The center of gravity in AI training data is shifting. It used to be simple labeling — tagging images, classifying text — a volume problem you solved with headcount. What frontier and agentic systems need now is different: expert-level agentic tasks, environments, and rubrics that take even the most qualified humans hours or days to build. Designing them well is research work in itself, where quality and complexity determine value.

Snorkel AI sits at the front of that shift. Spun out of the Stanford AI Lab in 2019, the company was built by the team that first introduced weak supervision — a technique for refining data at scale without having humans label every example by hand — to the field. The wider academic community that grew out of that work has now published more than 250 papers, cited over 25,000 times.
Growth has accelerated sharply since Snorkel launched its expert “data-as-a-service” offering in September 2025 — supplying finished datasets and evaluation data directly to frontier AI labs and enterprises, rather than just selling labeling software. The company says its annualized revenue run rate is up eighteenfold over the past year, to $375 million. That figure isn’t directly comparable to headline numbers from rivals like Mercor or Handshake, though: those companies run marketplaces that pay out 60% to 70% of gross revenue straight to the domain experts doing the work, so their published figures sit closer to gross transaction volume. Snorkel sells finished datasets and RL environments instead, so expert payments show up as cost of goods sold rather than revenue.
CEO Alex Ratner started and led the open-source Snorkel project as a PhD student at Stanford, advised by Christopher Ré — more than 60 papers came out of that project before it was incorporated as a company in 2019. “The teams pushing the frontier want a research data partner who pioneers the science of data development,” Ratner said. “That’s what Snorkel was built to be: the frontier lab for agentic data, combining human excellence with over a decade of research and technology.”
Snorkel said it has raised $350 million in a Series E co-led by Insight Partners and S32, with significant participation from existing investor Addition. New investors March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard, and Third Point Ventures joined, alongside existing investors Greylock, Lightspeed, GV, Factory, Prosperity7, Walden Catalyst, and Wells Fargo. The round values the company at $3.5 billion — nearly triple the $1.3 billion valuation it held after its Series D 17 months ago.
Lonne Jaffe, Managing Director at Insight Partners, said: “Snorkel’s research-grade approach to AI data, environments, and measurement is becoming an increasingly important ingredient in building capable and reliable AI systems — and their demonstrated growth at scale reflects that. We’re excited to back Alex and the team as they continue to bring their frontier AI data factory capabilities to longer-running, complex work in high value industries like healthcare, law, and software engineering.” Andy Harrison, CEO and General Partner of S32, said Snorkel’s “expert-agentic environments create a unique flywheel between human expertise and AI, delivering data at the speed and quality that enables a new era of efficiency and accuracy for model development.”
The funding will grow the capacity of Snorkel’s agentic data factory, accelerate investment in vertical and enterprise AI, and extend its research and technology into new domains and modalities, including deeper investment in open research programs like its Open Benchmarks Grants initiative.
The competitive landscape looks nothing like it did a year ago. Scale AI, the category’s original pioneer, was upended in June 2025 when Meta invested $14.3 billion for a 49% stake and hired away founder Alexandr Wang to its own AI lab — pushing core Scale customers like OpenAI and Google to seek “neutral” data partners instead. Mercor, founded by three 22-year-olds in 2023, moved fastest to fill that gap and is now a $10 billion decacorn with gross annualized revenue reportedly near $2 billion — though, like Snorkel noted about its rivals, that figure reflects marketplace throughput rather than a directly comparable revenue base. Surge AI, less flashy but reportedly the segment’s actual revenue leader, crossed $1 billion in revenue in 2024 and roughly $1.4 billion in 2025 without ever raising outside capital. Handshake, the college-recruiting platform turned AI data supplier via its network of student experts, has also entered the market and reportedly crossed $1.1 billion in gross annualized revenue.
What separates Snorkel from that pack is its research pedigree: rather than running a pure marketplace of human experts, the company says it uses its own software and models to generate data synthetically alongside subject-matter experts, a hybrid it’s been refining since its Stanford roots. For the broader AI training-data market, see wowtale’s data-labeling landscape. Wowtale previously covered Snorkel’s $100 million Series D and unicorn debut in June 2025.
MORE FROM THE POST
- Generative Media Inference Platform fal Raises $1.3B Series F
- AI Training Data & Agents Firm Labelbox Raises $100M+ Series E
- AI Training-Data Startup Micro1 Reportedly Hits $4B Valuation
- Firecrawl Raises $75M Series B, Launches Alexandria
- Factory Raises $200M at $5B Valuation, Tripling in Five Months
Share
Most Read
- 1
- 2
- 3
- 4
- 5


Leave a Reply