Databricks Bolsters Lakebase Database ‘Vision’ With Latest Acquisition

Databrick’s purchase of startup Mooncake will accelerate the ability of the Lakebase OLTP database to provide data for AI agents and applications.

Data and AI platform giant Databricks has acquired startup Mooncake Labs in a move to enhance the agentic AI capabilities of the Databricks Lakebase database the company launched in June.

Databricks, which announced the acquisition of San Francisco-based Mooncake Labs on Wednesday, said in a blog post that the deal would accelerate the company’s vision for Lakebase.

Terms of the acquisition were not disclosed.

Mooncake Labs was co-founded in 2024 by Zhou Sun, Cheng Chen and Pranav Aurora with a plan to develop next-generation data infrastructure technology—specifically a data ingestion engine purpose-built for the Apache Iceberg data table format that is the foundation of many data lakehouse systems today.

Databricks launched Lakebase, a Postgres-based operational OLTP database, at its Data + AI Summit in June. The goal is to provide a fully managed database, built on a modern lakehouse architecture, for developing data-intensive applications and AI agents.

Lakebase is itself based on Postgres database technology that Databricks acquired in May when it bought database startup Neon.

According to the blog post, co-written by Databricks co-founder Reynold Xin and Databricks vice president (and former Neon CEO) Nikita Shamgunov, the combination of Lakebase and Mooncake “redefine the operational database for the age of AI.”

The Mooncake Labs acquisition, the blog says, enhances Lakebase and “accelerates the Lakebase vision, making Postgres data immediately usable across workloads, enabling developers and agents to run applications, analytics, and AI on the same fresh data in the Lakehouse and Agent Bricks, without ETL pipelines required.”

Agent Bricks is Databricks unified workspace for building AI agents, another new product that the company launched at the Data + AI Summit in June.

With the Mooncake technology incorporated into Lakebase, Postgres changes are mirrored in real time to a lakehouse, ensuring that analytics and AI stay current as agents create new tables, events and workflows, according to the blog.

The Mooncake team, which is joining Databricks, “brings deep knowledge of Postgres internals, distributed systems and open table formats like Iceberg,” the blog said. The experience of Mooncake’s founders “at the intersection of OLTP [online transaction processing] and AI-scale workloads strengthens Lakebase’s ability to deliver on its vision: a unified foundation for applications, analytics and AI agents,” according to the blog.