Manifesto
A European AI laboratory — independent and sovereign.
I. The reckoning
Artificial intelligence has become an infrastructure of power. It decides credit, shapes diagnosis, filters information, assists defence, structures government. Yet this infrastructure is today held by a handful of non-European actors, subject to foreign jurisdictions, opaque in their workings, and indifferent to Europe's law, values and interests.
A technological dependency of this magnitude is not a matter of convenience. It is a matter of sovereignty. A continent that does not master the intelligence flowing through its institutions, its companies and its security no longer fully decides for itself.
Europe has not lost the race. It has not yet truly run it.
II. The refusal
We refuse the notion that Europe must choose between two dependencies. We refuse to let the data of a European patient, citizen, army or bank pass through models we do not control, hosted under laws that are not our own.
We equally refuse the opposite illusion: believing that a gap of several orders of magnitude in capital and compute can be closed by imitating those ahead of us. You do not win by copying. You win by moving the ground.
III. The conviction
Sovereignty is not a constraint added to AI. It is a category that the dominant actors are structurally unable to serve.
A European administration, bank or defence actor cannot — legally or politically — entrust its most sensitive data to models governed by extraterritorial law. Where others see a regulatory burden, we see the foundation of an industry.
Frontier is built on this conviction: the future of AI in Europe will not be a sovereign copy of American models, but an intelligence designed from the ground up for European law, values and interests.
IV. The principles
Sovereign. European data, European infrastructure, European supply chain — and verifiable. No Cloud Act, no black box, no hidden dependency.
Open. We favour open-weight models. Transparency is not a competitive weakness: it is the precondition of trust, and trust is the currency of critical AI.
Aligned with Europe. Built natively for the AI Act, GDPR, NIS2, DORA — not as a compliance layer bolted on afterwards, but as an architecture. European law as the specification, not the obstacle.
Specialised where it matters. We do not seek to do everything better than everyone. We build intelligence of excellence for the sectors where sovereignty is non-negotiable: defence, health, finance, energy, public sector.
Independent. European governance, mission-compatible funding, the freedom to say no. A laboratory serving the strategic autonomy of the continent, not an outside agenda.
V. The ambition
Frontier advances in stages, without overstating what it is not yet.
First, a voice: to make the case for sovereign, documented, credible AI exist within the European debate.
Then, a capability: to deploy for regulated European actors a sovereign, compliant and operational intelligence — and to prove it.
Finally, a frontier: to push specialisation to the highest level of capability attainable, so that Europe holds a state-of-the-art intelligence that is its own.
VI. The call
Strategic autonomy is not decreed. It is built, brick by brick, by those who refuse to delegate it.
Frontier is open to the institutions, companies, researchers and decision-makers who share this demand. Europe has the talent, the data, the law and the reasons. What it lacks is the will to build.
Let us build it.