Databricks has announced the general availability of Lakebase, a serverless PostgreSQL-based OLTP database tailored specifically for AI workloads. This innovative database solution adopts a lakehouse architecture that separates compute from storage, providing users with unprecedented flexibility and scalability.
Built on cutting-edge technology acquired from Neon and Mooncake Labs, Lakebase supports up to 8TB per instance and runs on Postgres 17, featuring enhancements such as pgvector, instant data branching, and point-in-time recovery. With unified access controls, users can seamlessly query transactional data using Spark and Databricks SQL, eliminating the need for traditional ETL processes.
Designed with modern use cases in mind, Lakebase facilitates real-time machine learning feature serving, AI agent memory management, and embedded analytics. It integrates seamlessly with the broader Databricks Data Intelligence Platform, offering robust data processing capabilities critical for contemporary AI applications.
Lakebase is available in both Autoscaling and Provisioned tiers, currently in general availability on AWS, with Azure in public preview and plans for GCP support expected in 2026. This launch underscores Databricks’ commitment to bridging operational database capabilities with their renowned data lakehouse architecture, positioning Lakebase as a game-changer for AI-driven data workloads.

Collections

Databricks has launched Lakebase, a serverless PostgreSQL-based OLTP database built for AI workloads, now generally available on AWS. Built on technology from Neon and Mooncake Labs, it runs on Postgres 17 with support for pgvector, instant data branching, and point-in-time recovery. The lakehouse architecture separates compute from storage, supports up to 8TB per instance, and integrates with Spark and Databricks SQL to eliminate ETL overhead. Key use cases include real-time ML feature serving, AI agent memory, and embedded analytics. Azure is in public preview, with GCP support planned for 2026.

Databricks Launches Lakebase: A PostgreSQL Database for AI Workloads