What you need to know
- $500M at $10.5B — Supabase closed a Series F led by GIC in early June 2026, with the valuation inclusive of the fresh capital. Total raised now exceeds $1 billion.
- Roughly doubled in eight months — the round landed about seven months after the October 2025 Series E, with the valuation roughly doubling over that period.
- Agents are the growth engine — the user base has more than doubled since the Series E and databases are up around 600% year over year, driven by AI-assisted and "vibe" coding.
- Claude Code leads the inflow — Supabase says Claude Code has been the single largest contributor to new databases since the start of the year, and AI agents now deploy most databases on the platform.
If you build with AI coding agents, you are probably already a Supabase user without having chosen it — the agent picks the backend. Take ten minutes to read the schema and Row Level Security (RLS) policies your agent generated. The default public access patterns that ship from a one-shot prompt are the most common security gap in vibe-coded apps.
What actually happened
On roughly 4 June 2026, Supabase announced a $500 million Series F at a $10.5 billion valuation. The round was led by Singapore's sovereign wealth fund, GIC, with every existing investor participating — Accel, Y Combinator, Craft, Felicis, Peak XV and Coatue. Stripe made its second investment in the company, and Salesforce Ventures joined for the first time. With this round, Supabase has now raised more than $1 billion in total.
The pace is the story as much as the size. The Series F came around seven months after the October 2025 Series E, and the valuation roughly doubled in about eight months. That is not the cadence of a company raising to extend a runway; it is the cadence of a company raising to capture a step-change in demand. Supabase has been explicit about where that demand comes from: AI-assisted development, and increasingly, agents acting on a developer's behalf.
The "agents-as-users" thesis, now with numbers
For two years the line "agents will provision their own infrastructure" sat in conference keynotes as a prediction. Supabase has turned it into a metric. The company reports that its user base has more than doubled since the Series E, that databases are up roughly 600% year over year, and — the line that matters most — that AI agents now deploy the majority of databases on the platform. Claude Code, by Supabase's account, has been the largest single contributor to new databases since the start of the year.
Read that carefully, because it reframes who the customer is. When a developer in Pune or Manchester asks an agent to "build me a booking app with logins and a dashboard," the agent does not file a ticket for a human to spin up Postgres. It calls Supabase, provisions the database, wires up Auth, creates the tables and ships. The human never opens the dashboard. The agent is the user; the human is the intent. Supabase is pricing itself as the default landing pad for that intent, and GIC is paying $10.5 billion to bet that the pattern holds.
It also explains why the investor list reads the way it does. A sovereign wealth fund leading, with Stripe doubling down and Salesforce Ventures joining, is not a syndicate chasing a consumer growth story. It is strategic money positioning around a piece of infrastructure that sits beneath a great many other businesses. When the people who process payments and the people who sell enterprise software both want a stake in your database company, the bet is that the backend is becoming a toll road for the agentic economy — and that the tolls are only going up.
| Round | Date | Valuation | Note |
|---|---|---|---|
| Series E | October 2025 | ~$5B | Pre-agentic-surge baseline |
| Series F | June 2026 | $10.5B | $500M raised; led by GIC; valuation roughly doubled in ~8 months |
| Total to date | — | — | >$1B raised across all rounds |
Why this matters for Indian and UK builders
Supabase — Postgres plus Auth, Storage, Edge Functions and pgvector — has quietly become the default backend for AI-built and indie apps. If you are a solo founder in Bangalore shipping a SaaS over a weekend, or a five-person startup in London standing up an MVP for a pilot client, the odds are high that your agent reached for Supabase before you had an opinion. This raise validates that reflex rather than questioning it.
The dual-market read is genuinely symmetric here. Indian indie hackers benefit from a generous free tier and Postgres skills that transfer to any enterprise job; UK startups benefit from a managed service that handles GDPR-relevant data residency choices and lets a small team punch above its weight. Neither market is a niche for Supabase — both are squarely in the "vibe coding plus serious production" middle that the company is funding its expansion to serve. The same pressure that pushed Cursor to a $50B valuation on the editor side is now flowing to the backend.
Open source means you are not hard-locked in — but the cost curve bites as you scale. The free tier is generous and the early bills are small; the surprise arrives when an agent provisions extra projects, leaves connection pooling misconfigured, or your read traffic grows faster than your revenue. Set spend alerts on day one, and treat the managed conveniences (Auth, Edge Functions, the dashboard) as the parts you would have to rebuild if you ever self-hosted — the Postgres database itself is portable.
The balanced take: portability, cost and the RAG angle
Because Supabase is open source and sits on standard PostgreSQL, your data layer is genuinely portable. You can self-host, you can dump and restore into any managed Postgres, and your SQL travels. That is a real hedge against the platform-risk anxiety that follows every large infrastructure raise. The honest caveat is that portability lives mostly in the database. The auth flows, edge function runtime and dashboard ergonomics are where convenience accrues — and where a migration would actually cost you time. Keep your schema and migrations clean and that bill stays small.
The retrieval angle deserves a measured mention. With pgvector, Supabase lets you store embeddings beside your relational rows and run similarity search in the same database — no second system to operate. For small-to-mid retrieval workloads that is a credible, low-operational-overhead RAG store, and it is one reason agents reach for it: one backend covers the app and its memory. Do not over-claim it, though. Once you are into very large corpora, high query-per-second loads or strict latency budgets, a purpose-built engine may still earn its keep — a trade-off we lay out in our comparison of Pinecone, Qdrant, pgvector and turbopuffer, and in the practical patterns in our production RAG hybrid-retrieval guide.
| Decision | Managed Supabase | Self-hosted Postgres + pgvector |
|---|---|---|
| Time to first app | Minutes — agents provision it | Hours to days of setup |
| Auth, Storage, Edge Functions | Built in | You build or assemble |
| Cost at small scale | Free tier, then predictable | VPS + your time |
| Cost at large scale | Can climb fast — watch it | Lower spend, higher ops burden |
| Lock-in risk | Low on data, moderate on conveniences | None — but you own everything |
Every article here is written by a Verified Builder. Want your name on the next one?
AI Tech Connect lists AI engineers, founders and researchers across India and the UK — and the people hiring browse it to find them. Adding your profile is free.
Become a Verified Builder →Where the agentic-infra wave is heading
The Supabase round is one data point in a broader 2026 pattern: capital is flowing hard into the infrastructure that agents consume, not just the models that drive them. Serverless GPU is seeing the same dynamic — Modal Labs raised a $355M Series C off $300M ARR as agents push inference and training jobs onto on-demand compute. And the metering of agent activity is becoming a first-class product concern, as Anthropic's move to metered credits in the Agent SDK shows. The throughline is simple: every layer an agent touches — editor, compute, database, billing — is being repriced around machine-driven consumption rather than human seats.
For a builder, the practical implication is not to chase the headline valuation. It is to design as though an agent is your busiest user. That means tight RLS policies, spend caps, clean migrations and an explicit answer to "what would it cost to move?" before you scale. The companies raising at these multiples are betting you will not move. The way to keep that as your choice rather than your fate is to keep your data layer boring, standard and portable — which, conveniently, is exactly what Postgres gives you.
The bottom line
Supabase at $10.5 billion is a vote of confidence in a thesis that builders have been living for a year: agents now provision infrastructure, and the backend is the next layer to be repriced around them. Use it — it is genuinely good, and your agent already likes it. Just read what the agent generated, set your spend alerts, and keep your schema clean enough that the door out stays open. For primary detail, see the Supabase/PRNewswire release, plus coverage from CNBC and TechCrunch.