Databricks Lakebase Reaches General Availability on Azure as Serverless PostgreSQL Expands to 14 Regions
Databricks has made Lakebase, its serverless PostgreSQL database built on acquired Neon technology, generally available on Azure with autoscaling compute, high availability, and compliance certifications across 14 regions.
Databricks has announced the general availability of Lakebase on Azure, marking the second major cloud platform to receive production-ready support for the company’s serverless PostgreSQL database. The March 2, 2026 release extends Lakebase to 14 Azure regions worldwide, up from just three regions during the public preview period.
Lakebase is a serverless, PostgreSQL-based online transaction processing (OLTP) database that separates compute from storage, allowing each to scale independently. The product emerged from Databricks’ approximately $1 billion acquisition of Neon in May 2025, followed by the acquisition of Mooncake Labs in October 2025 to strengthen PostgreSQL-lakehouse integration.
Azure GA Brings Autoscaling and High Availability
The Azure GA release introduces Lakebase Autoscaling as the default for all new instances, whether created through the UI, API, Terraform, or SDKs. Autoscaling projects include scale-to-zero capability, database branching, and instant restore. Existing provisioned instances retain their current configurations and connection strings.
Alongside the GA announcement, Databricks shipped high availability for Lakebase Autoscaling, providing automatic failover across availability zones with no connection string changes required. The compliance security profile is now supported for workspaces using HIPAA, C5, TISAX, or no compliance standard, a requirement for regulated industries.
The 11 new Azure regions added at GA include Australia East, Brazil South, Canada Central, Central India, South Central US, Southeast Asia, East US, Central US, North Europe, UK South, and West US 2, joining the existing East US 2, West Europe, and West US regions.
March Updates Add Cost Controls and Data Sync
Several updates followed the initial GA announcement throughout March. On March 12, Databricks added OAuth role management through the Lakebase UI and REST API, along with serverless budget policies and custom tags for cost attribution. The same update also introduced Lakehouse Sync in beta, which continuously replicates Lakebase Postgres tables to Unity Catalog Delta tables using change data capture.
On March 16, Databricks enabled Lakebase Autoscaling databases as resources within Databricks Apps, allowing developers to incorporate operational databases directly into their application workflows.
Building the Operational Database Layer
Lakebase represents Databricks’ push to add an OLTP layer atop its analytics-focused lakehouse platform. The product supports PostgreSQL 17 with the pgvector extension for vector search and PostGIS for geospatial queries. Databricks has described the architecture as running ephemeral compute over durable data lake storage, capable of processing over 10,000 queries per second with a 35-day data recovery window.
Integration with Unity Catalog provides unified governance across operational and analytical data. CTO Matei Zaharia has said the approach enables branching, snapshots, and point-in-time rollback for both human operators and AI agents, positioning Lakebase as infrastructure for agent-driven applications that need persistent state.
Lakebase first reached general availability on AWS in early February 2026. Google Cloud support is expected later this year. Inbound Private Link for Lakebase Autoscaling entered public preview on March 9, addressing enterprise requirements for network-isolated database access.