Both Snowflake and Databricks converged on open data architectures in the past year, with Snowflake shipping Iceberg v3 to general availability at its June Summit and rebuilding its Horizon Catalog on Apache Polaris, while Databricks pushed Managed, Foreign, and v3 Iceberg through Unity Catalog. Agentic analytics needs four things: governed context, reusable semantics, fast query access, and portability. A single closed platform can provide the first three but not portability, which decides who wins because models change every few months. Analysts summarized that Snowflake is preparing for a world where it no longer needs to own the data format to own the customer relationship, as Iceberg v3 support is table stakes. Snowflake describes a future where metadata, lineage, identity, and policy travel with the agent rather than staying locked inside the platform. Databricks markets Unity Catalog on “write once, read anywhere” and bidirectional federation across Snowflake, Glue, and other catalogs. The value is no longer the storage but governed and portable access to data in open formats.