Blacksmith Hits $1 Billion as AI Code Creates a New Testing Bottleneck
Blacksmith Hits $1 Billion as AI Code Creates a New Testing Bottleneck
Blacksmith, a continuous-integration startup built around automated software testing, said Wednesday it had raised $66 million in a Sequoia-led Series C at a $1 billion valuation. The round arrives only two months after a $45 million Series B valued the company at $550 million—and roughly a year after its $10 million Series A put its value at $60 million.
Founded in 2024 by chief executive Aditya Jayaprakash, Blacksmith sells the infrastructure that builds, tests and validates code before it is merged and deployed. Its pitch has sharpened as coding tools such as Cursor, Codex and Claude Code make producing software faster: the choke point, Jayaprakash says, is no longer writing the code. “But now the bottleneck has shifted to the second piece, which is: how do I actually validate all this code?”
That shift has translated into brisk growth. Blacksmith says it now serves more than 10,000 companies, double its August total; annual recurring revenue has risen 17-fold into the tens of millions; and validation runs have climbed from 5 million a week at the start of the year to 90 million. The company counts Granola, Handshake, Supabase, LangChain and Mercury among its customers.
Blacksmith’s answer is unusually hardware-heavy. Rather than relying solely on rented capacity from cloud giants, it buys and operates tens of thousands of its own machines in data centers globally, while writing software designed to avoid repetitive setup work during tests. The new capital will largely fund more computing power, even as supply constraints make capacity planning difficult.
The company’s bullish argument is that stronger AI models expand, rather than threaten, its market. Jayaprakash said a July Claude Fable 5 launch drove Blacksmith’s CI volume up 40%, while warning signs remain: model makers, GitHub and coding-agent companies are all moving deeper into automated testing. Blacksmith’s counter is straightforward: “As agents write more of the world’s software, the question that matters is whether it can be trusted.”
Write a comment