A Database for Every Agent: Supabase Raises $150M and Buys Turso

Supabase, the open-source Postgres development platform that has quietly become the default backend for the AI coding boom, announced on October 2 that it has raised $150 million in new funding led by Singapore’s sovereign wealth fund GIC, with participation from Alphabet’s CapitalG, IronArc, and SquarePeg. Alongside the round, the company confirmed it is acquiring Turso, a database startup built from the ground up for agentic workloads. The deal terms for the acquisition were not disclosed.

The speed of this round is the story’s first signal. This is Supabase’s second nine-figure raise in four months — it closed a $500 million Series F at a $10.5 billion post-money valuation in June 2026, bringing total capital raised to more than $1 billion. Companies rarely raise again this quickly unless growth is outrunning even an aggressive plan, and Supabase’s own numbers explain why: the platform is now adding more than 1 million users and 4 million databases per month, and 70% of those new databases are created by agents or AI-driven tools rather than by human developers clicking through a console.

The agents are already here

That 70% figure deserves a moment of pause. It means the majority of new database provisioning on one of the world’s fastest-growing backend platforms is no longer initiated by people. Claude Code, OpenAI’s Codex, and other coding agents spin up a Supabase project as a routine step in scaffolding an application — the agent writes the code, the agent provisions the backend, and the human reviews the result. Supabase reported a 600% year-over-year increase in databases back in June, when agents were already deploying the majority of new databases; that wave has only compounded since.

Supabase frames its own trajectory as two successive AI-driven expansions. First, AI app builders opened software development to a much broader audience — people who could describe what they wanted and have a model build it. Then coding agents became capable of constructing increasingly sophisticated applications on their own, multiplying the amount of software (and therefore backends) being created. The company’s stated goal is to be “the first place developers and agents turn when they need backend support” — a deliberate reordering in which the agent is treated as a first-class customer.

This is also why the round includes a liquidity component for employees, which the founders describe as “an important and vital part of being the best company for engineers.” Secondary liquidity inside a primary round is a common hallmark of companies with leverage to pick their investors rather than chase them.

What Turso brings

The Turso acquisition targets the specific economics of agent-created databases. An agent should be able to create a database instantly, and platforms should be able to launch enormous numbers of isolated databases without paying the overhead of a dedicated machine for each one. That is a fundamentally different requirement from the traditional model of one carefully-tended database serving an entire application for years.

Turso’s answer is built on SQLite — the ubiquitous embedded database with a tiny hardware footprint — extended with features agents need. Turso supports embeddings for AI workloads, encrypts each database page in a slightly different way to reduce attack surface, and is designed so an agent can spin up a lightweight instance nearly instantaneously. Turso’s architecture enables on-demand per-agent databases both in the public cloud and in bring-your-own-cloud (BYOC) deployments, and it already serves customers including Superhuman, Sauna.ai, CTO.new, and Mastra.

Turso founder Glauber Costa joins Supabase as Head of Agentic Services, alongside co-founder Pekka Enberg and the rest of the Turso team. Costa’s stated ambition frames the scale Supabase is planning for: “From the start of our journey, I’ve had as a goal to serve upwards of a billion databases to our customers. Being a part of Supabase, with its gargantuan developer success and ever-growing partner ecosystem will bring this goal closer.” A billion databases is not a typo-level stretch in a world where every agent session might warrant its own isolated store.

The Turso platform will continue operations, with what Supabase calls “a clear graduation path into the rest of the Supabase ecosystem” — suggesting Turso’s edge-first SQLite lineage will live beside, not inside, the core Postgres offering.

Supabase Compute: a runtime for long-running agents

The same day, Supabase also debuted Supabase Compute, a new cloud service providing hosted sandboxes for long-running AI agents. Where a full Postgres instance is overkill — an agent that just needs to persist notes, cache intermediate results, or track task state between steps — a lightweight Turso-style instance inside a Compute sandbox is the right-sized tool.

The combination repositions Supabase from “Firebase alternative” to something more ambitious: an agent-infrastructure backbone. Managed Postgres for the application tier, instant per-agent SQLite for the agent tier, and sandboxes for the compute tier, all under one platform with more than 13 million registered developers and a marketplace of 100+ integrated partner tools.

Analysis: the pick-and-shovel layer consolidates

Two structural shifts are visible in this deal. The first is that agent-created infrastructure has become a load-bearing assumption for backend platforms. When 70% of new databases come from agents, provisioning cost per database, isolation between agent workloads, and instant spin-up stop being nice-to-haves and become the core product requirement. SQLite-based architectures are a rational response: tiny footprint, embedded operation, and per-page encryption map well onto thousands of ephemeral, security-sensitive agent stores.

The second is where the capital is coming from. GIC leading, CapitalG (Alphabet’s growth fund) participating, four months after a $10.5 billion Series F — sovereign wealth and strategic growth money is consolidating around the infrastructure layer of the AI stack, echoing the BIS’s recent warning about circular AI financing. Notably, Supabase’s valuation in this round was not disclosed, which often indicates terms set for growth and liquidity rather than a new benchmark.

There are open questions. Turso’s cloud product has existing paying customers, and migrations of that kind rarely go perfectly. Postgres and SQLite are philosophically different databases, and Supabase will need to keep the story coherent — “which database should my agent use?” is a question the platform now answers with “it depends.” And the entire thesis leans on continued agent-driven growth; if agent-created database volume plateaus, a billion-database architecture is expensive to carry.

But the direction is clear. The buyers of backend infrastructure increasingly are not humans, and the platforms that treat agents as their primary customer — with instant provisioning, isolated lightweight stores, and hosted runtimes — are where the growth capital is flowing. Supabase just made that bet explicit, and bought the team whose architecture was built for exactly this future.