Snorkel AI just raised $350 million at a $3.5 billion valuation, the company said in a September 22 press release. Insight Partners and S32 co-led the round. That price is nearly three times the $1.3 billion valuation Snorkel carried after a $100 million raise in May 2025, according to Reuters. Snorkel is private, so there’s no stock to trade here. What the round does offer is a clear read on where AI money is going: toward training data that is slow, expensive, and hard to make. One caution up front. The revenue growth behind this valuation is self-reported, and the press release doesn’t include it at all.

Who’s Writing the Checks

The round drew significant participation from existing investor Addition. New investors include March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard, and Third Point Ventures. Returning backers include Greylock, Lightspeed, GV, Factory, Prosperity7, Walden Catalyst, and Wells Fargo. GV is Alphabet’s venture capital arm.

From Labeling to “Data 2.0”

For years, AI training data meant labeling. People tagged photos, flagged spam, rated chatbot answers. It was a volume game. More workers, more labels.

Snorkel’s release calls that era “Data 1.0.” Its pitch is that today’s most advanced models need something harder, which it brands “Data 2.0”: expert agentic tasks, environments, and rubrics. Those terms need unpacking.

An agentic task is a multistep job an AI agent has to finish on its own, like tracking down a bug buried in a large codebase. An environment is a simulated workspace, such as a sandboxed computer terminal, where a model can take actions and get scored. Labs use these for reinforcement learning, or RL, a training method where a model improves through trial, error, and reward. A rubric is the grading guide that spells out what a good result looks like.

Snorkel says building these well can take qualified experts hours or days. The branding is Snorkel’s. But the shift underneath it is showing up across the industry.

The Numbers Behind the Price

The press release skips revenue entirely. CEO Alex Ratner shared figures with reporters instead.

He told Reuters that Snorkel’s annualized revenue run rate had topped $350 million, up from about $20 million a year earlier. TechCrunch was given a figure of $375 million, described as an 18-fold jump in 12 months. A run rate takes the current pace of sales and projects it across a full year. It isn’t the same as booked annual revenue. And as a private company, Snorkel doesn’t publish audited results.

Most of that growth, the company says, came from the expert data-as-a-service business it launched in September 2025. Rather than selling labeling software, Snorkel now delivers finished datasets and RL environments. Coding data is one of its biggest areas of demand, per Reuters.

There’s an accounting wrinkle worth knowing. TechCrunch notes that marketplace-style rivals typically pass 60% to 70% of their gross revenue straight to the specialists doing the work. Snorkel told the outlet that because it sells finished products rather than expert labor, it books payments to its experts as a cost of goods sold. The company presents that as a real difference from its rivals. Outsiders can’t compare the figures directly without seeing its books.

A Crowded, Well-Funded Field

Snorkel isn’t alone here. The category changed when Meta paid $14.3 billion for a 49% stake in Scale AI in June 2025, Reuters noted. Money has kept flowing since. TechCrunch reports that Mercor’s gross annualized revenue has reached $2 billion, Handshake crossed $1 billion earlier this year, and Micro1 hit $500 million. Surge AI is also drawing investor interest.

Snorkel’s angle is research. It spun out of the Stanford AI Lab, and the company says its team’s work spans more than 250 peer-reviewed papers cited over 25,000 times. That’s Snorkel’s own tally. According to TechCrunch, it also pairs human experts with its own software and models that generate synthetic data, instead of operating purely as a talent marketplace.

What the Company and Its Backers Are Saying

Ratner described Snorkel as “the frontier lab for agentic data, combining human excellence with over a decade of research and technology.”

The other two quotes in the release come from the round’s co-leads. Insight Partners Managing Director Lonne Jaffe pointed to Snorkel’s “demonstrated growth at scale” and its plans to push into healthcare, law, and software engineering. S32 CEO Andy Harrison went further, saying that working with Snorkel “ensures the highest frontier model success rate.” Both firms just put money into the company. The release offers no data to support Harrison’s claim, so read it as a backer’s view, not a measured result.

Where the Money Goes

Snorkel plans to expand capacity at what it calls its agentic data factory, invest more in vertical and enterprise AI, and extend its research into new domains and modalities, meaning new types of data. It also plans to deepen its spending on open research, including Open Benchmarks Grants. That’s a $3 million program, launched in February 2026, that funds public datasets and tests for measuring AI performance.

All of this is a plan, not a result.

The Bottom Line

The valuation tells you investors expect frontier labs to keep paying up for expert-built data. That’s the takeaway. The caveats matter too. The revenue figures are self-reported and unaudited. The release names no customers, so there’s no way to see how much of that revenue depends on a handful of large AI labs. And the field is filling up fast, with rivals reporting bigger top-line numbers of their own.

Sources

• Snorkel AI Raises $350M to Scale the Data Factory for Frontier AI, PRNewswire, September 22, 2026.
• Exclusive-Snorkel AI valued at $3.5 billion amid surging demand for complex AI training data, Reuters (Krystal Hu), via Yahoo Finance, September 22, 2026.
• Snorkel AI triples valuation to $3.5B as demand for AI training data booms, TechCrunch (Marina Temkin), September 22, 2026.
• Snorkel AI Highlights First Wave of Open Benchmarks Grants Projects, PRNewswire, July 24, 2026.

Editorial Disclosure

This article is based on a press release issued by Snorkel AI on September 22, 2026, distributed via PRNewswire. Securities discussed: None. Snorkel AI is a privately held company and its shares are not publicly traded. Next Gen Tech Stocks has not received compensation from Snorkel AI, its management, its investors, investor relations representatives, or any third party for this coverage. No staff member or principal of Next Gen Tech Stocks holds a position in Snorkel AI at the time of publication. Revenue run-rate and growth figures cited in this article were reported by Snorkel AI to third-party media, are not contained in the press release, and have not been independently audited or verified. Statements regarding Snorkel AI’s planned use of proceeds, capacity expansion, research programs, and market opportunity are forward-looking and involve risks and uncertainties; actual results may differ materially. References to companies are for market context and analytical purposes only and do not constitute an investment recommendation.

Coverage on Next Gen Tech Stocks is for informational and educational purposes only and does not constitute professional financial advice. For further details on our publishing standards, advertising relationships, and risk disclaimers, please see our full Terms & Disclaimers Page