Snorkel AI raised $350 million in a Series E that values the AI data company at $3.5 billion, nearly triple its valuation from 17 months ago, according to TechCrunch.
Insight Partners and S32 led the round, with additional participation from Third Point, Addition, Lightspeed, Greylock, GV and Wells Fargo, TechCrunch and Unite.AI reported. The prior round, a $100 million Series D, had valued the company at $1.3 billion.
Snorkel, founded seven years ago out of Stanford research by CEO Alex Ratner, has shifted from selling data-labeling software to selling finished training datasets, combining synthetic data generation with human subject matter experts under what it calls data-as-a-service, according to TechCrunch. That business has grown 18-fold over the past year and crossed a $375 million annualized revenue run rate, TechCrunch and Unite.AI reported.
Snorkel's jump tracks a broader run-up across companies that supply AI labs with training data. TechCrunch reported that Mercor has reached $2 billion in annualized revenue, Handshake has hit $1 billion and Micro1 has scaled to $500 million.
The money chasing data suppliers says something about where AI labs think their next gains come from. If frontier labs are paying this much for curated datasets and expert-built training data, model quality looks bottlenecked less by compute and more by what the model is fed.