Databricks acquires Electric to give every AI agent its own Postgres database
DatabricksAcquirer
PPL Electric UtilitiesTarget
Databricks announced on Aug. 11, 2026 that it has acquired Electric, the startup behind the WASM‑based Postgres engine PGlite, adding technology that has grown to 13 million weekly downloads; the financial terms were not disclosed.
Databricks has acquired Electric, the startup behind the WASM‑based Postgres project PGlite, adding technology that has seen weekly downloads jump to 13 million. The move expands Databricks’ Lakebase service, which already incorporates Neon, the server‑less Postgres platform Databricks bought for roughly $1 billion last year.
Deal Terms
The companies said the transaction’s financial details were not disclosed. Electric’s team will join the Neon group within Databricks, reinforcing the Lakebase stack that already includes Neon and the recently added transactional‑processing startup Mooncake. Databricks has been building a portfolio of database assets to support its AI‑first strategy.
What Electric Brings
Electric’s flagship offering is PGlite, a full Postgres implementation compiled to WebAssembly that runs in browsers, Node.js, or sandboxed AI agents. PGlite’s weekly downloads have risen from 1 million to 13 million in the past year, and it supports dynamic extensions such as pgvector. The core of Databricks’ interest is the Electric sync engine, which mirrors a central Postgres instance to browsers, mobile apps, or agents in near‑real‑time, a “multiplayer” model similar to Figma or Google Docs but for databases. The founders confirmed that all open‑source components—including the sync engine, PGlite, Durable Streams, and TanStack DB—remain open source, while the hosted Electric Cloud service will be wound down, requiring existing users to self‑host or migrate.
Strategic Fit for Agentic Workloads
Databricks argues that agentic applications create roughly four times more databases than human developers on Lakebase, with projects averaging ten database branches and some exceeding 500 branch iterations. For many AI‑driven tasks, the compute associated with a database lives for under ten seconds before being discarded. The combination of Neon’s branch‑on‑write architecture and Electric’s sync engine enables Databricks to spin up disposable, low‑cost Postgres sandboxes that agents can use, test, and then sync back to a persistent store without incurring a fixed compute floor.
Outlook
By reuniting the two halves of a concept that began as a Neon experiment—WASM‑compiled Postgres and real‑time sync—Databricks positions Lakebase as the de‑facto data layer for the emerging AI‑agent economy. The acquisition also signals that Databricks will continue to consolidate niche database technologies to tighten its AI‑centric cloud offering.
Why It Matters
For Databricks, integrating Electric’s sync engine directly into Lakebase gives the company a unique capability to serve the exploding volume of short‑lived, agent‑generated databases, a niche where competitors such as Supabase or Aiven lack native WASM‑based sandbox support. The move strengthens Databricks’ value proposition to enterprise AI teams that need instant, zero‑cost database instances for iterative coding agents, potentially boosting Lakebase’s ARR and net‑revenue retention as customers adopt more agentic workloads.
Neon gains a deeper technical stack and a clear path to differentiate its serverless Postgres offering through built‑in real‑time synchronization, while Electric’s former hosted customers must transition to self‑hosting or other providers, opening a migration opportunity for Databricks‑managed services. Competitors that rely on traditional, always‑on database instances may see pressure to add similar lightweight, sync‑first capabilities or risk losing AI‑focused developers to Databricks’ expanded ecosystem.
Key Points
- Databricks acquired Electric; deal value was not disclosed
- Electric’s PGlite grew from 1 million to 13 million weekly downloads in the past year
- The acquisition adds Electric’s sync engine to Databricks’ Lakebase service, which already includes Neon
- Electric’s hosted Cloud service will be wound down; all open‑source components remain available
- Agents on Lakebase now create roughly four times more databases than human developers, with many databases living under ten seconds
Analysis
Databricks’ purchase of Electric, though undisclosed financially, follows its $1 billion Neon acquisition and signals a willingness to pay premium valuations for niche database capabilities that enable AI‑agent workloads. By embedding a WASM‑based Postgres engine and a near‑real‑time sync layer, Databricks can offer developers disposable databases that cost virtually nothing to spin up, a model that could dramatically increase usage‑based revenue and improve net‑revenue retention as agents generate high‑frequency, low‑duration transactions. The deal reflects a broader market trend where cloud platforms are consolidating specialized data‑infrastructure assets—evidenced by recent buys like Mooncake—to capture the AI‑driven demand for ultra‑low‑latency, serverless data stores. For investors, the transaction underscores the premium placed on technologies that lower the compute floor for AI applications, suggesting that future SaaS valuations will increasingly factor in the ability to monetize micro‑service‑scale database instances. Operators should evaluate how integrating such sync‑first databases can accelerate product iteration cycles and expand addressable markets in the emerging agentic software segment.