I was in Switzerland when the Swiss AI model, Apertus, launched in September 2025 — giving me a front-row seat to something significant: watching a small nation deploy what might be the most comprehensive approach to AI sovereignty we've seen yet.

Apertus (Latin for 'open') is a 70-billion parameter language model trained on 15 trillion tokens across 1,811 languages. What makes it different isn't the scale — it's the uniquely Swiss approach.

AI as Public Infrastructure

Switzerland made a fundamental choice to build AI as a public utility, not just a commercial product. This academic-led, fully open development is built on four pillars:

Public Governance & Funding. Positioned as a shared national resource, the project received substantial public funding and utilised over 10 million GPU hours on the Alps supercomputer through a national effort by ETH Zurich, EPFL, and the Swiss National Supercomputing Centre.

Multilingual & Sovereign by Design. With 40% non-English training data — including Swiss German and Romansh — the model is designed to preserve cultural and linguistic identity. This avoids vendor lock-in and API dependencies, ensuring true digital sovereignty.

Radical Transparency. The training data, model weights, architecture, and methodologies are all public. This complete auditability is crucial for regulated industries and building public trust.

Compliance-First. Apertus was built from the ground up to meet the requirements of the EU AI Act and Swiss data protection laws, embedding regulatory compliance into its core.

A Global Contrast

Apertus stands in stark contrast to other national strategies:

  • USA: Proprietary models from Big Tech optimised for market dominance
  • China: Centralised development with state oversight and control
  • EU: A regulation-heavy approach focused on compliance
  • India: Public-private partnerships centred on digital public goods
  • Switzerland: Open, academic-led, positioned as public infrastructure

Why This Matters

The Swiss approach is more than technology policy — it's a statement about democratic control over the tools that will shape our future. As geopolitical tensions rise around AI, neutral, transparent alternatives like Apertus become strategically vital.

It offers a powerful template for other nations seeking to preserve their own digital, cultural, and economic sovereignty. The question remains whether this model can scale, but its existence provides a crucial alternative vision for our AI future.

Originally published on LinkedIn · View discussion →