Phoenix Technologies AG, Switzerland's leading sovereign cloud and AI provider, today announced the availability of Apertus on its sovereign cloud. Developed by EPFL, ETH Zurich, and the Swiss National Supercomputing Centre (CSCS), Switzerland’s first large-scale, open, and transparent Large Language Model (LLM) can now be deployed on Phoenix Technologies’ high-performance infrastructure. This milestone marks a significant step for Swiss digital autonomy, establishing the country’s first end-to-end sovereign AI solution.
For the first time, Swiss organizations in sensitive sectors such as finance, healthcare, and government can leverage a state-of-the-art LLM without their data ever leaving Swiss jurisdiction. This approach directly addresses the critical business need for innovation while mitigating the risks associated with foreign-owned AI platforms. It provides a trusted path for enterprises seeking to harness the power of generative AI while maintaining full compliance with Switzerland's stringent data protection standards.
"This is a declaration of Switzerland's digital independence. For too long, Swiss enterprises have faced a difficult choice - innovate with foreign AI and risk data sovereignty or fall behind. Today, that choice is obsolete. By uniting the Apertus model with our sovereign, high-performance infrastructure, we are providing the definitive Swiss answer to AI. We are empowering our industries to build the future on a foundation of trust, security, and Swiss values."
The technical foundation of the solution combines fully open-source AI with Phoenix Technologies' enterprise-grade infrastructure. Apertus is available in sizes up to a 70-billion-parameter scale and
is distinguished by its fully documented development process and multilingual capabilities. This model runs on Phoenix's Cloud platform, which is powered by enterprise-grade NVIDIA H100 and next-generation H200 GPUs. The platform is fortified with advanced security features, including confidential computing, which protects data not just at rest and in transit, but also during processing, ensuring an unprecedented level of security for sensitive workloads.