If you’ve dabbled in vibe coding or built AI applications, the name Supabase might already sound familiar. It’s one of the most trusted default backends for countless indie developers worldwide, and over the past year, it has become a hotly pursued open source giant in the venture capital world.
On June 5, Supabase announced the completion of a $500 million Series F round. With that, the open source database company reached a valuation of $10 billion, officially crossing the threshold into decacorn territory (a private company valued above $10 billion). Just a year ago, it was worth only $2 billion.
Over the past year, AI coding tools have enabled more non‑professional programmers to build applications, driving rising demand for backend services. Supabase’s user base has nearly doubled from about 5 million to almost 10 million. More than 60% of new databases on the platform are now created by AI tools.
In addition, major vibe coding platforms such as Bolt, Lovable, and Replit have successively set Supabase as a default integration. In the blog post announcing the funding, Supabase CEO Paul Copplestone gave special shout‑outs to Claude Code and Codex.
Building an open source backend that’s better than Firebase
An application consists of two parts. The frontend is what users see; the backend handles data storage, user identity management, file uploads, and state synchronization across multiple users.
For a long time, building the backend was one of the most time‑consuming tasks: setting up a database from scratch, writing authentication logic, configuring file storage. It could take up half of a project’s total development time. To improve development efficiency, Backend‑as‑a‑Service (BaaS) platforms emerged, packaging the backend into an out‑of‑the‑box product.
Launched in 2011, Firebase was one of the first BaaS platforms. Acquired by Google in 2014, it has since become the standard choice for mobile development, powering about 72% of Android apps. But Firebase has long been criticized for two issues: it is closed source and deeply tied to Google’s ecosystem; and its underlying NoSQL document database, Firestore, cannot support applications that need complex queries and relationships.
Supabase was born because Paul himself was “burned” by Firebase. While working at a startup in Singapore around 2019, he used Firebase to build a product for the first time. He was initially amazed by its convenience, but soon ran into unsolvable problems—limited scalability, difficulty customizing functionality, and so on. He tried switching to the open source relational database PostgreSQL (Postgres), but the data format incompatibility made migration extremely costly.
The idea of combining Firebase’s ease of use with Postgres’ flexibility became the starting point for Supabase. In January 2020, Paul and co‑founder Ant Wilson officially started the company. They kept Firebase’s out‑of‑the‑box backend convenience, but replaced the underlying database with Postgres, which had been evolving for 30 years. They also pledged that the platform would be fully open source, self‑hostable, and free of vendor lock‑in.
After joining Y Combinator, the company repositioned itself as an open source alternative to Firebase. In the spring of 2020, Supabase hit the front page of Hacker News for two consecutive days, and its user count grew from 80 to 800. Yet the buzz didn’t immediately translate into widespread adoption—by the end of April, the total number of databases hosted on the platform was only eight.
Over the next three years, Supabase gradually expanded its product portfolio, evolving from a pure real‑time database into a full‑featured backend platform offering authentication, storage, edge functions, vector databases, and more.
The force that truly made the flywheel spin didn’t appear until the second half of 2024.
Vibe coding ignites growth
From late 2024 into 2025, a wave of AI coding tools—Bolt.new, Lovable, Cursor, Replit, and others—exploded onto the scene. People with no programming background could generate working applications through natural language conversations. That’s vibe coding.
At the same time, countless indie developers got their first taste of using AI to generate code and build personal applications. To give their apps a reliable backend, they turned to Supabase.
One reason is that AI model training data already contains a large amount of relevant open source code, making calls to Supabase far more successful than calls to other backend options. Consequently, major vibe coding platforms have progressively set Supabase as the default integration: when a user builds an application and publishes it, the moment an end‑user triggers a registration, Supabase automatically takes over a cascade of backend operations—creating tables, configuring authentication, generating APIs, and more.
Thus the data flywheel began to turn. At the end of 2024, Supabase’s annual recurring revenue (ARR) was about $30 million. By August 2025, that number had jumped to $70 million, a 250% year‑over‑year increase. The number of developers doubled from roughly 5 million to nearly 10 million, and the pace of fundraising accelerated.
The winner in a crowded field
BaaS is a highly competitive space. Firebase remains the benchmark for mobile development. Appwrite, a popular open source alternative that emphasizes full self‑hosting, has already earned over 50,000 stars on GitHub. Developers who want the ultimate Postgres experience can choose Neon; those who advocate extreme simplicity can turn to PocketBase, which packages the entire backend into a single executable. Then there are platforms with different focuses, such as AWS Amplify, Hasura, and Nhost.
Supabase has stood out for two reasons: the enormous tailwind from the vibe coding era, and its active cultivation of the developer community. Paul has said that serving the broad community of developers generates greater long‑term returns than catering to enterprise customers.
As a developer once frustrated by Firebase, Paul has expressed caution about large enterprise deals in many interviews. He believes that once a company starts accepting custom‑built enterprise contracts, product direction can become distorted by those customers’ specific needs—adding complex permission systems, rarely used compliance features—ultimately hurting the experience for ordinary developers.
But at a $10 billion valuation, investors have legitimate reasons to demand returns to match. Supabase must find ways to profit from the enterprise market without sacrificing developer experience, in order to hedge against volatility in the consumer market.
Its solution is to attract enterprise customers through product strength. For instance, its white‑label 2B product, Supabase for Platforms, serves as the default backend foundation of choice for AI code platforms, and its customer base has grown 370% over the past six months, indirectly securing revenue from a vast number of downstream enterprises.
Additionally, on the same day it announced the funding, Supabase released a tool called Multigres, aimed at solving operational challenges as users’ projects scale. Postgres’s longstanding difficulty with horizontal scaling has never been fully resolved; projects built on it, once they grow large, have to migrate to more complex distributed databases. Multigres is described as an operating system for Postgres, providing an open source horizontal scaling layer with sharding and zero‑downtime migration capabilities. With it, Supabase hopes to retain customers who started with AI coding and may one day grow into the next unicorn.
The flywheel is spinning, but it could reverse course
GitHub hosts hundreds of millions of repositories, but a large fraction of them are “graveyards”—codebases uploaded on a whim and never touched again.
Similar challenges await Supabase. Over the past year, the number of databases launched on Supabase has increased by 600%. But behind those impressive numbers, it’s impossible for outsiders to know what proportion of the users drawn in by the hype—and the applications they build—represent active, paying accounts versus zombie projects.
Security is another issue. Supabase relies heavily on Postgres’s Row Level Security (RLS), which determines users’ read/write permissions. In vibe coding, however, AI agents often automatically disable RLS policies or write rules with security holes just to get a project working—something typical users without security awareness find hard to detect.
The result is that a large number of AI‑generated Supabase backends are essentially “naked” in terms of data protection. These risks haven’t yet led to widespread incidents, but if not addressed in time, they could eventually backfire.
AI is rapidly expanding the population of developers, and the infrastructure market that serves them is growing accordingly. Supabase’s open source, disciplined, and community‑friendly stance is, for now, an optimal strategy during a phase of hypergrowth. But in the evolution of developer tools, many products have gone from open source darlings to community cast‑offs.
The real test will come after growth slows. When the tide of vibe coding recedes and investors begin demanding profits, we will see whether Supabase remains developers’ “most loyal friend” or becomes yet another story of community disappointment.