Supabase acquires Turso for SQLite

This deal pushes Supabase downmarket into the cheapest possible database unit, a database per experiment, agent, or preview app instead of a shared Postgres instance from day one. Turso’s SQLite system is built so many tiny databases can wake up only when code touches them, then go idle again, which fits the bursty pattern of AI generated software and agent workloads far better than reserving heavier database capacity upfront.1 2 3 4

• Supabase is not replacing Postgres with SQLite. Both companies said they will keep building their existing platforms, with Turso serving lightweight workloads and Supabase framing Postgres as the destination when an app needs a fuller production database.1 2 3
• The practical gain is server density and cost. Turso’s many database architecture lets one server handle large numbers of small, mostly idle databases, which is useful when coding agents or app generators create huge volumes of short lived projects.2 3
• This mirrors a broader market move toward agent friendly data infrastructure. Databricks bought Electric in August 2026 to bring PGlite and sync technology into Lakebase, while Supabase chose SQLite economics for the same core problem, cheap isolated state for lots of small workloads.3 5 6

The next step is a more explicit ladder inside Supabase, starting with dormant by default SQLite databases for prototypes and agents, then graduating successful apps into Postgres, compute, analytics, and storage as usage becomes steady. That makes Supabase harder to displace because it can serve both the first draft of an app and its production version.1 2 3 4