Databricks introduced FILE, a new native column type designed to store documents, images, audio and video directly alongside structured data in lakehouse tables. The feature is currently available in beta.

FILE applies the same governance and fine-grained access controls used for traditional enterprise data to unstructured content, while supporting SQL and Python processing. Databricks is also working to bring FILE support to Parquet, Delta Lake, Apache Iceberg and Spark, expanding its platform for multimodal AI workloads.

Source: databricks.com