Last Updated on July 7, 2026 by Taya Ziv
In 2014, Paul Copplestone walked into an investor meeting in Christchurch, New Zealand, and pitched a database idea. Nobody bit. Not a single check. Not a follow-up call. The investor didn’t think databases were interesting enough to fund.
Twelve years later, Copplestone’s database company just raised $500 million at a $10.5 billion valuation. And the craziest part? His biggest customer in 2026 isn’t a person. It’s Claude Code.
This is not a funding story. This is a map of where the actual money in AI is going, and it’s not where most founders think.
The Numbers That Break the Narrative
Supabase, the open-source Postgres platform Copplestone co-founded with Ant Wilson in 2020, just closed a Series F led by GIC, Singapore’s sovereign wealth fund. Stripe doubled down. Salesforce Ventures joined. The valuation jumped from $5 billion in October 2025 to $10.5 billion today. That’s a 110% increase in eight months.
But the valuation isn’t the story. This is: AI agents now deploy the majority of databases on the Supabase platform. Not some. Not a growing share. The majority. Claude Code, Anthropic’s terminal-based coding agent, is the single largest contributor since January. Database launches grew 600% year-over-year. More than 60% of all new databases on the platform were created by AI tools.
The developer community doubled from 4 million to over 9 million in eight months. And Supabase’s Platforms product, which powers most of the top AI app builders, saw 370% customer growth in six months.
Read those numbers again. The platform didn’t grow because humans loved it. It grew because machines loved it.
The Accidental AI Infrastructure Play
Copplestone didn’t set out to build AI infrastructure. He built a developer tool. An open-source alternative to Google’s Firebase, with the tagline “build in a weekend, scale to millions.” Postgres database, authentication, storage, edge functions, all bundled into something a developer could start using in minutes.
Then vibe coding happened.
When Lovable, Bolt, Cursor, and every other AI coding tool needed a backend to spin up databases for the apps their users were building, they defaulted to Supabase. Not because Supabase marketed to them. Because the developer experience was already so clean that AI agents could navigate it without human help.
This is the same pattern we saw when Notion quietly became the biggest AI agent platform on Earth with 1 million agents in three months. Notion didn’t build for agents. It built a workspace so intuitive that agents adopted it by default. Supabase did the same thing, just one layer deeper in the stack.
The lesson is uncomfortable for anyone building an “AI-native” product: the companies winning the AI era didn’t build for AI. They built great developer experience, and AI agents are the most demanding developers on the planet.
The Vibe Coding Supply Chain
Here’s how the money actually flows.
A non-technical founder opens Lovable or Bolt. They describe what they want in plain English. The AI generates a full-stack application. That application needs a database. The AI spins up Supabase because it’s the default. Every new app is a new Supabase customer. Automatically. No sales call. No demo. No onboarding email.
When Cursor reached a $50 billion valuation and declared the SaaS playbook dead, the takeaway for most founders was “AI coding tools are the future.” The smarter takeaway was “who supplies the infrastructure that AI coding tools rely on?”
Cursor is the pickaxe. Supabase is the general store.
Databricks acquired Neon, Supabase’s closest serverless Postgres competitor, for roughly $1 billion in May 2025. That removed the most credible independent alternative. Firebase, owned by Google, is built on NoSQL, an architecture AI agents struggle with because relational data models are what they prefer to work with. Amazon’s Aurora Postgres and MongoDB compete for enterprise workloads, but neither has the open-source, developer-first positioning that makes Supabase the default for the AI app building community.
So the competitive landscape is clearing at exactly the moment when demand is exploding. That’s how you get a 110% valuation increase in eight months.
What Founders Should Actually Learn From This
Everyone’s building AI products. Agents, copilots, chatbots, assistants, wrappers. The pitch decks all say “AI-native.” And most of them will fail because they’re competing on intelligence, which is commoditizing, instead of competing on defaults, which compound.
Supabase’s moat isn’t technology. Postgres existed before Copplestone was born. The moat is that when an AI agent needs a database, the path of least resistance is Supabase. And in a world where agents are creating millions of applications per month, being the default is worth more than being the smartest.
The same logic applies across the stack. Authentication, payments, hosting, monitoring, analytics. Every layer of infrastructure that AI agents rely on has a default winner forming right now. Most of those winners aren’t AI companies.
Anthropic’s revenue grew 80x in one year, and the product that drove it was Claude Code, not the chatbot. But Claude Code doesn’t exist in isolation. It’s spending its customers’ money at Supabase, at Vercel, at Stripe, at every infrastructure provider in the vibe coding supply chain. The AI company captures the intent. The infrastructure company captures the transaction.
If I were starting a company today, I wouldn’t build another AI agent. I’d look at what AI agents buy when they build things, and I’d make sure I was the easiest option on the shelf.
The Multigres Bet
Alongside the funding, Supabase launched Multigres, an open-source horizontal scaling layer for Postgres. Today, companies that outgrow a single database instance have to migrate to a completely different system. Multigres fixes that by bringing sharding and high availability to Postgres natively.
Copplestone told CNBC the goal is to scale “up to the size of OpenAI or even larger.”
That’s not engineering ego. It’s positioning. If Supabase is already the default for the vibe coding wave, and vibe coding is producing millions of new applications, some of those applications will become real companies. And when they do, Supabase wants to make sure they never need to leave.
The database market is projected to reach $285 billion by 2030 according to Grand View Research. Supabase’s 600% year-over-year growth on its own platform outpaces that. The question the industry has to answer now is straightforward: if AI agents are the primary creators of new software, and those agents default to Supabase for their backend, what does the database market look like in five years?
Paul Copplestone couldn’t find a single investor for a database in New Zealand in 2014. In 2026, AI agents are his biggest customer, and they never even needed a pitch.
