
Supabase CEO on the "painful" decisions that built a $5B company
- 29m
- Transcript
Summary
Written by a model from the transcript; not yet reviewed.
Paul Cobblestone, CEO and co-founder of Supabase, discusses how the open source Postgres platform became the backend of choice for vibe coding tools like Bolt, Lovable, and Cursor, driving the company from a $2B to a $5B valuation in four months. He explains the deliberate strategy behind saying no to million-dollar enterprise contracts that didn't fit the product roadmap, drawing on a pattern he's seen in developer tools that lose their developer base when they pivot to enterprise. The only counterexample he cites is Stripe.
Cobblestone outlines Supabase's growth model: product-led growth for the base, with product-led sales for larger organizations. The company is fully remote across 40 countries with about 220 people, over 5 million developers on the platform, 10 million databases launched, and 100,000+ paid accounts. He describes plans to build a new storage engine and a sharded Postgres solution, both to be open-sourced, and notes that LLMs are particularly well-suited to SQL because it's an old, well-established language. He also envisions a 'self-driving database' where AI handles routine administration tasks like failover based on traffic patterns.
- Supabase deliberately rejects enterprise contracts that don't align with its product roadmap, even million-dollar ones, to preserve the developer experience. 4:22
- The company's strategy mirrors Firebase and MongoDB: capture developers with an easy start, then make the platform so good they never want to leave. 9:46
- LLMs excel at SQL because it is an old, well-established language, making databases more accessible to non-specialists who can prompt rather than code. 22:15
- Supabase is fully remote across 40 countries with about 220 people, hiring only autonomous senior engineers to make async collaboration work. 11:57
- The 'entrepreneur' role is growing as non-developers use AI tools to build applications they previously could not without hiring an engineer. 26:14
Highlights
- 3:33
Dev Tools Losing Developer Focusinsight
The guest describes a recurring pattern where developer tools gain enterprise customers, swing hard toward that audience, and lose touch with the developer segment. He cites Stripe as the only counterexample that has stayed relevant for developers for a long time.
- 4:22
Saying No to Million-Dollar Contractsstory
The guest recounts that it has happened many times that enterprises offered large contracts and the company said no because the request didn't fit the roadmap. He describes it as painful each time, from $10,000 support contracts early on to million-dollar deals now.
- 9:46
The Firebase and MongoDB Playbookstrategy
The guest explains that only two database companies have succeeded in the last two decades: Firebase and MongoDB. Neither built a bigger database to migrate workloads; both targeted developers building from day one and made them never want to leave.
- 12:44
The guest shares that Supabase has around 220 people (grew from 95 at the start of the year), over 5 million developers, 10 million databases launched, and over 100,000 paid accounts on the platform.
- 17:12
Zero to a Million in Ten Daysdata
The guest mentions a customer that scaled from zero to a million users in 10 days, built by a developer. He questions whether a 10-minute prompt from a non-developer could manage the same scale, while noting AI models are starting to write database code better than he can.
- 22:15
Why LLMs Are Good at SQLinsight
The guest explains that relational databases like Postgres are written in SQL, an old and esoteric language, and LLMs are very good at languages that have been around for a long time. This means non-database developers can now build on Supabase by simply prompting.
- 22:55
The Self-Driving Database Visioninsight
The guest describes a concept where AI diagnoses database problems and makes small corrections throughout the day, such as failing over from Singapore to Europe as the sun rises, handling the boring work data engineers used to do manually.
- 26:14
Non-Developers Becoming Buildersstory
The guest shares that friends who never wrote a line of code are now showing him applications they've built with AI tools, excited about their customer signups. Previously they would have needed to convince a $230,000 engineer to help with their project.
About this episode
Vibe coding has taken the tech industry by storm, and it's not just the Lovables and Replits of the world that are winning. The startups building the infrastructure behind them are cashing in too. Supabase, the open-source database platform that's become the backend of choice for the vibe-coding world, raised $100 million at a $5 billion valuation just months after closing $200 million at $2 billion. But co-founder and CEO Paul Copplestone has a surprising strategy: he keeps turning down million-dollar enterprise contracts, betting instead that vibe coding will birth the next generation of companies. Today on TechCrunch's Equity podcast, Julie Bort sat down with Copplestone to explore Supabase's rise and what it means for the database race. Listen to the full episode to hear about: Why Copplestone believes "the death of Oracle won't take a generation" The technical moonshots Supabase is funding to make Postgres even more scalable How he decides which enterprise deals to turn down, and why it still "feels very painful" Subscribe to Equity on Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod. Learn more about your ad choices. Visit megaphone.fm/adchoices
Transcript
5,001 words · 4 speakers
Topics
- Open Source Databases
- Vibe Coding
- Product-Led Growth
- Postgres
- AI and Development
- Startup Strategy
- Remote Work
- Developer Tools
Page updated · transcript and summary are machine-generated