How Supabase Became One Of The Fastest Growing DevTool Companies In The World

Paul argues that money alone cannot fix fundamental issues like product-market fit, citing that most YC companies close with plenty of cash remaining. He emphasizes that startups should not exist if incumbents like AWS are already good enough, and that persistence and finding the right market are more critical than capital.

Paul shares his background as a New Zealander with an entrepreneur father, having worked one corporate job he hated before moving to startups. Supabase is his third venture-backed startup, but he has done many others in between. His first startup failed due to lack of product-market fit, and he left because of strategic differences with the CEO.

Paul agreed to be CTO for a co-founder's business for two years, during which he incubated Supabase as a side project. He knew he wanted to build a dev tools and database company but wasn't sure of the exact form. The company he helped launch is still profitable, but Supabase became his life's work.

While using both PostgreSQL and Firebase in a previous company, Paul hit scaling limits on Firebase and decided to migrate to PostgreSQL. He wrote an open-source tool for this migration, posted it on Hacker News, and it gained traction within 3-4 months. This momentum led him to launch Supabase, combining his desire for an open-source database company with the growing interest in PostgreSQL.

By late December 2024, Supabase noticed a surge in database launches from AI coding agents like Bolt and Lovable. Initially, the team debated whether to support this use case, as early apps were small and might not grow. However, by the time of the interview, 60% of database launches were measured as agent-initiated, with the actual figure likely around 90%, totaling millions per month. This shift forced Supabase to create a dedicated 'Supabase for Platforms' product.

The success of Lovable and Bolt led Supabase to iterate on a product that allows companies to launch millions of databases. They realized this same capability—launching many databases, monitoring security, and shutting down overgrown instances—was exactly what enterprise innovation labs wanted. By merging the roadmaps for AI agents and large enterprises, Supabase created a unified platform that serves both millions of small agent-launched databases and fewer but larger enterprise databases.

After Christmas 2024, Supabase saw all metrics accelerate simultaneously: user count rose from 6.5 million to 10 million, while conversion rates and activation rates also increased. This was unusual because typically at that scale, marketing spend would dilute user intent and lower conversion. The growth was driven entirely by AI coding agents building more and larger applications, leading to better product-market fit without any marketing expenditure.

Supabase deliberately avoids spending money on marketing, preferring to give free databases to developers instead. The CEO states, 'If I could give a free database, I'd rather give that to developers than spending on marketing.' This approach has paid off because AI agents naturally drive high-intent users, so conversion and activation rates remain high even as the user base grows rapidly.

Paul Copplestone explains that while many AI agents currently focus on the build stage—prompting to create an application—the harder problem is operating the application once built. For Supabase, this means developing self-driving databases that handle operations like patching, security, and uptime automatically, so developers don't have to wake up at night worrying about database failures. He emphasizes that the operate stage is a much harder area to launch in compared to the easily replicable build stage, and advises focusing on such hard problems.