Nov 18, 2026: Schema detection and evolution for structured types (General availability)

With this release, Snowflake can create structured OBJECT, ARRAY, and MAP columns from the nested data in your Parquet, JSON, Avro, and ORC files. When you load these files with COPY INTO or Snowpipe, schema evolution adds new structured columns and fields as your data evolves.

Key use cases include:

  • Create tables with structured columns from your files: INFER_SCHEMA infers the structured column definitions, so you don’t write them by hand.
  • Add new nested columns and fields as they appear: Schema evolution adds new nested columns as structured columns and adds new fields to existing structured columns.
  • Evolve nested columns in Snowflake-managed Apache Iceberg™ tables: Like standard tables, these tables now gain new nested columns and fields.

To get started, set INFER_NESTED_TYPES to TRUE in INFER_SCHEMA and in your COPY INTO statements or pipes. INFER_NESTED_TYPES is FALSE by default, so existing loads aren’t affected.

For more information, see Detect and evolve structured columns.