A central store that keeps raw data of any kind in its original format until a team or a model needs it.
Data Lakes hold much of the data that AI systems are trained on and retrieve from. Unlike structured data warehouses, lakes ingest raw telemetry, PDFs, videos, and logs. Many architectures use a Data Lake as the source that continuously feeds unstructured data into Vector Databases for RAG.
See also Data Mobilization, Data Lakehouse and Vector Database.