With Nicheformer Helmholtz Munich and TUM present the world's first large-scale Foundation Model. Single cell analysis with spatial transcriptomics united - trained on over 110 million cells. This solves a central problem in modern molecular biology: How can molecular profiles of individual cells be correctly interpreted when their spatial context is lost? "With Nicheformer, we are taking the first steps towards universally applicable AI models that represent cells in their natural context," explains Fabian Theis, Director of the Computational Health Center | Helmholtz Munich and Chairman of the Bavarian AI Council. This creates a new basis for virtual tissue models, which can revolutionize our understanding of health and illness - and at the same time, shows how AI in Biomedicine opens up completely new realms of knowledge.