Databricks launches Lakebase, a Postgres layer for AI-ready lakehouse analytics
Databricks rolled out Lakebase, a new Postgres-compatible database layer built into its lakehouse platform for running AI apps and agents on operational data. It also introduced Agent Bricks, a tool to automate building AI agents.
The announcements came at Databricks’ annual Data + AI Summit and investor session in San Francisco. 
Ali Ghodsi, Databricks CEO, said:
“We’ve spent the past few years helping enterprises build AI apps and agents that can reason on their proprietary data with the Databricks Data Intelligence Platform.
Now, with Lakebase, we’re creating a new category in the database market: a modern Postgres database, deeply integrated with the lakehouse and today’s development stacks.
As AI agents reshape how businesses operate, Fortune 500 companies are ready to replace outdated systems. With Lakebase, we’re giving them a database built for the demands of the AI era.”
Powered by Neon technology, Lakebase separates compute and storage. It’s fully managed, supports 300+ data sources, and handles operational data in low-cost data lakes with autoscaling compute for AI workloads. Databricks calls traditional operational databases outdated, costly, and vendor-locked.
William Blair analyst Jason Ader summarized Databricks’ roadmap in three chapters:
- Chapter 1: Establish the lakehouse as the base architecture
- Chapter 2: Embed AI to broaden data insights
- Chapter 3: Launch Lakebase as a transactional layer for operational use cases inside the lakehouse
Ader sees this as unifying analytics, AI, and transactional workloads into one platform—timely as companies want simpler, AI-ready infrastructure.
Databricks also launched Agent Bricks to let business users build customized AI agents automatically generating synthetic training data. This cuts trial-and-error and need for specialist tools.
Meanwhile, Managed Iceberg Tables entered Public Preview. They support Apache Iceberg’s REST Catalog API and Unity Catalog governance, enabling fast, cost-efficient queries with external engines like Spark, Flink, and Kafka.
Informatica, soon acquired by Snowflake, is a launch partner for both Lakebase and Managed Iceberg Tables, letting its customers handle Iceberg data at scale on Databricks.
Informatica also announced new AI connectors and GenAI Recipes to speed enterprise AI agent adoption via Mosaic AI, Databricks’ AI suite.
Graph engine supplier PuppyGraph announced native integration with Managed Iceberg Tables. This enables live, zero-ETL graph queries on petabyte-scale datasets for fraud detection, root cause analysis, and network path tracing—all governed by Unity Catalog.
Databricks confirmed two major cloud partnerships. Its Azure collaboration got a multi-year extension for tighter ties with Azure AI Foundry and Power Platform. Google Cloud’s Gemini 2.5 models (Pro + Flash) are now natively available in Databricks, running on customer data with built-in governance and billing via Databricks contracts.
Databricks also committed $100 million to AI education, democratizing access through free platform editions and training for students and professionals.
Additional launches include Lakeflow Designer, a no-code ETL builder with GenAI help, entering Private Preview soon. Databricks One, a code-free AI/BI dashboard tool for business users, will hit public beta later this summer.
Singapore’s national health agency Synapxe adopted Databricks for its HEALIX cloud analytics platform and signed an MoU to boost healthcare AI talent in the country.
Databricks expects to hit $3.7 billion in annualized revenue by July, with 50% year-over-year growth. Fiscal 2025 ended at $2.6 billion revenue and near cash flow breakeven. The company plans to hire 3,000 staff in 2025 on top of an 8,000+ workforce. IPO remains unannounced.
Lakebase is available in Public Preview now with more upgrades coming. Learn more on the official Lakebase blog.
