Snorkel AI, a startup specializing in the creation of training datasets and simulated environments for artificial intelligence laboratories and corporations, has closed a $350 million Series E financing round. The investment values the seven-year-old company at $3.5 billion.
The latest round was co-led by Insight Partners and S32, bringing Snorkel’s valuation to nearly three times the $1.3 billion figure achieved during its Series D raise 17 months ago. Existing backers, including Addition, Lightspeed, Greylock, GV, and Wells Fargo, also participated in the transaction.
In its early years, Snorkel focused on software designed to automate data labeling. However, the company pivoted last year toward a “data-as-a-service” model, delivering finished datasets directly to clients. Rather than functioning solely as a marketplace for human experts, Snorkel employs a hybrid strategy that combines synthetic data generation through its own software and models with oversight from subject matter experts.
According to the company, its current annualized revenue run-rate has reached $375 million, representing an 18-fold increase over the past 12 months. This surge is largely driven by the growing hunger among AI labs for premium training data.
Other firms in the AI data sector have experienced comparable growth trajectories. Mercor reported gross annualized revenue of $2 billion, while Handshake surpassed the $1 billion mark earlier this year. Additionally, TechCrunch noted that Micro1 scaled to $500 million in revenue. These figures are particularly significant given that such companies typically distribute 60% to 70% of their top-line income to the domain specialists performing the work, meaning their net annual revenue is considerably lower than these gross totals suggest.
Snorkel distinguishes itself by selling reinforcement learning environments and complete datasets rather than raw human labor. Consequently, payments to its human experts are classified under cost of goods sold rather than contributing to headline annualized revenue figures, according to the company.
Snorkel originally launched commercially in 2019, following four years of research conducted by co-founder and CEO Alex Ratner and his team at Stanford University’s AI lab.
Hitting unicorn status three times over in under two years? Snorkel is playing a different financial game now.
Did they really pivot to synthetic data or just rebrand? The line between software and service feels blurry here.
Gross revenue isn’t net profit. Distributing 60-70% to experts probably eats most of that margin.
An 18-fold revenue jump is insane. The hunger for quality training data is clearly outpacing the supply.