Snorkel AI, a company specializing in building training datasets and simulated environments for AI laboratories and enterprises, has closed a $350 million Series E funding round that values the startup at $3.5 billion.

The round was co-led by Insight Partners and S32, and the valuation it produced represents nearly triple the $1.3 billion figure attached to the company just 17 months ago, when Snorkel raised $100 million in its Series D. A group of existing backers also joined the latest round, including Addition, Lightspeed, Greylock, GV, and Wells Fargo.

Founded seven years ago, Snorkel AI originally focused on software tools designed to automate the labeling of data. Over time, however, the company pivoted its core offering. Last year, it began delivering fully completed datasets directly to clients - a model it markets as data-as-a-service. Rather than functioning as a straightforward marketplace connecting customers with human annotators, Snorkel employs a hybrid methodology that combines its proprietary software and machine learning models to generate synthetic data, while also drawing on contributions from domain-specific subject matter experts.

Revenue Growth Reflects Surging Demand

Snorkel reports that its current annualized revenue run rate has reached $375 million - an increase of roughly eighteen times over the previous twelve months. The company attributes this sharp growth trajectory to the rapidly intensifying demand from AI laboratories for high-quality, specialized training data. As frontier model development accelerates, the need for curated, domain-specific datasets has become a central bottleneck in the industry, and companies like Snorkel are positioned directly at that intersection.

Snorkel is not alone in experiencing this kind of explosive momentum. Several other companies that have repositioned themselves as AI-focused data providers have reported similarly dramatic revenue figures. Mercor has disclosed a gross annualized revenue figure of $2 billion, while Handshake crossed the $1 billion threshold earlier in 2025. Separately, Micro1 has reportedly scaled to $500 million in annualized revenue.

An important distinction applies when evaluating those headline numbers. For companies that operate primarily as human expert marketplaces, a significant share of gross revenue - typically between 60% and 70% - flows directly to the specialists performing the work. That means their actual net revenue is considerably lower than the top-line figures being cited publicly.

A Different Revenue Structure

Snorkel's financial structure differs from those marketplace-style competitors in a meaningful way. Because the company sells complete datasets and reinforcement learning (RL) environments rather than human labor as a direct service, the payments made to its human contributors are recorded as a cost of goods sold. This means those expenses do not inflate the company's headline annualized revenue figure in the same way that gross marketplace payouts would, according to the company. The $375 million run rate Snorkel is reporting is therefore more directly comparable to a traditional software or services revenue figure.

Reinforcement learning environments, in particular, have become an increasingly critical component of advanced AI development. These simulated settings allow AI systems to be trained through trial-and-error interactions, and building them at scale requires both sophisticated tooling and carefully curated data - areas where Snorkel has invested heavily since its inception.

Origins in Academic Research

Snorkel AI's commercial operations launched in 2019, but the company's roots stretch back further. Co-founder and CEO Alex Ratner and his colleagues spent four years conducting foundational research at a Stanford University AI laboratory before bringing the technology to market. That academic foundation has informed much of the company's approach to programmatic data labeling and synthetic data generation, which remain central to how Snorkel delivers its data-as-a-service offering to clients today.

With the latest funding secured and revenue growing at a pace few technology companies achieve at this stage, Snorkel AI appears well-positioned to expand its footprint in a market that shows no signs of slowing. The global race to develop more capable AI systems continues to drive demand for the kind of high-quality, structured training data that Snorkel specializes in producing.