What actually happened
On 2 October 2026 Supabase said it is acquiring Turso. Paul Copplestone’s post says existing users see no change. Supabase keeps building on Postgres. Turso keeps working on SQLite. Glauber Costa and Pekka Enberg join, and Glauber leads the agentic infrastructure effort. No price and no close date are in the post.
The scale line is Supabase’s: it is already launching more than one million databases a week, because agents create prototypes, dashboards, and isolated environments. Turso’s pitch, repeated by both companies, is many small databases on one server, loaded when needed and suspended when idle, including a database per agent. Named users of that pattern include Superhuman, Sauna.ai, CTO.new, and Mastra. Those are examples, not a new Supabase SKU.
Why a small team cares
A solo builder running five coding agents on one Supabase project already has the failure mode this deal points at. Shared dev data, one agent’s migration breaking another’s preview, and a dashboard you have to clean up by hand. A database that appears with the branch and disappears with the task would remove that babysitting.
It is not available as a switch today. The post says they are building that evolution. Turso’s own note says its databases, APIs, and open source stay as they are, with a path to Postgres when a workload outgrows SQLite. That path is described, not shipped as a migration button.
Hype vs useful
Useful as a signal: infrastructure vendors now assume the caller may be an agent, and Supabase is a reasonable long-term default if you already pay for it. Not useful as a reason to re-platform, buy Turso seats, or split production into a database per agent. Production still wants one Postgres database with backups. Per-agent databases belong in previews, and only after the product exists.