Databricks Acquires Electric to Bring WASM Postgres Sync to AI Agent Sandboxes
Databricks has acquired Electric, pairing its WASM-based local Postgres, PGlite, with a real-time sync engine to give AI agents fast local data alongside its centralized Lakebase database.
Overview
Databricks has acquired Electric, the startup behind the WASM-based local Postgres database PGlite and a real-time data synchronization engine, in a move the two companies say will extend Databricks’ Postgres capabilities “from the lakehouse to the edge,” according to Databricks. The deal closed for an undisclosed sum, according to InfoWorld.
What We Know
Electric has “pioneered data primitives purpose-built for agents, including PGlite and a real-time sync engine to keep data synchronized between distributed agents and a centralized Lakebase,” Databricks said in its announcement. PGlite runs a lightweight Postgres instance directly inside an application or an AI agent’s own sandbox, while the sync engine keeps that local data consistent with Lakebase, Databricks’ centralized, at-scale Postgres database, according to Databricks. Databricks compared the sync architecture to the technology underpinning “collaborative apps like Google Docs, Figma and Notion,” according to Databricks.
The combination creates a two-tier setup: PGlite handles data locally within an application or agent environment, while Lakebase serves as the centralized store for shared and persistent data, according to InfoWorld. Databricks argues that agentic applications — where multiple agents can work independently for minutes or hours, repeatedly querying data — create latency problems for a single centralized database that local, synchronized data can avoid, according to InfoWorld.
PGlite’s usage has grown fast: Databricks says the project went “from 1M to 13M weekly downloads in just twelve months,” a figure Electric’s own blog independently confirms, calling PGlite the company’s “most widely-used project at 13M downloads a week.” The technology traces back to Neon, the Postgres company Databricks acquired in 2025 and built Lakebase on top of, according to InfoWorld — as previously reported. PGlite was built on “foundational WASM Postgres work” by Stas Kelvich, who co-founded Neon, Databricks said, and Electric “took that proof of concept and turned it into the embeddable Postgres that millions of projects run every week,” according to Databricks. Electric’s own account of that history says the company was first pointed toward the idea by Nikita Shamgunov, Neon’s co-founder and CEO, after Electric had been building Postgres-to-SQLite sync and he suggested syncing Postgres to Postgres instead; Kelvich then shared an early WASM Postgres repo that became the basis of PGlite, according to Electric.
Electric said everything it has previously open-sourced — Postgres Sync, PGlite, TanStack DB and Durable Streams — stays open source, but Electric Cloud, its hosted service, is winding down, with existing customers directed to self-host or move to another provider, according to Electric. Lakebase itself is described as serverless Postgres for apps and agents with scale-to-zero, native branching, and support for both transactional (OLTP) and analytical (OLAP) workloads, according to Electric.
Analyst Reactions
Industry analysts told InfoWorld the local-database approach could make agents faster and cheaper but raises new governance questions. “Running a database inside an agentic application or inside an agent’s sandbox can help make agents faster, especially for complex and longer-running tasks, as local access cuts down on network hops, in turn reducing wait times,” said Pareekh Jain, principal analyst at Pareekh Consulting, adding that reliability could also improve “when connectivity is poor.” Chandrika Dutt, research director at Avasant, said reduced network calls could lower infrastructure costs for CIOs, though the magnitude would depend on workload and state size. Amit Kumar Jena, AI development manager at Kanerika, said the approach could also cut costs by reducing the need to provision a fully managed database instance for every agent.
Manoj Chandra Jha, principal analyst at Nord-IQ Research, cautioned that “cost and reliability benefits remain unproven,” since Databricks has not yet deployed the architecture at production scale, and said distributing state across agent sandboxes “expands the attack surface and fragments governance, forcing enterprises to extend access control, audit, and compliance frameworks beyond a single database perimeter into hundreds of ephemeral local instances.” Dutt echoed the governance concern, saying CIOs will need to weigh “what enterprise data can be materialized in an agent environment, how that data is secured and retained, how local state is audited and deleted, and how synchronization and conflicts are managed.” Jena added that while “central warehouse governance is a solved problem,” sandbox-level state isn’t, and said every CIO evaluating agent platforms should ask vendors “how does state get secured and torn down inside an agent, not just inside the central database.”
On competitive positioning, Dutt said the dual-database architecture could give Databricks a temporary edge, noting that rivals including Snowflake do not currently offer comparable WASM-Postgres capability; Jena said the same is true of Google Cloud and Teradata. Dutt cautioned, however, that a differentiated architecture “does not necessarily translate into a lasting competitive advantage,” and that its significance will depend on whether Databricks can deliver the security, governance, observability and consistency controls enterprises need to adopt it.
What We Don’t Know
Neither company disclosed the acquisition price. Databricks has not said when the merged PGlite/Lakebase architecture will be broadly available in production, and analysts noted the cost and reliability claims remain unproven at scale.
Analysis
The acquisition folds Electric into Neon, the Postgres company Databricks bought in 2025 to build Lakebase, reuniting a lineage that began when Neon co-founder Stas Kelvich’s early WASM Postgres experiment became the seed for Electric’s PGlite, according to Databricks and Electric. For Databricks, it extends a strategy of assembling a full-stack Postgres offering — from Neon’s managed cloud database to Electric’s edge-side sync — squarely aimed at the emerging pattern of AI agents that need fast local state without losing a consistent, centrally governed record.