PgDog raises $5.5M to scale Postgres with a proxy in front of it

What PgDog announced

PgDog, an open-source proxy that sits in front of PostgreSQL, announced on June 10, 2026 that it has raised $5.5 million. The post was written by the project's founder, Lev Kokotov, who says the money comes from Basis Set, Y Combinator, Pioneer Fund and other investors, and gives the three-person team years of runway.

The pitch is deliberately narrow. “Postgres is the only database you need,” Kokotov writes, arguing that databases such as MongoDB and DynamoDB exist mainly because PostgreSQL has a scaling problem. The goal is to make Postgres handle 100 TB-plus tables and one million queries per second, so that teams do not have to reach for anything else.

Same Postgres, with a proxy in front

PgDog does not fork or modify PostgreSQL. It is a proxy that is deployed in front of an existing database and makes it horizontally scalable. Installation is meant to be trivial: pull the Docker image, change the DATABASE_URL, and let the proxy do the work.

According to the announcement, PgDog runs wherever it is needed — on-premises, in a customer's own cloud account, in a colocation rack or on a laptop — with no dependencies and no hidden serverless costs. The proxy is multithreaded and will use every CPU it is given.

Production numbers

The project reports that it serves more than two million queries per second in production across dozens of deployments, and has sharded more than 20 TB of data that the team knows about. PgDog is open source and has passed 1.4 million Docker pulls on GitHub. A new release ships every week, on Thursdays, and the team says it answers questions in its Discord community daily.

Why the team says it can be trusted

Kokotov addresses the obvious objection for a small startup handling production data: PgDog was built by three infrastructure engineers who ran Postgres at large scale before. He says he ran Postgres at Instacart, where the company grew five-fold in April 2020, and the hardest problem was making the database serve hundreds of thousands of grocery delivery orders per minute. The team solved it by sharding Postgres on RDS, Aurora and EC2, and the same technology is now packaged as an open-source product.

The company is also building an enterprise edition meant to make PgDog easier to run on AWS, with SLA-backed support, and is inviting interested teams to get in touch.