A large machine learning model trained on broad, diverse data, which can be adapted to a wide range of downstream tasks.
Foundation Models are the general-purpose base layer for Private AI systems. Training one from scratch is expensive, so enterprises typically download open-weight Foundation Models and adapt them to their own domains through fine-tuning or RAG.
See also Private AI, Digital Transformation and Enterprise Architecture.