Databricks acquires ElectricSQL: PGlite, AI agent sandboxes, and the $6.9B revenue engine behind the deal

What Databricks acquired from ElectricSQL

On August 11, 2026, Databricks acquired ElectricSQL (often called Electric), the startup behind PGlite and a real-time sync engine . Acquisition terms were not disclosed .

ElectricSQL brought two core technologies into Databricks:

• PGlite — A full PostgreSQL database compiled to WebAssembly (WASM), under 3MB in size, capable of running directly inside a browser or an AI agent sandbox with no external server required . Over the past year, PGlite grew from 1 million to 13 million weekly downloads .
• A real-time sync engine — Keeps data synchronized between distributed AI agents and a centralized Databricks Lakebase .

The Electric team joined the Neon group inside Databricks — the serverless Postgres company Databricks acquired for roughly $1 billion, which now powers Lakebase .

Why this acquisition matters for AI agent infrastructure

The acquisition is a direct play to embed Postgres inside AI agent sandboxes . Databricks CEO Ali Ghodsi explained: "Agents need super-fast Postgres" that runs locally inside the agent environment but can sync asynchronously back to a central Postgres instance .

Key strategic elements:

• Bridge the gap — Agents currently work in sandboxes separated from Databricks' Lakebase (its managed Postgres layer). PGlite closes that gap by putting a real Postgres instance inside the sandbox, wired to Lakebase through real-time sync .
• Edge-to-lakehouse architecture — Extends Databricks' Postgres capabilities "from the lakehouse to the edge," as the company's official announcement stated .
• Every agent gets its own database — PGlite gives each AI agent a dedicated, lightweight local Postgres instance rather than contending for a shared cloud database, enabling fast agent-native data operations .

The bet is that enterprises standardizing agent infrastructure will pay for one SQL dialect end to end rather than maintain two .

Recent enterprise and financial developments

Databricks' financial trajectory has accelerated sharply in 2026:

• $5B funding at $134B valuation (Feb 2026): Databricks raised $5 billion in equity plus $2 billion in new debt capacity, with investors including Microsoft, JPMorgan, Goldman Sachs, and Qatar Investment Authority. Annualized revenue then stood at $5.4 billion, up 65% year-over-year .
• Revenue surged to $6.9B ARR (June 2026): By June, annualized revenue had jumped over 80% year-over-year to $6.9 billion, placing Databricks at a private market valuation reportedly in the $165–175 billion range .
• AI product revenue: Over $1.4 billion in AI product revenue, representing about 26% of total revenue; net retention rate above 140% .
• IPO on the horizon: The company is widely expected to file its S-1 in mid-2026, and it remains one of the most anticipated enterprise software IPOs in recent history .

Other notable developments include Databricks finalizing the acquisition of Panther, an AI-powered security operations platform, on August 3, 2026 S.

Major AI infrastructure news reported alongside the deal

The Databricks-Electric announcement landed in one of the most active weeks for AI infrastructure deals in 2026:

• Anthropic's $10B compute deal (Aug 4): Anthropic signed a $10 billion contract for computing capacity from a seven-month-old infrastructure startup, Volta Infra Holdings, managed by Nvidia S.
• Nvidia invests up to $3B in Lancium (Aug 7): Nvidia invested up to $3 billion in power infrastructure developer Lancium, behind the Stargate data center campus in Texas — an initial $2 billion for roughly a 20% stake .
• JPMorgan leads $441M debt deal for Global AI (Aug 10): JPMorgan Chase arranged a $441 million credit facility for Global AI, a two-year-old firm building AI data center capacity .
• Broadcom's $14.2B acquisition of OptiFabric Systems for co-packaged optical interconnects in large-scale AI clusters .
• Palo Alto Networks' $8.8B acquisition of AgentGuard AI for autonomous agent security .

Common theme: The AI infrastructure market is consolidating rapidly, with massive capital flowing into power and energy capacity (Nvidia/Lancium, Anthropic), data center financing (Global AI), and data-layer infrastructure for AI agents (Databricks/Electric). Power has become the acknowledged bottleneck across nearly every major AI infrastructure deal .