Snorkel AI triples valuation to $3.5B as demand for AI training data booms

Snorkel AI has landed a $350 million Series E and says the round values it at $3.5 billion. Insight Partners and S32 led the deal, and backers including Addition, Lightspeed, Greylock, GV, and Wells Fargo joined in again.

That’s a sharp jump from where it stood 17 months ago. In its Series D, Snorkel raised $100 million at a $1.3 billion valuation. Now the seven-year-old startup says it’s worth nearly three times as much.

The company started out with software for automating data labeling. Last year, it moved into something broader: complete data sets, plus simulated environments for AI training. Snorkel calls that data-as-a-service, and it says the model blends its software, synthetic generation, and subject matter experts instead of just running a human expert marketplace.

The growth story is hiding in the numbers. Snorkel says its annualized revenue run-rate is now $375 million, up 18-fold over the last 12 months. It also says payments to human experts sit in cost of goods sold, since it sells reinforcement learning environments and finished datasets rather than human labor.

This is the kind of business AI labs are flooding money into right now. Snorkel’s rise comes alongside similar spikes at other data companies, including Mercor, Handshake, and Micro1. The real takeaway is simple: the bottleneck in AI is no longer just chips or models. It’s getting high-end training data, and that’s suddenly very expensive.