Agents – not humans – have picked the next $100b company

When AI software firm Lovable set the vibe coding scene on fire in 2024, Singapore-based Supabase thought its own platform was under attack.

The company, which hosts the back-end infrastructure for Lovable, suddenly saw an exponential number of new databases being created.

Supabase co-founder and CEO Paul Copplestone at the 2026 Tech in Asia Conference in Singapore / Photo credit: Tech in Asia

“We thought we were having a DDoS attack when Lovable was launching,” said Supabase co-founder and CEO Paul Copplestone. He spoke at the 2026 Tech in Asia Conference in Singapore.

Instead, it was a sign that people were beginning to build software with AI. The supporting database was no longer being chosen by a developer but by an AI agent.

Recipe for success

If an AI agent is becoming the customer, what makes it choose Supabase?

Copplestone points to two reasons, which are bets the company made before anyone knew how quickly AI coding would take off.

The first is open-source database system Postgres. Supabase built a suite of developer tools around Postgres, betting that developers would want more than just a database.

To meet those needs, the company bundled together many of the services required to build a functioning backend, including authentication, storage, real-time capabilities, and edge functions. For an AI agent, that matters because it can set up everything an app needs to run in one go.

According to Copplestone, AI coding tools choose Postgres as the host around 58% of the time, while Supabase itself is chosen roughly 22% to 24% of the time.

The second reason is community. From its earliest days, Supabase encouraged open-source communities to use, discuss, and contribute around the product. Developers posted about the platform on social media and shared how they used it, with its user base growing to nearly 12 million developers globally

That turned out to be valuable once LLMs began consuming vast amounts of information from the internet. The community had effectively created a large body of material that made it easier for AI models to understand and use Supabase.

This is what turned the platform into the “default for tools like OpenAI and Claude,” according to Copplestone.

When the company announced its latest fundraise, where it secured US$500 million in a round led by Singapore sovereign wealth fund GIC, it said that database launches had grown more than 600% in a year, with more than 60% of new databases being launched by AI tools rather than humans.

Road to US$100 billion

Going forward