A Startup That Tests Software Built by AI Just Hit a $550 Million Valuation

Blacksmith, a company that provides cloud infrastructure for testing software, has raised $45 million in a funding round led by Peak XV Partners, valuing the company at $550 million. The round brings Blacksmith's total funding to $58.5 million and marks a roughly 10x jump from the $60 million valuation the company held less than a year earlier. Existing investors GV and Y Combinator participated. The valuation and revenue figures were disclosed in an exclusive TechCrunch interview with co-founder and CEO Aditya Jayaprakash (TechCrunch).

Founded in 2024, Blacksmith began as a cloud provider for what developers call "continuous integration," or CI. CI is the automated process that takes new code, compiles it, runs tests against it, and checks whether it is safe to release to users. Think of it as a quality-control checkpoint on a factory line: every batch of code has to pass inspection before it ships. Blacksmith has since added Codesmith, an AI tool that automatically fixes code when those checks fail.

Blacksmith's customer base has grown from more than 700 to more than 5,000 in less than a year, according to Jayaprakash. Customers include Mercury, Supabase, Clerk, Ashby, and Expensify. Some of Blacksmith's largest customers spend more than $1 million annually on the platform.

The revenue numbers are notable even by the standards of recent tech startups. Blacksmith reached a $10 million annual revenue run rate with just 10 employees and has since grown to approximately 30 employees, with revenue now in the "tens of millions of dollars." That kind of revenue-per-employee ratio tends to attract investor attention, particularly in a funding environment where growth-at-all-costs narratives have given way to more disciplined deployment.

Blacksmith's prior round was a $10 million Series A led by Google Ventures, reported in September 2025 (TechCrunch). GV's continued participation in the latest round, alongside Y Combinator, signals sustained conviction from early backers as the company's valuation scaled.

The broader context here is that AI-assisted coding tools are generating substantially more code, which in turn generates substantially more testing work. Every line of code produced by an AI assistant still needs to be compiled, tested, and validated before it reaches real users. If anything, the volume of code flowing through these testing pipelines may be growing faster than the capacity of engineering teams to review it manually. That creates direct demand for both faster testing infrastructure and automated fixing of build failures. Codesmith, Blacksmith's tool that auto-fixes failed checks, sits squarely in that gap.

This combination illustrates a second-order effect worth understanding. The primary products in the AI coding space are assistants and agents that write code. But the infrastructure layer that has to absorb all that new code is itself a market. When an AI coding tool generates code that fails a test, the cost of that failure — in developer time and computing resources — does not disappear. It shifts downstream. Companies like Blacksmith are positioning to capture the value of handling that shift.

Peak XV Partners leading the round also carries a signal. The firm, formerly Sequoia Capital India, has been expanding its investments in U.S.-based developer infrastructure and enterprise software. For a two-year-old company with fewer than 50 employees to command a half-billion-dollar valuation, the implicit thesis is that testing infrastructure augmented by AI can become a defining layer in the developer toolchain.

The risk, as with any infrastructure startup riding a demand surge, is that the underlying driver could shift. If the major AI companies and software-tool vendors build testing and auto-fix features into their own platforms, companies like Blacksmith could lose their edge. Blacksmith's bet is that specialized, high-performance testing infrastructure combined with AI-powered fix capabilities will retain value even as general-purpose AI coding tools mature.

For now, the numbers speak for themselves. A company founded in 2024, with roughly 30 employees and tens of millions in revenue, has convinced investors to place a $550 million bet that the testing layer is where the AI coding boom lands.