Supabase Raises $150M, Buys Turso as AI Agents Drive Databases

A raise, an acquisition, and a new kind of customer

Supabase, the company built around open-source PostgreSQL, has raised $150 million and agreed to acquire fellow database startup Turso, announcing both moves together. 13 Singapore's sovereign wealth fund GIC led the round. Alphabet's independent growth fund CapitalG, IronArc and SquarePeg also took part. 13 Supabase did not disclose a valuation for the round, and the price paid for Turso was also kept private. 13

The timing is notable. The round comes just four months after Supabase closed a $500 million Series F. 12 The company has been raising at a rapid pace. It reached unicorn status with a $200 million Series D in April 2025, added a $100 million Series E that October, and then raised the Series F at a $10.5 billion valuation in June. 1

The main story is who is now creating databases on the platform. According to WOWTALE, AI agents now account for 70% of new databases created on Supabase. 1 Ventureburn uses a different figure: a 600% increase in database creation by autonomous agents over the past year. 2 These numbers measure different things, one a share of new databases and the other a growth rate. Both point the same way. Software agents, not human developers, are becoming the main source of new databases on the platform.

Why Turso fits

Supabase packages the backend pieces an application typically needs into one service: databases, authentication, storage and edge functions. 1 That model was designed for human developers building apps. Agent-driven use looks different. Agents create large numbers of small, often short-lived databases for prototypes, experiments, dashboards and quick apps.

CEO and co-founder Paul Copplestone described the shift directly in a blog post. He wrote that agents are "spinning up millions of databases" for the things they build, and that this pattern, at this scale, "requires an evolution in database infrastructure." 3

Turso is meant to fill that gap. The startup rewrote SQLite in Rust and built a cloud platform designed so that a single server can manage very large numbers of databases. 1 Ventureburn describes the deal as giving Supabase a low-cost architecture for launching micro-databases almost instantly. 2 Postgres is a full-featured database, but it is costly to provision for every throwaway task. A lightweight, embedded engine like SQLite fits better when the workload is millions of tiny databases. Supabase appears to be pursuing a two-tier approach:

• Postgres for durable production applications
• Turso's SQLite-based engine for the agent-generated long tail

SiliconANGLE also notes that Supabase already offers an MCP (Model Context Protocol) server, which lets external AI agents read and edit records. 3 Taken together, the Turso deal looks less like a side bet and more like the next step in a strategy to make Supabase a default backend for agents.

Where the money goes

The sources agree on two uses for the new capital. Part of it will provide liquidity for employees who hold shares. The rest will go toward building new features for AI agents. 13

The employee liquidity piece is worth noting. Secondary sales like this are common at late-stage, highly valued private companies that are in no hurry to go public. It suggests some of this round is about rewarding staff after a steep run-up in valuation, not only about funding operations. Supabase took on a large sum only a few months earlier, so the round seems driven by investor demand and strategic positioning more than an urgent need for cash.

Reading the hype

The three outlets differ in tone. SiliconANGLE presents the news in measured business terms. 3 Ventureburn's coverage is much more enthusiastic. It calls traditional provisioning for agent workloads "financially ruinous" and says Supabase is building the core engine for an "AI-first internet." 2 The basic facts are consistent across all three. The larger claims about rewiring backend architecture are interpretation.

One caution applies to the headline statistics. A database created by an agent is not the same as a database that becomes a lasting, paying workload. Many agent-created databases are likely prototypes or one-off experiments. That is exactly why Supabase needs cheaper infrastructure, and it is also a reason to be careful about turning database counts into revenue projections.

The takeaway

The clearest signal here is structural. A leading developer database platform now reports that most of its new databases come from software agents, not people. 1 It is spending acquisition money to rebuild its stack around that fact. 3 If the trend holds, the database market's competitive question may shift. The old question was which platform developers prefer. The new one is which platform agents can use most cheaply and quickly. Supabase is betting it can answer the second question while keeping its lead on the first.