A million databases a week: Supabase bets everyone becomes a builder
Key points
Key takeaways from Y Combinator Lightcone episode with Supabase co-founder and CEO Paul Copplestone (October 2026):
Explosive growth. Supabase has passed 13 million signups (roughly 3 million a year earlier), expects revenue to grow 4x this year, and manages close to 50 million databases — up from 8 million last year — while spinning up over a million new databases a week.
Agents are now the primary user. Around 70% of new platform activity comes from agents; Supabase's three chapters — dashboard, then API, now code-first — were chosen so CLI-and-agent-first builders get the modalities they want, and the company became a default database inside leading LLM coding tools.
Open source is critical. Everything ships open source because agents debug by reading source code, and Copplestone expects enterprises to demand inspectable code on security grounds — closed-source vendors will lose their trust.
The builder wave. Non-engineers with domain expertise can now build their own software; for the first time signups, activation and conversion are all rising simultaneously, and Copplestone sees no ceiling on PLG this year.
Personal databases. Copplestone predicts everyone with AI access becomes a builder, with personal data stores for finance, health and relationships; he built his own "control tower" app — email triage, spending, health and a personal CRM — on Supabase, dogfooding his agents-first thesis (and giving a Brazilian taxi driver the means to build a local ride-hailing app).
Architecting for billions. With ChatGPT-scale usage implying over a billion databases a month, Supabase is going both ways: Postgres scaling up to petabytes and, via the Turso acquisition, a SQLite-compatible "Supabase Spark" tier for prototyping — "prototype to petabytes" is the new tagline.
AI inside. Supabase runs no AI policy internally; the share of engineers writing 80% of their code with coding agents jumped from 10% to 60% in about six months, including heavy testing-tool adoption as the company shifts into an AI-coded development model.
Read more: Supabase · Watch the episode
