Supabase closes $500M Series F at $10.5B as AI agents take over Postgres provisioning

Supabase raised a $500 million Series F on June 4, 2026, at a $10.5 billion post-money valuation, led by GIC with all existing investors (Accel, Y Combinator, Craft, Felicis, Peak XV, Coatue) participating alongside new entrants Salesforce Ventures and Georgian. Stripe doubled into the company for a second time. The headline is the multiple: $10.5B is up from $5B at the October Series E, a 2x markup in roughly eight months. The substance sits underneath: Supabase says 60% of new databases on the platform are now provisioned by AI coding agents, with Claude Code the largest single contributor in 2026.

What the round looks like

The Series F is the seventh institutional round and pushes total capital raised past $1 billion. CEO Paul Copplestone framed the use of funds across three buckets: accelerating open-source Postgres tooling, supporting operational growth, and providing employee liquidity. The liquidity policy is structured rather than ad hoc. Employees get a 10-year stock option exercise window and can sell up to 25% of vested stock in each funding round, regardless of employment status. That last clause is the unusual one, and it shapes how Supabase is positioning itself against later-stage AI-infrastructure peers still on standard 90-day exercise windows.

Supabase disclosed the operating numbers behind the markup. Database launches grew 600% year-over-year. Nearly 10 million developers now build on the platform, more than double the figure at the Series E eight months ago. Customer count is past 250,000. Supabase for Platforms (the white- label product powering most of the top AI app builders) grew customers 370% in six months. Headcount is north of 350 across 50-plus countries.

Accel partner Arun Mathew was unusually specific in his investor commentary, calling the growth rate “phenomenal” and noting Accel has not seen a database-layer company grow at this pace before. That framing, with its implicit comparison to Accel’s prior database bets at Confluent and parts of MongoDB’s trajectory, is the analyst-facing data point the round is leaning on.

Where this lands in the Postgres market

The structural read is that Supabase is not in the database race anymore so much as in the agent-deployment race. The 60% figure for AI-provisioned databases reframes what Supabase is to a meaningful share of its own pipeline. New projects do not come from a developer reading the docs and signing up; they come from Claude Code or Codex spinning up a backend as part of an agent’s normal workflow. That changes the surface that has to be optimized. Spin-up latency, project-API ergonomics, branching, and the rate at which the agent can iterate are now operating concerns rather than DX polish.

Against Neon, the practical positioning gap narrows. Neon’s branching and serverless Postgres story has been the agent-friendly pitch for two years, and Supabase’s database branching default (shipped in May) plus @supabase/server clearly target that same lane. Against PostgreSQL itself, the question is whether Supabase’s hosted layer plus auth, storage, and edge functions is a sticky enough wrapper to keep self-managed Postgres deployments from clawing back at scale. The Series F gives Supabase another runway to answer that.

The Multigres v0.1 alpha launched alongside the round is the longer-arc bet. Supabase open-sourced it under Apache 2.0 as a horizontal scaling layer for Postgres, with the alpha shipping single-shard high availability on unmodified Postgres replication plus context-aware connection pooling. Sharding, the headline feature, is held back for a later release. Copplestone said the goal is supporting companies “up to the size of OpenAI or even larger.” A hosted Multigres tier on Supabase is on the roadmap. The bet here is that the Postgres-to-other-database migration path most teams hit at scale gets closed off before they walk it.

What’s worth watching

1. Agent-provisioned database revenue mix. 60% of new databases coming from AI coding tools is striking, but the question is what share of monetized revenue those agent-created projects drive. If agent deployments convert to paid usage at parity with developer deployments, the Series F valuation has a clean justification. If they skew toward short-lived prototypes, the valuation depends on a different growth thesis.

2. Multigres adoption curve. The single-shard alpha is the easiest cut. The flagship sharding work is what would justify the “operating system for Postgres” framing. Watch whether a hosted Multigres tier lands inside the next two quarters and whether large customers actually migrate workloads onto it instead of moving off Postgres altogether.

3. The Postgres-vendor consolidation question. Two large Postgres- adjacent raises in twelve months (Neon’s Q4 2025 round and now Supabase’s Series F) plus Multigres open-sourcing positions the category for a compression. The question is whether one of these companies absorbs another by 2027, or whether the database-vendor field stays fragmented and the competition shifts to who ships the better agent integration.

Stackmaven’s follow-up on Supabase’s Series F lands on or around September 3, 2026.

Sources cited

1. Supabase: Supabase Series F supabase.com
2. CNBC: Vibe-coding phenomenon lifts AI startup Supabase to $10.5 billion valuation www.cnbc.com
3. SiliconANGLE: Supabase raises $500M as AI coding tools drive phenomenal growth siliconangle.com
4. Supabase: Multigres v0.1 Alpha supabase.com