Snorkel AI triples valuation to $3.5B on AI training-data demand
Snorkel AI, a seven-year-old startup that builds training datasets and simulated environments for AI labs and corporations, has closed a $350 million Series E at a $3.5 billion valuation. The round was led by Insight Partners and S32, with existing investors Addition, Lightspeed, Greylock, GV, and Wells Fargo also participating. The new valuation is nearly triple the $1.3 billion mark Snorkel reached in its Series D, a $100 million round closed 17 months earlier.
Snorkel originally sold software for automating data labeling. Last year it shifted to a data-as-a-service model, delivering customers finished datasets rather than just the tools to build them. Its approach is hybrid: the company uses its own software and models to generate data synthetically, working alongside human subject matter experts rather than running a pure human-expert marketplace.
The company says its annualized revenue run rate now stands at $375 million, an eighteenfold increase over the past 12 months, which it attributes to surging demand from AI labs for high-end training data. Other companies positioning themselves as AI data labs report similar growth: Mercor's gross annualized revenue has reached $2 billion, Handshake passed the $1 billion mark earlier this year, and Micro1 has scaled to $500 million, according to prior TechCrunch reporting. Because these companies typically pay out 60% to 70% of that top-line income directly to the domain specialists doing the work, their actual net revenue runs well below the headline gross figures. Snorkel says its own case differs on the accounting: since it sells RL environments and complete datasets rather than human labor directly, payments to its human experts are booked as cost of goods sold rather than folded into its headline revenue number.
Snorkel launched commercially in 2019, following four years of research by co-founder and CEO Alex Ratner and his team at a Stanford AI lab.
Key facts
- Snorkel AI raised a $350 million Series E at a $3.5 billion valuation, led by Insight Partners and S32.
- The new valuation is nearly triple the $1.3 billion valuation from its $100 million Series D, closed 17 months earlier.
- Snorkel's annualized revenue run rate now stands at $375 million, an eighteenfold increase over the last 12 months.
- Snorkel shifted last year from data-labeling software to a data-as-a-service model, generating data with a hybrid of its own models and human subject matter experts.
- Peers Mercor ($2 billion gross ARR), Handshake ($1 billion) and Micro1 ($500 million) show similar growth, but typically pay 60% to 70% of that revenue to domain specialists, unlike Snorkel, which books those payments as cost of goods sold.
Why it matters
A nearly threefold valuation jump in under a year and a half, alongside an eighteenfold revenue increase, signals how acute the demand for high-quality AI training data has become as labs push to improve their models. Snorkel's pivot from selling labeling software to selling finished datasets and RL environments mirrors a broader shift in the sector: from tools that help humans produce data to companies that sell the data itself.
Who it affects
AI labs and corporations that need large volumes of specialized training data and reinforcement-learning environments are Snorkel's direct customers. The subject matter experts who help produce that data, Snorkel's investors (Insight Partners, S32, Addition, Lightspeed, Greylock, GV, Wells Fargo), and rival data companies such as Mercor, Handshake, and Micro1, which are seeing comparable growth in the same market, are all affected by how this segment is now being valued.
How to use it
Snorkel sells directly to AI labs and enterprises as a data-as-a-service provider, delivering completed training datasets and simulated RL environments rather than just labeling tools. The article gives no pricing or contract terms.
How solid is it
The figures, the $350 million raise, the $3.5 billion valuation, the prior $100 million Series D at $1.3 billion, and the $375 million revenue run rate, are Snorkel's own disclosures as reported by TechCrunch; none are independently audited in the source. The comparisons to Mercor, Handshake, and Micro1 rely on TechCrunch's own prior reporting on those companies.
Risks and caveats
The eighteenfold revenue jump is a run-rate figure, not audited annual revenue, and run rates can move quickly in either direction. Cross-company revenue comparisons in this sector are complicated by differing accounting: peers that pay 60% to 70% of gross revenue to domain specialists see much lower net revenue than their headline numbers suggest, while Snorkel says it books similar expert payments as cost of goods sold rather than in revenue, a difference that makes its $375 million figure not directly comparable to competitors' gross figures without knowing more about each company's cost structure.