Google Cloud Unifies Data for Agents

Google Cloud is betting that agent infrastructure will be won at the data layer, where governance and context matter more than another model benchmark.

• –Knowledge Catalog supplies business definitions, lineage, and permissions-aware context to improve grounding and reduce token waste.
• –Borderless Lakehouse queries data across AWS, Azure, Databricks, Snowflake, and SaaS systems without forcing physical migration.
• –Smart Storage enriches PDFs, images, scans, and audio where they already live, turning “dark” unstructured data into agent-ready assets.
• –Open Iceberg interoperability reduces lock-in and lets BigQuery, Spark, and other engines share distributed data.
• –Zero-copy does not mean zero cost: cross-cloud connectivity, caching, previews, and governance still introduce operational and pricing complexity.