Google renamed its lakehouse layer on 20 April 2026: BigLake is now called Lakehouse, and the BigLake metastore is now the Lakehouse runtime catalog. The catalog is serverless, implements the Iceberg 129 REST Catalog API with credential vending ("vending short-lived access tokens directly to client engines"), and serves Spark 129 , Flink 129 , Hive 129 , Trino 403,499 , BigQuery 1 and AlloyDB from one set of tables in Cloud Storage. Governance across projects sits in Knowledge Catalog (formerly Dataplex Universal Catalog).
Inside BigQuery, Apache Iceberg managed tables (formerly BigLake tables for Apache Iceberg in BigQuery) store Parquet 129 data and Iceberg metadata in your own Cloud Storage bucket, while BigQuery runs "automatic storage optimization, including adaptive file sizing, automatic clustering, garbage collection, and metadata optimization". They accept high-throughput streaming through the Storage Write API, and Spark and other engines read them after BigQuery exports Iceberg metadata.
Billing has three parts: Cloud Storage bills the data at its usual rates, background maintenance consumes Data Compute Units per second, and queries use BigQuery's on-demand or slot pricing (BigQuery). For a team already on BigQuery, managed Iceberg tables are the cheapest way to make the same data readable by Spark; for a multi-engine lakehouse, the Lakehouse runtime catalog is the piece to evaluate.