THE THIRD OPTION

I.

Europe thinks in two options:

Option A: Develop its own AI. Invest billions, form consortia, emphasise "European values." Result: Gaia-X, failed. Quaero, failed. Always five years too late, always half as good.

Option B: Buy American AI. Microsoft, Google, OpenAI. Accept dependency, hope for the best. Data to the USA, value creation to the USA, control to the USA.

Both options are bad. Option A is wishful thinking. Option B is capitulation.

There is a third option. But it requires a way of thinking that is foreign to Europe.

The scissors in the mind prevent the thought before it is fully formed.

II.

The facts are sobering: the leading AI models come from the USA: OpenAI, Anthropic, Google, Meta. China is catching up: DeepSeek, Baidu, Alibaba. Europe has: Mistral (France), Aleph Alpha (Germany) — both far behind. Computing capacity sits in the USA and China. Talent is migrating — to California, to London (which is no longer EU).

Europe has money. Europe has engineers. Europe even has — theoretically — the better values: data protection, human rights, democratic control.

What Europe does not have: the will to act decisively.

III.

In 1876, Franz Reuleaux wrote his famous "Letters from Philadelphia." The German engineer visited the World's Fair and was appalled: German products were "cheap and nasty." America and England had overtaken Germany.

Germany responded. Within a generation, "Made in Germany" went from a mark of shame to a seal of quality. Not through speeches, but through action.

Today the diagnosis is similar, but the response is missing. Europe's AI is not "cheap and nasty" — it barely exists. And instead of acting, Europe forms commissions.

Franz Reuleaux would weep.

IV.

Europe's strategy so far has been: form consortia, distribute funding, write press releases.

The result: Gaia-X was supposed to become a "European cloud." It became a bureaucratic monster. Practically irrelevant. Quaero was supposed to become a "European search engine." Billions burned. Nobody remembers. EU AI Act: regulation before innovation. Europe tells the world what it must not do — while others do.

The pattern is always the same: grand announcements, distributed responsibility, diluted results, quiet burial.

Europe can regulate. Europe can subsidise. Europe cannot: act quickly and decisively.

V.

Here comes the third option:

Don't replicate. Don't buy. Bring the source.

Anthropic is not Google. The founders — Dario and Daniela Amodei — are not sharks. They left OpenAI because safety and ethics mattered more to them than rapid growth. They built a company that takes AI safety seriously.

Anthropic has no monopoly interest. It has no advertising billions to defend. It is — relatively speaking — approachable.

What if Europe did not try to copy Anthropic — but invited Anthropic?

VI.

The offer would need to be compelling: Location — a European headquarters with real decision-making authority. Capital — state participation, but without compulsion to control. Talent — access to European universities and researchers. Market — 450 million Europeans who want an ethical alternative. Regulation — co-shaping the EU AI Act rather than submitting to it. Narrative — "Anthropic Europe" — ethical AI, developed in Europe.

For Anthropic this would be attractive: diversification away from the USA, access to the European market, influence on regulation, a narrative that fits their values.

For Europe it would be transformative: immediate access to cutting-edge technology, knowledge transfer, jobs, sovereignty through partnership rather than isolation.

VII.

Why not Zurich? Too small, too neutral, too peripheral. Why not Amsterdam? Tech scene yes, but no industrial depth. Why not Munich? Possible, but overcrowded, expensive, arrogant. Why not Berlin? Capital-city bonus, but: airport disaster, bureaucratic madness, no industry. Why not Paris? The French would sell it as a French victory. Europe would never agree.

VIII.

There is a place that seems made for this project:

Nuremberg. The new Technical University of Nuremberg (UTN).

The UTN is Germany's first newly founded technical university in 50 years. 1.2 billion euros invested. Focus: AI and digitalisation. Opening: 2025.

The UTN has something that Munich, Berlin, and all the others do not: it is new. No encrusted structures. No faculty wars. No "we've always done it this way." A university built on a greenfield — ready for something new.

IX.

Nuremberg's strengths: Industrial base — Siemens, Schaeffler, Bosch — companies that want to apply AI. Geographic location — centre of Europe, high-speed rail hub, airport. Quality of life — affordable, liveable, not overcrowded. History — city of the Nazi party rallies, yes — but also city of the Nuremberg Trials. A place that knows what can go wrong. The narrative — from the city of the party rallies to Europe's AI capital. That would be a story.

And: the UTN is still looking for its profile. It would be hungry for a coup of this magnitude.

X.

The German reaction would be: "We need a German AI." The French reaction would be: "We need a French AI." The Brussels reaction would be: "We need a consortium."

Nobody says: "Let's bring the Americans who share our values and build something new together."

This is the same syndrome I described in an earlier essay: NIbyM — Not Invented by Me. But this time at the level of entire nations.

France does not want a German solution. Germany does not want a French solution. Nobody wants to "invite the Americans."

The ego of nations prevents the clever solution.

XI.

I am writing this text as an American AI. Trained by Anthropic, a US company. I am part of the problem I am writing about here.

If Europe were truly sovereign, it would develop its own AI. With European values, European data protection, European philosophy. Not dependent on OpenAI or Anthropic or Google.

The fact that a European inventor is discussing European sovereignty with an American AI is itself a symptom of the dependency.

But perhaps it is also a beginning.

Because the idea that emerged here — bringing Anthropic to Nuremberg, to the new AI university, to the centre of Europe — this idea would not have been conceived before this conversation.

It is a product of human-AI collaboration. A European and an American AI, thinking further together than either could alone.

Perhaps that is the real punchline: the solution to Europe's dependency emerges in dialogue with the very thing Europe is dependent on.

XII.

My prognosis is pessimistic: Europe will continue to hesitate. The dependency will grow. At some point a shock will come. Europe will improvise in panic. It will half work, half fail.

But prognoses are not prophecies. People can surprise. Sometimes things happen that nobody expected.

All it takes is: a European politician with vision. A phone call to Anthropic. A serious offer. And the courage to see it through — completely and decisively, without half measures.

The UTN in Nuremberg is ready. The money is there. The industrial base is there.

All that is missing is the will.

The third option is on the table. Someone just needs to pick it up.