Command Palette
Search for a command to run...

Turbopuffer Crosses $100 Million Run Rate in March, Remains Profitable, Company Says

aiai-infrastructureai-inference-platformsbusiness 4 posts · 1 accounts

Turbopuffer said it crossed a $100 million run rate in March, 19 months after crossing $1 million, while remaining profitable after raising less than $1 million. The company, which builds a search engine on object storage for AI workloads, said customers include Cursor, Anthropic, Notion, Cognition, Harvey and Atlassian.

Turbopuffer said it started as a blog post and project in a co-founder's living room in Ottawa, aiming to deliver search at one-tenth the cost of existing systems. After its October 2023 launch, Cursor moved its production workload to the platform after two requested features were shipped, cutting costs by 95%; Turbopuffer said Cursor now searches more than 1 trillion chunks of code on the service.

From the sources (4 posts)

@sirupsen

turbopuffer crossed $100M run-rate in March. 19mo after $1M. Profitable & <$1M raised. Cursor・Anthropic・Notion・Cognition・Harvey・Bridgewater・Ramp・Linear・Legora・Superhuman・Atlassian・Granola We’d be nowhere without them. We work like

@sirupsen

turbopuffer started as a blog post, became a project, and then a startup out of my co-founder Justine’s living room in Ottawa. We were consumed by the idea of building a 10x cheaper search engine on object storage. This architecture had on

@sirupsen

When tpuf launched in October ‘23, it immediately clicked for @cursor_ai. I went to Hayes Valley to a tiny office of ~8 with one of the largest search workloads I had ever seen. We connected on databases, search, and scaling infra. Justin

@sirupsen

We saw this story again and again. The teams at the frontier know the magic happens with AI when it draws in just the right context. They know AI disappoints most when it misses data. That’s a search problem. Our customers use turbopuffer

Preview built on a synthetic news corpus (16 weeks, Apr–Jul 2026). Impact calls are model reads, not price data.

About Archive