Twenty euros a month. That is what it costs today to access AI tools that three years ago would have required entire teams of analysts, writers or programmers.
A tempting price. Too tempting. Because it has nothing to do with the real cost of what lies behind it: the data centres, the energy, the engineers, the years of research. It is a penetration price. The same one any supplier uses to create dependency before adjusting the terms.
Meanwhile, companies do their sums in silence. Why hire when AI does the work? Vacancies go unfilled. Departments slim down. Margins improve. The trap closes slowly, and what is more, with the enthusiastic complicity of those being trapped.
The problem is what comes next. A company that integrates AI into its core processes — customer service, document management, decision-making — cannot unplug it the way you switch coffee brands. If tomorrow its licence is denied, it does not lose a tool. It loses the ability to function. It fails within weeks.
And this is where the conversation stops being about technology and becomes a conversation about power.
What happens if a major American tech company decides — or is pressured — not to renew licences for European companies in a moment of political tension? What happens if the administration of an EU country loses access to the models with which it manages healthcare, justice or security?
This is not science fiction. It is the usual logic applied to a new resource. We lived it with gas. We saw it with semiconductors. Every time Europe has outsourced control of something critical, it has ended up paying a disproportionate price when circumstances changed. And circumstances always change.
We are still treating AI as if it were a Spotify subscription. But AI is infrastructure. It is the layer on which the economy and public administration will be built for the coming decades.
Europe has world-class universities, engineers, data and a market of four hundred million people. It has everything it needs to depend on no one in something this critical.
What I cannot understand — and would like someone to explain to me — is how it is possible that to this day there is no European consortium working on a shared AI model. France, Germany, Spain, the Netherlands, Sweden: advanced economies, with technical talent and much to lose if this goes wrong. Where is that project? What exactly are we waiting for?