Supabase Acquires Turso: $150M Round Bets on Agent Databases

Supabase has raised $150 million and agreed to buy Turso, the company behind the SQLite fork libSQL. The two announcements came together and point in one direction: Supabase is building database infrastructure for AI agents that create databases far faster than human developers ever did.

What was announced

Singapore's sovereign wealth fund GIC led the round. Alphabet's growth fund CapitalG, along with IronArc and SquarePeg, also took part.31 Supabase began as an open-source Firebase alternative built on PostgreSQL.2 It said it agreed to acquire Turso for an undisclosed sum.3

Turso's founder, Glauber Costa, will become Supabase's Head of Agentic Services.1 Both companies told customers that existing users will see no changes. Supabase says it will keep building around Postgres, and Turso will keep working on SQLite.1

Accounts of the timeline differ slightly. One report says the deal was announced on October 2, 2026.1 Another says the funding round closed in late September and that the company's blog post went up on October 3.2 The gap probably reflects time zones or the difference between when the round closed and when it was announced. The core facts are the same everywhere.

One outlet goes further on financing. It calls the round a Series E extension that pushes Supabase's total funding past $500 million and reportedly values the company at $5 billion to $5.5 billion.2 That outlet frames the valuation as a modest rise from 2025, driven partly by AI demand and partly by competition from Neon, PlanetScale, and databases offered directly by cloud providers.2 Those valuation figures come from people familiar with the deal, not from an official disclosure, so they should be treated as unconfirmed.

The agent thesis

CEO and co-founder Paul Copplestone explained the deal in terms of AI agents. In his blog post, he said agents are spinning up millions of databases for the prototypes, dashboards, and apps they build. He argued that a pattern at that scale requires database infrastructure to evolve.3

Turso fits that argument. Its product consists of embedded, on-demand databases compatible with SQLite and designed for edge and serverless workloads.2 Postgres is powerful, but it is relatively heavy to provision. A lightweight SQLite-style database suits a world where agents create and discard thousands of short-lived data stores.

Supabase already exposes an MCP server so external AI agents can read and edit records.3 It says the new money will fund more agent features and provide liquidity for employees.3 Costa's title of "agentic services" suggests Turso's team will focus on that use case rather than on general-purpose SQLite hosting.

What wasn't said

Commentary aimed at independent developers points out what the announcements left out: no purchase price and no closing date.1 The same commentary accepts that nothing changes for users in the short term, but treats the long-term promise with more caution.1

That caution is reasonable. Promises that nothing will change are standard after an acquisition. Product roadmaps, free tiers, and pricing often shift once the acquired company is folded into a larger business strategy, and here that strategy is clearly about agents.

A consolidating market

The more important story may be market structure. One analysis notes that the default Postgres choice for a new project now sits with one of two well-funded companies: Neon, which is now part of Databricks, or Supabase. With this deal, the best-known hosted SQLite becomes a feature of one of them.1

This suggests that developer-focused serverless databases are being absorbed into larger platforms. Independent providers are joining companies with bigger strategic aims, whether that means Databricks' data and AI platform or Supabase's agent-focused backend. Competition among Neon, PlanetScale, and the cloud providers' own databases2 also helps explain why Supabase wanted a second database engine rather than relying on Postgres alone.

What it means

On the evidence so far, this deal is less about SQLite versus Postgres and more about who provides the infrastructure for machine-generated software. Supabase is betting that agents, not individual developers, will drive database demand. It is paying to own both a full relational engine and a lightweight engine that can be created in large numbers.

For developers, the practical advice is about portability. Commentary focused on independent developers recommends keeping Postgres and libSQL workloads portable so that future pricing or product changes don't leave you stuck.1 libSQL's open-source roots and Postgres's broad ecosystem make that achievable. It just takes some planning now, while the companies still say nothing will change.