Blacksmith has raised $66 million in a Series C led by Sequoia Capital at a $1 billion valuation, according to the company’s October 21 funding statement. Premji Invest joined existing backers Peak XV Partners, GV and Y Combinator in the round, which takes total funding to $125 million.
In a separate founder announcement, Blacksmith says the capital will expand its team and computing capacity. It manages just under 10 megawatts across data centres in the United States and Europe, and aims to multiply that footprint fivefold in 2027. The planned expansion targets the infrastructure that builds and tests software before changes are merged.
Sequoia’s investment note reports more than 10,000 companies using Blacksmith, compared with 700 a year earlier. It puts weekly continuous integration (CI) jobs at 90 million, up from 5 million earlier this year, and says annual recurring revenue has risen 17 times. The largest customers now spend more than $1 million a year, it says.
The investor’s account describes infrastructure built from bare metal to absorb sharp changes in demand. It reports faster jobs and lower costs, but those are backer-published claims, not independent test results. Its central argument is that coding agents produce successive attempts at software changes, each requiring checks before it can be merged. Founded in 2024, Blacksmith has three co-founders: Aditya Jayaprakash, Aayush Shah and Aditya Maru.
Power availability shapes expansion
An interview published by Premji Invest on October 21 gives more detail on the operating challenge. Co-founder and CEO Aditya Jayaprakash described power availability as a constraint. He discussed smaller sites with at least a megawatt of capacity, alongside the work of opening them, routing jobs between regions and maintaining storage.
The interview’s recording date is not stated. Jayaprakash explained that a code submission can require thousands of virtual processors for ten to thirty minutes, followed by demand falling to zero. He said Blacksmith deliberately prioritises execution speed over durability for these short-lived workloads, unlike a cloud built to support many different applications.
His commercial argument also separates the software that organises CI jobs from the infrastructure that executes them. Customers can spend much more on the latter, he said, while that cost appears within a broader cloud bill. Blacksmith is asking teams to move that particular workload out of their general-purpose cloud rather than replace their entire computing setup. He also described self-service onboarding, with sales teams intended to remove obstacles rather than sell the product.
The investor case for specialist infrastructure
Premji Invest’s separately authored investment thesis addresses why that distinction matters to the financing. It says the orchestration layer was often bundled with existing platforms, making life difficult for independent suppliers. The firm believes the opportunity is in the underlying computing capacity, particularly when many coding agents work simultaneously and push demand above its previous peaks.
The thesis also identifies waiting time as a cost: companies can pay for AI-generated code and then leave it queued for validation. Premji’s bet is that faster infrastructure will take a larger share of this work from general-purpose providers. That is a forecast by an investor participating in the round, not a confirmed market-share outcome.
The firm expects adoption to begin with developers because Blacksmith fits into tools they already use. It presents compatibility as an advantage over asking engineering teams to change their whole workflow. The investment case therefore depends on winning a specific workload inside existing software operations, rather than persuading customers to abandon every incumbent platform.
What the funding does and does not establish
There is separate evidence of the pressure on test systems. In a September 14 engineering account cited by Blacksmith and Sequoia, Anthropic reported a 25-fold rise in CI jobs over six months. It redesigned its own test-selection service after temporary fixes stopped keeping pace. This describes Anthropic’s internal infrastructure, not a Blacksmith customer endorsement or an independent performance benchmark.
Sequoia says Blacksmith has further launches planned in the coming months, without identifying them or giving release dates. The funding announcement supplies a clearer near-term direction: more staff and computing capacity to support the validation workload.
Related Dalimss News coverage includes Flipkart Ventures’ backing of Keploy for software testing and Autoheal’s seed round for work after code is written.
