For a decade, enterprise data teams argued about warehouses versus lakes. That argument is settled: the lakehouse pattern — open table formats over cheap object storage with warehouse-grade transactions — is the default for new platforms.
But winning the architecture argument does not win the outcomes. The organizations getting value are the ones treating data as a product: clear ownership, versioned contracts, SLAs, and discoverability. The table format is table stakes.
-- Open table formats give warehouse guarantees on lake storage:-- ACID, schema evolution, and time travel. Reproduce a regulator's-- question from six weeks ago, to the exact row.SELECT * FROM main.gold.positionsVERSION AS OF 142; -- state the report was filed on SELECT * FROM main.gold.positionsTIMESTAMP AS OF '2026-04-21T16:00:00Z'; -- as-of any point in timeNext articleRAG in Production: Lessons From Regulated EnvironmentsRead Article