Supabase is buying Turso and will keep running SQLite beside Postgres for agent databases: Reveneau

Why it mattersA team building an AI agent that wants a separate database per user, per session or per run now has a clear path from the Postgres they already use on Supabase to a per-agent SQLite file on the same account, without picking a smaller vendor and hoping it survives.

For a team building an agent that spins up a new database for every user or every run, the market just moved from two separate vendors to one.

Supabase said on 2 October 2026 that it is acquiring Turso, the SQLite-in-Rust company founded by Glauber Costa and Pekka Enberg. The announcement sits on Supabase's own blog, with a matching post from Costa on Turso's blog titled "Turso is joining Supabase to give every agent its own database". No price was given.

What each company brings

Supabase packages Postgres with authentication, storage and edge functions, and says in the announcement that it is now launching more than one million databases per week. Turso rewrote SQLite in Rust and built a cloud platform where, in Supabase's words, "a single server can manage millions of databases, loading them when needed and suspending them when they're not". Named Turso customers in the post include Superhuman, Sauna.ai, CTO.new and Mastra.

The two products do different jobs. Postgres is the long-running, shared database that a web app or an internal tool sits on top of. SQLite on Turso's platform is a per-user or per-session file that a Supabase account can create as cheaply as writing to disk.

What changes for current customers

The post is explicit on this: "For existing users, nothing changes. Supabase will continue building around Postgres, while Turso will continue its work on SQLite." Both brands keep shipping, and Supabase says Turso's team continues to run Turso. The merger of the two products into a single ecosystem is the direction, not a date.

Glauber Costa will lead what Supabase calls its "agentic infrastructure" effort. Pekka Enberg joins with the engineering team.

Why this is being framed as an agent story

The announcement puts the acquisition inside one specific shift. Supabase's argument is that agents are "spinning up millions of databases to power the prototypes, explorations, dashboards, and apps they're building", and that this breaks the assumption that each application has one long-lived database with a human team around it. In their words, an agent "should be able to create a database as easily as creating a file, with just as little concern about cost". Turso's architecture, where many small databases live on one server and sleep when idle, was designed for that pattern.

That framing is Supabase's own, and it matches a trend other platforms have been naming: Vercel, Cloudflare and others have been shipping per-session storage primitives for the same reason. Whether this is the right architecture for a given agent is a judgement call, not a settled question.

A team currently running Postgres on Supabase for its product can now reach for a SQLite file per agent run on the same account, without onboarding a second vendor and without the smaller company getting bought by somebody else next year. A team already on Turso has a bigger platform behind the roadmap. A team on libSQL who has not committed to a vendor at all has one fewer independent SQLite host to choose from.

Source

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.