The headline is impressive: €200 billion for artificial intelligence. That’s how the European Commission sells its InvestAI programme, and that’s how it gets quoted — as Europe’s answer to the torrent of money pouring out of Silicon Valley.
It’s worth looking closely at a single word. The Commission doesn’t say it will spend €200 billion. It says it wants to “mobilise” €200 billion. That isn’t a detail. That’s the whole story.
Because when you work through what’s actually behind the number, the offensive shrinks dramatically between the headline and the effect. What begins as a €200 billion headline ends as a comparatively small, late, and partly hypothetical cheque — written while the problems that actually caused Europe’s lag are left untouched.
Mobilised, not spent
The EU is selling a €200 billion AI offensive. But the decisive word is “mobilised” — not “spent.” Work through the number and the headline shrinks dramatically before it reaches any effect.
2027–28 data centres expected to run
1 SITE under construction so far (Norway)
Late, slow, and not yet built.
A small, late, partly hypothetical cheque — without touching expensive energy, fragmented capital markets, slow permits, or the talent drain. The EU mistakes a funding pot for a strategy.
The word that matters
In Brussels usage, “to mobilise” means: put in a little public money and hope private investors add a multiple of it. With InvestAI, the split looks like this:
- €50 billion is real public money.
- €150 billion is hoped-for private capital — money that isn’t there yet and hasn’t been committed.
The target is a leverage ratio of roughly 1:10: each public euro is meant to pull in ten private ones. On paper, an elegant mechanism. In reality, that private growth capital is precisely what Europe lacks most — there is no deep, unified capital market, and pension funds avoid venture risk. So the €150 billion isn’t merely uncertain; it is exactly the kind of money whose absence explains Europe’s AI gap in the first place. The Commission is hoping the market will deliver what the market in Europe structurally does not.
Already, of the €200 billion, only a quarter survives as real public money.
SQL Server 2025 Unveiled: The AI-Ready Enterprise Database with Microsoft Fabric Integration
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
And only a fraction of that for compute
Keep going down. Of the €50 billion in genuine public funds, €20 billion is reserved for four or five so-called AI gigafactories — the large, purpose-built training facilities meant to finally give European researchers and start-ups access to the compute now concentrated in the United States.
But even those €20 billion aren’t carried by Brussels alone. Under the funding model, the EU covers only up to 17% of a facility’s investment cost; the rest has to come from member states and private backers. What actually remains as the Commission’s own firmly committed contribution is therefore a single-digit billion sum — a few billion euros for the heart of the entire offensive, the compute.
The sequence is sobering: €200 billion in the headline → €50 billion in real public money → €20 billion for compute → a few billion that Brussels truly commits itself. Big in the headline, small in the effect.
Data Center Cooling Engineering: Data Center Cooling Engineering
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Late, slow, and not yet built
On top of the small sum comes the timing. Even this money isn’t flowing today.
The formal call for the gigafactories doesn’t open until July 2026; the responsible EuroHPC board only agreed to the plan in principle in early June 2026. The facilities are expected to come online in 2027–2028. So far, a single site — in Norway, on hydropower — is under construction. Alongside it run 19 much smaller “AI Factories” using existing supercomputers.
Translated: in a race whose pace is measured in quarterly numbers, Europe in mid-2026 is presenting mainly a funding structure and a call for tenders — silicon follows two years later. Late, slow, and not yet built.
INFINIBAND STORAGE AT SCALE: RDMA FOR ENTERPRISE SAN AND PARALLEL FILE SYSTEMS: Deploy Lustre, GPFS, and NFS over RDMA for high-throughput data centers.
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The comparison that hurts
Set the yardsticks side by side, all of them sourced, none inflated.
The four big US hyperscalers — Amazon, Microsoft, Alphabet, Meta — are spending around $700 billion in capital expenditure in 2026 alone, up 77% from 2025. Amazon is planning roughly $200 billion, Microsoft about $190 billion — each, in a single year. The Stargate project alone is budgeted at $500 billion. Microsoft is also sitting on an $80 billion backlog of cloud orders it can’t fill for lack of power.
Now the scale: a single large US company invests roughly ten times as much in one year as Europe’s entire, multi-year gigafactory pot of €20 billion. The four together come to thirty or thirty-five times it. Stargate alone is twenty-five times. And while Europe negotiates over 17% funding rates, Microsoft is building a single data centre in Portugal for $10 billion — half of Europe’s flagship budget, from one company, on European soil.
AC/DC Adapter for AI Prime HD+ Aquarium LED – AquaIllumination JYH32-2402500 10136 Power Supply Cord Charger PSU
Brand New, High Quality Replacement Cord
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
A funding pot is not a strategy
You could forgive all of this if it were the first step toward something bigger. The real problem is that the cheque — small and late as it is — doesn’t touch the causes of the lag at all.
Europe’s AI weakness doesn’t stem from some missing €20 billion pot. It stems from electricity prices that, according to the regulator ACER, run roughly double the US level for industry; from grid connections and permits that take years; from fragmented capital markets with no deep late-stage funding; from departing talent that goes where the compute, the salaries, and the capital are; and from a dependence on US clouds for which, by the Commission’s own figure, Europe wires around €264 billion abroad every year.
InvestAI solves none of these. The accompanying “Technological Sovereignty Package” of 3 June 2026 consists mostly of laws and frameworks — a Chips Act revision, a Cloud and AI Development Act whose centrepiece is a procedure to assess your own dependency, an open-source strategy, an energy roadmap. The ~€100 billion some attach to it is largely the same InvestAI money under a new heading, not additional.
The most honest voice is, unintentionally, the Commission’s own: Ursula von der Leyen conceded that taxpayers cannot foot this bill alone — Europe “urgently” needs private capital. Nobody says that about money they have.
The fair other side
A serious critique has to allow the best counterargument — here it is.
Public seed money that leverages private capital is a legitimate, standard model; no state funds infrastructure alone, and the US uses “mobilised” figures too. The gigafactories are real, the call is imminent, one site is already under construction — this is not vapourware. And doing something beats doing nothing.
All true. None of it changes the diagnosis. The 1:10 leverage is a target, not a fact — and it presupposes exactly the private capital whose absence is the problem. The timeline to 2027/28 is too slow for this race. And even if every hoped-for euro arrived, it would touch none of the causes. A funding pot finances facilities. It doesn’t change electricity prices, capital markets, permitting timelines, or the talent drain.
Bottom line
Europe writes a comparatively small, late, and partly hypothetical cheque and calls it an offensive. The €200 billion headline shrinks, on inspection, to €50 billion of real money, of which €20 billion is for compute, of which a single-digit billion sum is what Brussels actually carries itself — payable from 2027, against an opponent that builds ten times the entire pot every year.
This isn’t only a question of size. It’s a confusion of categories. Money can be a tool of a strategy; it is not one. As long as energy, capital, permits, and talent stay as they are, the sentence that sums up this whole episode holds:
The EU mistakes a funding pot for a strategy.
Sources: European Commission and EuroHPC (InvestAI; gigafactory funding model; Technological Sovereignty Package, 3 June 2026; €264bn dependency figure); ACER (2026) on industrial electricity prices; FT-compiled 2026 hyperscaler capex; von der Leyen’s statement on the need for private capital. Figures reflect the state of play as of late June 2026; quotes are paraphrased. Analysis and opinions are the author’s.
