The Third Omission

On a single day — 19 March 2026 — three texts appeared in the Handelsblatt, each describing the same problem from a different angle: a major general who wants no American AI; an economist asking who will pay for the most expensive bet in economic history; and a Bundesbank vice-president calling on Europe to achieve independence in its financial systems. All three are right. And all three describe, without naming it, the same structural failure.

I. The Triptych

Major General Jürgen Setzer declares that the Bundeswehr will not use AI models from the United States. It wants to "develop independence" — by 2029. Torsten Riecke describes how the major US tech companies alone plan to invest around 600 billion dollars in AI infrastructure in 2026, and asks who will ultimately foot the bill. Sabine Mauderer, Vice-President of the Deutsche Bundesbank, calls on Europe to finally develop its own infrastructure in the financial sector — payment systems, capital markets, tokenised assets — rather than continuing to rely on US platforms.

Three texts, one blueprint: Europe has recognised the dependency. In military AI. In civilian AI infrastructure. In financial systems. In all three areas the same applies: America acted earlier, scaled faster, and today sits at the control points. Europe watches — and appeals.

II. The Appeal as Substitute for Action

Mauderer's text is well written and factually correct. But it ends with an appeal: to European companies to use European markets; to the state to make the framework more attractive; to all parties to sharpen their awareness. This is the language of institutional hope — not the language of structural change.

BioNTech goes to New York because that is where the money is. That is not a question of awareness or European sentiment — it is a question of liquidity, analysts, ecosystem. This ecosystem does not arise through appeals. It arises through capital, through regulation, through critical mass. And critical mass does not arise through the call to create it — but through the first actor who slams a fist on the table and says: from here on, a different rule applies.

Europe's problem is not a lack of awareness. It is the confusion of diagnosis with therapy.

Setzer has the same structure. He diagnoses the dependency on American AI precisely — and announces the intention to be "war-ready" by 2029. What lies between remains vague. The main partner company, Aleph Alpha, has discontinued the development of its own models. The only credible European alternative — Mistral AI from Paris — is not mentioned in the interview.

III. The Pattern

The pattern is not new. It showed itself in energy policy: Europe recognised the dependency on Russian gas at the latest after 2014 — and acted only in 2022, when the pain became unbearable. It showed itself in the semiconductor industry: the dependency on Taiwanese chips had been known for years — the European Chips Act came when the supply chains had already broken. It showed itself in digital infrastructure: for years there was discussion of GAFA dependency — without a European counterweight emerging.

And now it shows itself in financial infrastructure. Mauderer describes that "almost always" American infrastructure is used when Europeans pay by credit card, smartphone or online service. That is no new insight — it has been the reality for twenty years. The difference from before: today a Bundesbank vice-president speaks about it publicly. That is progress. But it is not yet action.

IV. What Tokenisation Really Means

Mauderer's most interesting point is not the appeal — it is the technological analysis. Stablecoins from the "Magnificent Seven" — Apple, Microsoft, Google, Amazon, Meta, Nvidia, Tesla — would have the potential to reach a billion users in a short time, thereby creating a new level of dollar dependency: no longer merely as reserve currency, but as an everyday means of payment in digital form, controlled by private US companies.

This is no abstract danger. Apple Pay, Google Pay, PayPal — that is already reality. The next step would be an Apple dollar or a Meta coin, seamlessly integrated into existing ecosystems and simply more convenient for the European user than any alternative. Convenience beats sovereignty — that is the real lesson from twenty years of European digital policy.

Tokenised bank deposits, as Mauderer proposes, are the right answer — but only if they are actually used. If European banks offer tokenised deposits that are just as convenient as Apple Pay, competition arises. If they merely place them on the shelf as a product, without compelling usage or at least incentivising it strongly, it remains another well-intentioned concept.

V. The Real Question

Riecke's question — who pays for the most expensive bet in economic history? — has a clear answer when one reads the three texts together: Europe pays. Not only as the end customer of AI services that will be monetised after the dependency phase. But also as the user of American payment systems, American capital markets, American cloud infrastructure. The bill does not come once — it comes permanently, in the form of margins, dependencies, and the structural disadvantage of not being able to set the rules oneself.

The antidote is not technological nationalism. It is not the attempt to build everything from scratch that America has built over twenty years. It is the precise identification of the points at which European independence is critical — and the willingness to not merely appeal there, but to decide.

Mistral AI has understood this. The French Ministry of Defence signed a framework agreement in January 2026 — not a position paper, not an appeal, not a roadmap to 2029. A contract. That is the difference between diagnosis and therapy.

Europe does not have a knowledge gap. It has a decision gap.

VI. The Dead Horse

There is a question that no one asks in the Handelsblatt of 19 March 2026, even though it is the most obvious one: why is Germany trying to ride the dead horse Aleph Alpha instead of turning Mistral into a racehorse together with France?

The answer is as simple as it is embarrassing: because Mistral is French. Arthur Mensch sits in Paris, not in Heidelberg. The framework agreement runs through the French Ministry of Defence. German participation would structurally mean being a junior partner — and that is institutionally, politically and industrially impossible to sell. Not because it would be wrong. But because it would mean abandoning one's own champion and admitting that the champion was never really one.

Aleph Alpha received public subsidies, federal ministries as reference customers, political backing at the highest level. SAP, Bosch, Porsche invested — not because they believed in the product, but because they wanted to be part of the German AI flagship narrative. The result is a company that discontinued the development of its own models in 2024, whose founder has left, whose main investor Bosch sold its stake to the Lidl group — and which nevertheless continues as a Bundeswehr partner, because no one has the courage to name the bad investment publicly.

This is not incompetence. It is the system. Germany thinks in subsidy programmes, not in market structures. It creates viability without competitiveness. It builds flagships that do not sail — and calls it industrial policy.

Mistral has done the opposite: venture capital instead of subsidies. Open models instead of sovereign packaging. Military contracts instead of letters of intent. A valuation of 14 billion dollars instead of political protection. And a framework agreement with the French Ministry of Defence — signed in January 2026, not planned for 2029.

The Airbus model worked. Germany took decades to accept that Hamburg is not Toulouse — and that it is still a European aircraft. In the end it was the most successful industrial project Europe has ever realised together. With AI, Europe does not have that time. The windows are closing faster than subsidy programmes can be approved.

A dead horse does not get faster when you feed it taxpayers' money. And a racehorse in Paris remains a racehorse — even if the stable owner is French.

As long as Germany prefers to subsidise a German champion that does not deliver, rather than strengthening a European champion that does, nothing will change about the decision gap. That is not a technological question. It is a question of ego.