands to benefit from the EU’s industrial turn — and who does not.

Contribution by: EDRi member, Open Future