Snorkel AI, which supplies training and evaluation data to AI labs and large companies, raised $350m in a Series E round at a $3.5bn valuation, it said on 22 September 2026.
Insight Partners and S32 co-led the round, with significant participation from existing investor Addition. New investors March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard and Third Point Ventures joined, along with existing backers Greylock, Lightspeed, GV, Factory, Prosperity7, Walden Catalyst and Wells Fargo.
According to TechCrunch, the valuation is nearly three times the $1.3bn the company was valued at when it raised a $100m Series D 17 months earlier, in May 2025.
“The teams pushing the frontier want a research data partner who pioneers the science of data development. That’s what Snorkel was built to be: the frontier lab for agentic data, combining human excellence with over a decade of research and technology.”Alex Ratner, Co-founder and CEO, Snorkel AI
What Snorkel AI does
Snorkel was founded in 2019 out of the Stanford AI Lab. TechCrunch reported that it launched commercially after about four years of research by co-founder and chief executive Alex Ratner and his team, and that it originally sold software to automate data labelling.
The company changed its focus after launching an expert data-as-a-service offering in September 2025. It now sells finished datasets, reinforcement-learning environments, benchmarks, evaluations and grading rubrics to AI labs and enterprises, which it describes as an agentic data factory. According to TechCrunch, it uses a hybrid method in which its own software and models generate data synthetically alongside human subject-matter experts.
Snorkel argues that frontier and agentic AI systems need complex, expert-built tasks and environments that can take skilled people hours or days to construct, rather than the simple labelling work of earlier years. It says its founding team's research spans more than 250 peer-reviewed papers. Techstrong.ai reported that its customers include AI labs, enterprises and government agencies.
Revenue growth
Snorkel said its annualised revenue run rate had passed $375m, more than 18 times higher than less than a year earlier. The figure was provided by the company and is not audited annual revenue. According to TechCrunch, Snorkel said payments to the human experts it works with are counted in cost of goods sold rather than in the headline run rate.
TechCrunch noted that other companies selling expert data to AI labs report larger gross figures: it cited Mercor at $2bn in gross annualised revenue and Micro1 at a $500m gross run rate. It added that such firms typically pass 60% to 70% of their top-line income to the specialists who do the work, so their net revenue is much lower.
Snorkel did not disclose its headcount or whether it is profitable.
What the money is for
Snorkel said it would use the funding to expand the capacity of its data factory to meet demand, invest more in vertical and enterprise AI, and take its research and technology into new domains and modalities. It also plans to put more money into open research, including its Open Benchmarks Grants programme.
Lonne Jaffe, a managing director at Insight Partners, said the firm expected Snorkel to apply its approach to longer-running, complex work in industries such as healthcare, law and software engineering. Andy Harrison, chief executive and general partner of S32, said working with Snorkel creates a partnership between model developer and data provider.
Company website: snorkel.ai