Blacksmith just raised $45 million at a $550 million valuation, and the pitch is refreshingly unglamorous for an AI-adjacent startup. It’s not building a coding assistant. It’s building the cloud infrastructure that catches everything a coding assistant produces.
Here’s the problem nobody was really talking about a year ago. AI coding agents can generate pull requests fast, way faster than any human developer working alone. But every one of those changes still has to compile, run through tests, pass whatever checks a team has set up, and survive review, all of which depends on cloud compute sitting underneath the whole process.
Blacksmith says CI jobs on its cloud platform have grown 5 to 10 percent week over week since January, and its customer base jumped from roughly 800 companies last September to more than 6,000 now. Company-reported numbers deserve a little skepticism, sure, especially since that kind of percentage growth gets harder to sustain as the underlying cloud workload scales up. Still, the direction is hard to argue with.
The core issue is that generating code faster doesn’t automatically mean shipping software faster. If an AI agent can churn out several times more pull requests than a human developer, the bottleneck just moves downstream, into build pipelines, test suites, and the cloud infrastructure that has to actually validate all that new code. Blacksmith’s answer is to throw a lot of dedicated cloud compute at the problem. The company already manages hundreds of thousands of cores across its cloud fleet, and it plans to spend most of this new funding scaling that footprint up by roughly an order of magnitude.
That’s a genuinely capital-intensive bet for a software company to make, and it comes with its own balancing act unique to cloud economics. Too little capacity and CI performance suffers. Too much and expensive cloud hardware just sits idle between bursts.
Beyond raw infrastructure, Blacksmith has also built Codesmith, a cloud-based coding agent that can diagnose CI failures and attempt fixes on its own, plus a QA tool in development meant to test changes autonomously before merging. That’s arguably the more interesting bet long-term: using automation to validate automation, which only works if the underlying tests running in the cloud are actually good enough to trust.
Led by Peak XV Partners, with Y Combinator and GV returning as investors, the round suggests at least some confidence that this cloud validation layer keeps growing even as individual AI coding tools come and go.
