Supabase raised $500 million in June at a $10.5 billion valuation. Four months later it raised $150 million more -- and wouldn't name a price

Supabase announced two things at once on Oct. 2, 2026: a new $150 million funding round, and the acquisition of Turso, a startup that rebuilt the decades-old SQLite database for AI agents. Both moves answer the same question -- where does an agent's own database actually live, if it needs one just for itself rather than a shared one? The timing is the part worth sitting with: this is Supabase's second raise in four months, after a $500 million round in June priced the company at $10.5 billion -- and this time, it named no number at all.

$150M (Raised four months after a $10.5B valuation, with no new price attached) Turso's engineering pitch is the inverse of a normal database: instead of one long-lived instance serving an application, an agent can spin up a disposable, diskless SQLite database in milliseconds for a single task and let it vanish -- a single server hosting millions of them, loading each only when something actually calls it. The company rewrote SQLite's storage engine in Rust to make that work, adding concurrent writes the original format was never built for. (SQLite has run essentially unchanged inside phones, browsers and embedded devices for 25 years precisely because it almost never changes -- rewriting its storage layer for millions of ephemeral agent databases is a bigger engineering bet than the announcement makes it sound.) Superhuman, Sauna.ai and Mastra are already running on it, according to Turso's own announcement. Founder Glauber Costa becomes Supabase's Head of Agentic Services; cofounder Pekka Enberg and the rest of the Turso team join him.

"Agents have become the primary users of technology, and Supabase is the leading database for the agentic era." -- Paul Copplestone, Supabase co-founder and CEO

The Turso deal isn't Supabase's first move toward agent-shaped infrastructure this year. The company separately launched Supabase Compute, hosted sandboxes built for long-running agents rather than short human sessions, and Copplestone has described the platform's whole center of gravity shifting away from a dashboard built for human eyes toward CLIs and MCP servers an agent calls programmatically without a person ever looking at a screen. Turso fills the gap neither product addressed: the short-lived, single-task database an agent spins up and discards, rather than the long-lived one an application keeps for its whole life.

The case for urgency is a real number, even if its precision is softer than it looks. Supabase says it now adds more than 1 million users and 4 million new databases every month, and that 70% of those new databases are created by agents or AI-driven tools rather than a person typing commands -- up from a 600% year-over-year jump in database creation the company reported back in June. More than 13 million developers now build on the platform, Supabase says, up from roughly 10 million in June. The 70% figure is also firmer than the company's own past attempt at the same number: in a July podcast appearance, Copplestone said Supabase could reliably measure only 60% of new databases coming from agents, though he suspected the true share was closer to 90%.

What the $150 million doesn't come with is a price. Supabase has confirmed part of the round will fund employee share sales -- letting early staff and investors cash out existing stock -- alongside new money for the Turso integration, but hasn't said how the total splits between the two, or whether this values the company above, at, or below the $10.5 billion mark it set four months ago. A company that can raise again this quickly, at a number it doesn't have to defend publicly, has less reason to name one than a company that actually needs the market's verdict. The ladder getting there was already steep: a $2 billion valuation in April 2025, roughly $5 billion that October, then $10.5 billion in June -- more than doubling twice within about fourteen months before this latest round arrived without a number attached at all.

Supabase isn't the only platform chasing the same bet. PlanetScale remains independent, built on MySQL and Postgres rather than SQLite; Google's Firebase, which Google acquired back in 2014, already ships a backend-as-a-service inside a much larger company, built around Firestore rather than anything SQLite-shaped. Folding Turso in is Supabase betting that owning the agent-specific layer outright beats partnering for it -- consolidation, not just expansion.

Not everyone reacted to the deal as a clean win. Turso's own announcement thread on Hacker News surfaced real skepticism alongside the enthusiasm -- not about the strategic logic, but about what happens to a smaller team's product once it's inside a bigger platform's roadmap.

The acquisition announcement and the funding round landed in the same breath, each supplying the headline the other was missing: a product story with no price, and a price with no new product yet to show for it. Whether the integration actually ships what Costa has promised -- not just this week's blog posts -- is the part neither side can announce in advance.