Snorkel AI Triples Valuation to $3.5B on AI Data Demand

Snorkel AI has raised $350 million in a Series E round at a $3.5 billion valuation, nearly tripling its worth in 17 months as demand for high-end AI training data reshapes the market for dataset providers.

Insight Partners and S32 led the round. Existing investors Addition, Lightspeed, Greylock, GV, and Wells Fargo also participated, according to details of the deal.

The new valuation stands in sharp contrast to the $1.3 billion the seven-year-old startup commanded when it raised $100 million in its Series D roughly a year and a half ago. In that interval, Snorkel's business has transformed — and so has its revenue line.

The company says its annualized revenue run rate now stands at $375 million. That figure represents an eighteenfold increase over the last 12 months. The growth is fueled, in Snorkel's words, by AI labs' "insatiable appetite for high-end training data."

A business model rebuilt in a year

Snorkel launched commercially in 2019, after four years of research by co-founder and CEO Alex Ratner and his team at a Stanford AI lab. Its original product was software for data-labeling automation.

Last year, the company shifted its offering. Instead of selling tools, Snorkel began delivering completed datasets to customers — an approach it calls data-as-a-service. The company also builds simulated environments for training AI systems.

The delivery model is hybrid. Snorkel uses its software and models to generate data synthetically, working alongside subject matter experts rather than operating purely as a human expert marketplace.

That structure has accounting consequences that matter for anyone comparing revenue claims across the sector. Because Snorkel sells reinforcement learning (RL) environments and complete datasets rather than human labor, payments to its human experts are accounted for in cost of goods sold rather than in headline annualized revenue numbers, according to the company.

A crowded, fast-scaling field

Other companies positioning themselves as AI data labs report similarly steep growth curves. Mercor's gross annualized revenue has climbed to $2 billion. Handshake hit the $1 billion milestone earlier this year. Micro1 has scaled to $500 million, TechCrunch reported.

Those headline numbers deserve scrutiny. Such companies pay out roughly 60% to 70% of their top-line income directly to the domain specialists doing the work, which means their actual net annual revenue is substantially lower than the gross figures they publicize.

Snorkel's cost structure sidesteps some of that optics problem. By folding expert payments into cost of goods sold, the company presents a run rate that reflects its product revenue — datasets and RL environments — rather than a pass-through figure inflated by labor payouts.

Why the money is moving

The through-line across these deals is the same: AI labs need training data that is increasingly specialized, and they are willing to pay at scale to get it. Snorkel's eighteenfold run-rate growth in twelve months, and a valuation that jumped from $1.3 billion to $3.5 billion between its Series D and Series E, measure how quickly that demand has converted into contracts.

The question investors are now pricing is durability. Snorkel has $350 million in fresh capital and a business model that has already pivoted once — from labeling software to finished data products — in response to where its customers' needs moved. Whether synthetic generation plus expert oversight can sustain margins as labs' appetites evolve will determine whether this round marks a ceiling or a waypoint for the company.