Deutsche Telekom has secured a contract with German Leibniz University Hannover worth tens of millions to provide a sovereign large language model (LLM), trained and operated entirely in Europe.

In a push to create sovereign European alternatives to AI technologies from the U.S. and China, the project is funded by the German Federal Ministry for Economic Affairs. It also has scientists from six German research institutions, two startups collaborating, along with the German applied research organization Fraunhofer Society, the German AI Association, and the Center for Sovereign AI (CESAI).

Deutsche Telekom will use its Industrial AI Cloud to provide Leibniz University Hannover with the technical infrastructure for the SOOFI (Sovereign Open Source Foundation Models) research project in the AI factory. The goal of SOOFI is to develop a powerful and sovereign LLM with around 100 billion parameters in a move to outperform and replace the AI model Teuken 7B, previously dubbed a European alternative to American tech giants.

Nicolas Flores-Herr, site lead at Fraunhofer IAIS, said: “SOOFI brings together leading research teams from Germany to train large open-source LLMs, which represent an important area of a sovereign AI value chain. These are to be used, for example, in industry (small and medium enterprises) SMEs, and the public sector to strengthen the development of an independent European AI infrastructure.”

Deutsche Telekom’s subsidiary T-Systems is also participating, providing its Industrial AI Cloud for training the sovereign LLM. The associated data center is connected to four 400 GB/s fiber optic connections to meet the highest standards of data protection, security, and reliability.

Building on this, from March 2026, a network of approximately 130 Nvidia DGX B200 systems, featuring a total of over 1,000 GPUs, will operate exclusively for the European language model SOOFI.

“With SOOFI, we are laying the foundation for the next generation of European AI models – sovereign, powerful, and entirely in European hands,” Wolfgang Nejdl, professor at the L3S Research Center at Leibniz University Hannover, said. “Large AI models that respect European values are essential for building trust in AI, especially in sensitive areas such as education, medicine, administration, and production.”

Deutsche Telekom and its T-Systems arm have been making strides to meet the growing digital sovereignty demand. In September, the latter became one of the first vendors to appoint a chief sovereignty officer. A month later, T-Systems launched an AI-driven cloud migration service aimed at addressing failed digital transformation efforts by companies that have moved workloads to the cloud without seeing promised efficiency gains.

This latest development comes weeks after Deutsche Telekom and Nvidia teamed up to build a data center in Munich, Germany, jointly funding the $1.2 billion project, which will have enterprise software company SAP as a customer.

With more details revealed in October, the duo announced their agreement to launch an AI cloud for industrial clients, planned to go live in the first quarter of 2026. The cloud is set to operate from a refurbished Munich data center with up to 10,000 Nvidia Blackwell GPUs, where SAP will provide the software stack.

As part of the AI cloud arrangement, companies will be able to book computing power on demand to develop industrial AI applications, as the cloud will also serve public services and defense sectors.

The combination of the AI factory currently in creation and the AI Continent Action Plan, outlined earlier this year, is poised to elevate the EU’s position as a stronger AI leader. These developments are all seen as stepping stones to creating an independent and robust ecosystem in Germany and Europe.