A late-June review of the European Union’s InvestAI plan shows that the bloc’s EUR 200 billion AI offensive is still mainly a target to mobilise investment, not a confirmed public spending package, leaving the core gigafactory buildout dependent on private capital, member-state funding and procurement that has yet to produce operational large-scale AI training sites.

The Commission announced InvestAI at the Paris AI Action Summit on February 11, 2025, saying the initiative would mobilise EUR 200 billion for AI and include a new European fund of EUR 20 billion for AI gigafactories. That wording is central: mobilisation describes a target for public and private investment, not money already disbursed from the EU budget.

The funding breakdown supplied for this report puts public money at about EUR 50 billion, with the remaining EUR 150 billion depending on private investment. It also narrows the compute piece to EUR 20 billion, meant for up to five AI gigafactories. The Commission’s AI Factories policy page says those facilities would combine more than 100,000 advanced AI processors with power capacity, supply chains and high-speed networking.

Even the EUR 20 billion gigafactory figure is not the same as a direct Commission outlay. EuroHPC guidance says member-state commitments are required before a call, and refers to a 17% cap on eligible IT capital costs. As of late June 2026, EuroHPC’s public gigafactory calls page listed no open call, while the source material says tender activity is expected in July 2026 and operations are expected in 2027-28.

Compute Gap Meets Funding Gap

The issue for European AI firms is access to compute. Frontier model training is limited by chips, data centers, power and capital; startups without large cloud budgets depend on shared infrastructure or foreign providers. If InvestAI arrives mainly as leveraged finance, European companies may wait years for capacity while US providers keep spending directly.

The scale gap is large. Market reports citing Financial Times data put 2026 capital spending by Amazon, Microsoft, Alphabet and Meta around $700 billion to $725 billion, much of it for AI and cloud infrastructure. OpenAI and SoftBank said Stargate intends to invest $500 billion over four years. Those figures are not like-for-like public programs, but they show the pace and size of private infrastructure spending Europe is trying to answer.

From Paris Summit To Procurement

InvestAI was announced three weeks after the US Stargate announcement and during the Paris AI Action Summit, where European leaders tried to recast AI policy around infrastructure and adoption as well as regulation. The existing AI Factories program is already moving: the Commission says 19 AI Factories and 13 antennas are operational, using or extending EuroHPC supercomputing resources for startups, researchers and public-sector users.

The gigafactories are a different tier. The Commission describes them as large-scale sites for training next-generation AI models with over 100,000 advanced AI processors. That is why the procurement timetable, power availability and national co-financing are central to whether the headline becomes usable compute.

“to mobilise EUR 200 billion for investment in AI”

— European Commission, February 11, 2025

Private Capital Still Missing

Several facts remain unsettled. The final mix of EU, member-state and private financing has not been locked for each gigafactory; the locations, consortia, procurement model and delivery schedule depend on formal calls and awards. It is also not yet clear whether enough private capital will appear on the timetable assumed by the EUR 200 billion headline.

The wider constraints are also open. The source analysis says the plan does not directly solve Europe’s high energy costs, slower permitting, fragmented capital markets or AI talent outflow. Those are policy and market conditions, not line items that disappear once a fund is announced.

Tender Decisions Set The Pace

The next test is procedural and financial: EuroHPC must publish the gigafactory call, member states must provide commitments, and bidder consortia must show how they will finance, build and power the sites. Award decisions will clarify how much of the EUR 20 billion gigafactory fund turns into contracts and how much compute capacity European users can actually access.

If the current timetable holds, the first large facilities would not affect Europe’s compute supply until 2027 or 2028. Until then, InvestAI’s main test is whether the promised leverage turns into signed financing and construction rather than remaining a headline number.

Key Questions

Is the EU spending EUR 200 billion on AI?

No. The Commission says InvestAI will mobilise EUR 200 billion. That includes public money and expected private investment; it is not the same as a EUR 200 billion EU budget line.

How much is aimed at AI gigafactories?

The Commission has announced a EUR 20 billion European fund for up to five AI gigafactories. The source analysis calculates that Brussels’ direct share is only part of that figure because national and private co-financing are needed.

Are AI Factories and AI gigafactories the same?

No. The AI Factories program links existing or upgraded EuroHPC supercomputing centers; the Commission says 19 AI Factories and 13 antennas are operational. Gigafactories are planned as larger training sites with more than 100,000 advanced AI processors.

When could the gigafactories start operating?

The source material points to 2027-28 for operations, after procurement. That schedule remains subject to formal calls, awards, site readiness, power connections and financing.

Why does this matter outside Brussels?

Compute access affects whether European startups, researchers and industrial firms can train or adapt advanced AI models without relying mainly on US cloud platforms. Delays or funding gaps could leave that dependency in place longer.

Source: Thorsten Meyer AI