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

Snorkel AI has raised a $350 million Series E at a $3.5 billion valuation, nearly tripling its worth in 17 months, as AI labs' demand for high-quality training data continues to outstrip supply.

Insight Partners and S32 led the round. Existing investors Addition, Lightspeed, Greylock, GV, and Wells Fargo also participated. The new valuation towers over the $1.3 billion the company commanded when it raised $100 million in a Series D a year and a half ago.

From labeling tools to data-as-a-service

Snorkel, now seven years old, launched commercially in 2019 after four years of research by co-founder and CEO Alex Ratner and his team at a Stanford AI lab. The company originally sold software for automating data labeling. Last year, it changed course and began delivering completed datasets to customers, an offering it calls data-as-a-service.

The company also builds simulated environments for AI training. Unlike pure human-expert marketplaces, Snorkel uses a hybrid approach: its software and models generate data synthetically, with subject matter experts in the loop.

That model is scaling fast. Snorkel says its annualized revenue run rate now stands at $375 million, an eighteenfold increase over the last 12 months. The company attributes the growth to AI labs' insatiable appetite for high-end training data.

A booming, but complicated, market

Snorkel is not the only data company riding the wave. 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.

Headline gross figures understate the economics of these businesses, though. Such companies pay out roughly 60% to 70% of top-line income directly to the domain specialists doing the work, meaning their net annual revenue falls substantially below the gross numbers.

Snorkel claims a cleaner accounting story. Because it sells reinforcement learning environments and complete datasets rather than human labor, payments to its human experts land in cost of goods sold rather than in the headline annualized revenue figure, according to the company.

Why it matters

The round signals where the money is flowing in the AI stack right now. As frontier labs exhaust easily scraped internet data, the bottleneck has shifted to expert-generated and synthetic training data, and investors are paying premium multiples for companies that can supply it. A near-triple valuation jump in 17 months, against an 18x revenue ramp, shows how sharply the market has repriced data infrastructure.

The test ahead is whether data-as-a-service revenue proves durable as labs build more of their own synthetic data pipelines in-house, or whether Snorkel's hybrid model keeps it a step ahead of both the marketplaces and its customers.