A startup that checks AI-written code just jumped from a $60 million to a $550 million valuation in under a year
Blacksmith raised $45 million to solve a problem AI coding tools quietly created: who checks the code the machines wrote?
Key points
- Blacksmith raised a $45 million Series B in 2025, valuing the company at $550 million, up from $60 million less than a year ago.
- The startup grew from roughly 700 customers to more than 5,000 in under twelve months.
- Revenue reached "tens of millions of dollars" annually, with some customers spending over $1 million a year on the platform.
- Blacksmith now employs around 30 people, up from just 10 when it first hit a $10 million annual revenue run rate.
- Competitors include GitHub Actions, Cursor Automations, Amazon Web Services and Google Cloud.
AI tools can write code faster than most human developers. That sounds like a win. But someone still has to check whether that code actually works, and right now that checking step is a serious bottleneck.
Blacksmith, a San Francisco startup founded in 2024, thinks it has the answer. This week it announced a $45 million Series B funding round, led by venture firm Peak XV Partners, with existing backers GV and Y Combinator also putting money in. Total funding now stands at $58.5 million.
The valuation jump is striking. Less than a year ago, the company raised a $10 million Series A at a $60 million valuation. Today it is valued at $550 million, nearly tenfold in under twelve months.
What does Blacksmith actually do?
Blacksmith helps software teams catch bugs before code goes live. It started as a platform for continuous integration (CI), meaning machines automatically double-check code every time a developer makes a change. More recently it added Codesmith, an AI agent (software that carries out multi-step tasks on its own) that can automatically fix code that fails those checks.
CEO Aditya Jayaprakash told TechCrunch AI that coding tools such as Cursor, OpenAI's Codex and Anthropic's Claude Code are generating more code than ever, without guaranteeing its quality. "Validating code is still a bottleneck, and it's an even bigger bottleneck because people are writing even more," he said. More code means more to verify. That is Blacksmith's pitch.
Is the growth real?
The numbers suggest so, though context matters. Customer count jumped from about 700 to more than 5,000, with clients including financial software company Mercury and expense-management platform Expensify. Revenue moved from a $10 million annual run rate to "tens of millions of dollars," though Jayaprakash declined to give a precise figure. The company hit that first milestone with 10 employees and has since tripled its headcount to around 30.
For any developer already leaning on AI to write their code, the practical point is this: the check-your-work step did not disappear. It just got more complicated and, for startups like Blacksmith, more commercially interesting. We first reported on the company on 12 August 2026.
Should anyone be worried about the competition?
Blacksmith is not alone here. GitHub Actions, Cursor Automations, Amazon Web Services and Google Cloud all offer code-testing tools, and Anthropic bakes validation capabilities directly into Claude Code. That is a crowded field for a thirty-person startup.
Jayaprakash says Blacksmith competes on testing speed and price, and plans to expand into a fuller suite covering writing, validating and merging code. Whether that's enough to hold ground against the big cloud players is the question a $550 million valuation needs answered. Our earlier story on Indian AI startup Emergent, which hit a $1.5 billion valuation in July, showed how fast AI-tooling multiples can inflate. Blacksmith's ratio of valuation to headcount is even more extreme, and that is the number worth watching as the market cools.
