A June 2026 review of the European Union’s InvestAI program finds that the Commission’s €200 billion AI pledge is a target for mobilized investment, not money Brussels has already set aside to spend, leaving Europe’s planned AI-gigafactory build-out dependent on member states and private investors.

The Commission launched InvestAI at the AI Action Summit in Paris in February 2025, saying it would mobilize €200 billion for artificial intelligence and create a €20 billion European fund for AI gigafactories. The Commission described those facilities as infrastructure for training very large AI models and said they would give researchers, start-ups and companies access to large-scale compute.

The funding math is narrower than the headline. The review supplied as source material breaks the €200 billion into €50 billion in public money and €150 billion in expected private capital. That private portion is not the same as cash already disbursed to projects, and its final timing, allocation and legal commitments remain open.

For the most compute-heavy part of the plan, the source review says €20 billion is reserved for four to five gigafactories. Under the funding model described in the review, Brussels would cover up to 17% of a site’s investment costs, with the rest coming from member states and private backers. On that reading, the Commission’s direct contribution to the central compute effort would be a few billion euros, not €200 billion.

Compute Gap Meets Funding Gap

The distinction matters because advanced AI systems depend on scarce chips, power, data-center capacity and engineering talent. If Europe’s main number relies on leverage rather than direct spending, the program may not close the compute gap quickly enough for companies trying to train or run large models in Europe.

The scale difference is stark, though the categories are not identical. The source material cites a Financial Times analysis that put 2026 capital expenditure by major U.S. hyperscalers at roughly $700 billion, with Amazon and Microsoft each expected to spend amounts near Europe’s multi-year AI headline in a single year. It also cites the $500 billion Stargate project as another benchmark for the level of private U.S. AI infrastructure spending.

That comparison does not mean public money should match U.S. corporate spending euro for dollar. It does show why the structure of InvestAI matters: a smaller, slower public contribution may help de-risk projects, but it does not by itself solve high energy costs, fragmented capital markets, slower permitting or the loss of technical talent to better-funded markets.

From Summit Promise to Tender

The Commission framed InvestAI as a public-private partnership and said the gigafactories would contain around 100,000 latest-generation AI chips, about four times the AI factories being set up at the time. It also said the fund would use EU budget money to reduce risk for other investors, drawing on existing programs such as Digital Europe, Horizon Europe and InvestEU.

EuroHPC is the public-private body expected to run the infrastructure process. According to the source material, its governing board gave the gigafactory plan initial approval in early June 2026, with the formal tender due in July 2026 and the facilities expected to come online in 2027 or 2028. As of late June 2026, the review says one site, in Norway and powered by hydropower, was under construction, while 19 smaller AI Factories were tied to existing supercomputers.

“mobilise €200 billion for investment in AI”

— European Commission, February 11, 2025 press release

Private Capital Still Unsettled

It is not yet clear how much of the €150 billion private component is binding, how quickly it can be deployed, or which projects will receive it. The final locations, ownership structures, energy contracts, chip supply arrangements and public subsidy shares for the gigafactories have not been fully confirmed.

There is also uncertainty over whether the program can address problems outside the funding pot. The source review points to expensive electricity, fragmented European capital markets, slow approvals and talent outflows as constraints that InvestAI’s headline number does not resolve on its own.

Tender Will Set Sites

The next test is the planned EuroHPC tender in July 2026. Investors, member states and AI developers will be watching for which sites qualify, how much public funding is binding, and whether private backers sign onto the timetable needed to have gigafactories operating in 2027 or 2028.

The main markers to watch are concrete project awards, power and cooling plans, chip procurement, and enforceable co-financing commitments. Until those are in place, the €200 billion figure remains a mobilization target rather than proof that Europe has matched the AI infrastructure spending now led by U.S. cloud companies.

Key Questions

Has the EU spent €200 billion on AI?

No. The Commission’s wording is that InvestAI will mobilize €200 billion. The source review says €50 billion is public money and €150 billion is expected private capital, so the headline is not the same as direct EU spending.

How much is for AI gigafactories?

The Commission announced a €20 billion fund for four future AI gigafactories. The late-June review says the Brussels share of those projects may be much smaller because costs are split with member states and private investors.

When will the gigafactories be ready?

According to the source material, the formal tender is planned for July 2026 and sites are expected to operate in 2027 or 2028. As of late June 2026, the review identifies one site in Norway as under construction.

Why does this matter for European AI companies?

Access to large-scale compute affects whether European labs, start-ups and industrial companies can train or run advanced models without relying mainly on U.S. cloud providers.

What is still unconfirmed?

The binding private funding, site selection, public subsidy shares, power supply, GPU procurement and operating timeline all remain developing details.

Source: Thorsten Meyer AI