Germany drives sovereign AI infrastructure as US limits access to advanced models

German authorities and industry groups are accelerating programmes to create a European‑run artificial‑intelligence ecosystem. The federal government announced the creation of the AI Security Institute (AISI) to evaluate high‑risk frontier models, while the KI pioneer Jakob Uszkoreit called for €50 billion to fund European super‑computing capacity and joint data‑center projects with French and Norwegian partners. A joint demonstration by KEEQuant, Collaider and noris network showcased a quantum‑secured AI model that keeps data within trusted German infrastructure, targeting public‑sector, critical‑infrastructure and finance users. The EU’s Frontier AI Grand Challenge awarded the “Europa” consortium access to up to 2.5 % of EuroHPC compute for building an open 400‑billion‑parameter model in all 24 EU languages. At the same time, the United States imposed export controls that barred non‑U.S. entities from Anthropic’s latest models, prompting German officials to warn of Europe’s heavy reliance on American AI compute – only about 5 % of global capacity, according to Ifo Institute chief Clemens Fuest. Security agencies have also flagged rising threats to AI data‑centres from extremist and foreign actors. Collectively, these moves aim to secure European digital sovereignty, expand domestic compute resources, and reduce dependence on U.S. AI technologies.