Blacksmith, a US start-up that provides infrastructure for testing software before it is released, has reached a $1 billion valuation after raising $66 million from investors led by Sequoia Capital.
The Series C funding comes about two months after the company secured $45 million in a round led by Peak XV Partners, which valued it at $550 million. Blacksmith has now raised about $124.5 million in total.
Founded in 2024 by chief executive Aditya Jayaprakash, the company provides continuous integration, or CI, services. Its platform automatically builds and tests code before it can be merged into an existing codebase and deployed.
Blacksmith said more than 10,000 companies now use its service, up from about 5,000 in August. Its customers include Granola, Handshake, Supabase and LangChain.
The company was running about five million validation checks a week at the start of the year. That figure has since risen to 90 million, while annual recurring revenue has increased 17-fold, according to Jayaprakash, who said revenue was now in the tens of millions.
Blacksmith said its largest customers spend more than $1 million a year. The business has about 40 employees in San Francisco and New York and plans to increase its workforce to roughly 75 by the end of the year.
AI coding shifts the testing bottleneck
Jayaprakash said most of the new funding would be used to buy additional computing power, as the growth of AI-assisted software development increases demand for systems that can check code quickly.
Blacksmith buys its own servers rather than relying entirely on rented capacity from Amazon Web Services, Google Cloud or Microsoft Azure. It operates tens of thousands of machines in data centres around the world and has developed software to manage them, including tools that avoid repeating the same preparatory work for every test.
Jayaprakash said developers had traditionally spent roughly half their time writing code and the other half waiting for it to build and pass checks. Coding tools such as Cursor, OpenAI’s Codex and Anthropic’s Claude Code had made the first part “extremely easy”, he said.
“But now the bottleneck has shifted to the second piece, which is: how do I actually validate all this code?” he said.
Anthropic said last month that its own validation workloads had grown 25-fold in six months as Claude came to account for most of its code output.
The increased demand has also made capacity planning more difficult for Blacksmith. Jayaprakash said demand rose whenever a more capable model was launched, with CI volume on the platform increasing by 40% after Anthropic released Claude Fable 5 in July.
Owning its hardware gives Blacksmith greater control over performance, Jayaprakash said, but leaves it exposed to wider shortages of computing capacity and memory.
The company also faces competition from the AI model providers themselves. Codex and Claude Code can already run tests while working on projects, while Cursor launched a product in March allowing companies to set up agents to review and test code automatically. Microsoft-owned GitHub also sells hosted machines for running CI jobs.
Jayaprakash said those services would ultimately complement rather than replace Blacksmith, arguing that more capable coding models would produce more software requiring validation. “Usage on us actually goes up,” he said.
