Snorkel AI, a startup that helps AI labs and companies build training data sets and simulated environments, has raised $350 million in Series E at a $3.5 billion valuation.
The new round, led by Insight Partners and S32, values the seven-year-old startup at nearly triple the $1.3 billion valuation it received when it raised a $100 million Series D 17 months ago. Existing investors including Addition, Lightspeed, Greylock, GV, and Wells Fargo also participated in the round.
Snorkel initially offered software for automating data labeling, but last year it moved to providing customers with completed datasets, which it calls data-as-a-service. Rather than operating as a purely human expert marketplace, Snorkel relies on a hybrid approach that uses software and models to generate data holistically with subject matter experts.
Snorkel said its current annual revenue run rate is $375 million, an increase of 18x in the past 12 months. This growth is being driven by AI Labs’ insatiable appetite for high-end training data.
Other data companies that are positioning themselves as AI data labs are experiencing similar explosive growth. Mercor’s total annual revenue has risen to $2 billion, Handshake reached the $1 billion milestone earlier this year, and TechCrunch reported that Micro1 has scaled to $500 million. It’s important to note that these companies pay approximately 60% to 70% of their top-line revenue directly to the domain specialists who do the work, so their actual annual net revenue is significantly lower than the headline gross revenue figure.
The company says that because Snorkel sells reinforcement learning (RL) environments and complete datasets rather than human labor, payments to human experts are included in cost of goods sold rather than in headline-generating annual sales.
The snorkel was commercially launched in 2019 after four years of research by Stanford AI Lab co-founder and CEO Alex Ratner and his team.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
