Reversibility is often discussed but rarely built into the architecture from the design phase. Proprietary formats, specific APIs, and technical dependencies: switching providers quickly becomes complex and costly. We offer an alternative to proprietary solutions for the AI data lifecycle.

The goal is to help any company or government agency looking to design AI systems that are reliable, robust, sovereign, and reversible from the outset.

Inria has developed an ecosystem around Scikit-learn—the leading open-source machine learning library—that covers the entire data lifecycle.

This approach gave rise to the P16 project as part of France 2030. P16 aims to develop sovereign digital commons supported by innovative public-private governance.

In addition, the growing role of digital commons in European sovereignty strategies will also be presented, particularly through foundational initiatives developed at the European level, such as the AI Factories program.

In France, the AI Factory France (AI2F) project aims to bring together an ecosystem of startups, SMEs, major corporations, research organizations, universities, schools, data centers, incubators, and investors to facilitate access to cutting-edge computing infrastructure dedicated to artificial intelligence. Coordinated by GENCI, AI2F is deploying a national and European platform designed to address the three major challenges of AI: computing, data, and skills.