Snorkel AI announced on September 22 a $350 million Series E round at a $3.5 billion valuation, capping a year in which the company's data-as-a-service business grew more than eighteenfold and crossed a $375 million annualized revenue run rate. The round was disclosed in a blog post by co-founder and CEO Alex Ratner and reported by Unite.AI.
The raise is one of the largest data-infrastructure financings of the year and signals how quickly the market for specialized AI training data has matured from an academic side project into a revenue engine measured in the hundreds of millions. For more context on this story, see our ongoing latest AI developments.
Who Invested in the Snorkel AI Series E
Insight Partners and S32 led the round. New participants include Third Point, March, Blumberg, Allegis, Standard VC, and Frontline. A deep bench of existing investors returned as well: Addition, Lightspeed, Greylock, GV, P7, Wells Fargo, Walden Catalyst Ventures, and Factory.
The investor list spans growth equity, crossover funds, and strategic backers — a mix that reflects Snorkel's positioning between academic machine learning research and enterprise AI deployment.
The Numbers Behind the Round
Ratner said the company's data-as-a-service offering, launched nearly a year earlier, grew over 18x and crossed $375 million in annualized revenue run rate in the week of the announcement. He also said Snorkel now partners with frontier labs, hyperscalers, neolabs, vertical AI leaders, enterprises, and U.S. government agencies.
That customer roster explains the valuation jump. Snorkel was valued at $1.3 billion in May 2025; eighteen months later, its new price tag is roughly 2.7x higher, backed by a revenue figure that would have been unthinkable for a data-labeling business in the pre-LLM era.
From Research Project to Revenue Engine
Snorkel was founded in 2019 and is headquartered in Redwood City, California, emerging from years of academic research into programmatic data labeling. Its previous $100 million Series D, announced May 29, 2025, was led by Addition with participation from Prosperity 7 Ventures, Greylock, Lightspeed, and strategic investors including BNY and QBE Ventures. That round brought total funding to $237 million since the company's founding and coincided with the general availability of Snorkel Evaluate and Snorkel Expert Data-as-a-Service, two offerings on the company's AI Data Development Platform for evaluating and tuning specialized AI systems.
"Data 2.0" and the Agentic Data Platform
Ratner frames the new financing around what he calls Data 2.0 — the shift from assembling static training sets to running continuous, expert-guided data pipelines for systems that must be evaluated and tuned as their tasks grow more agentic. The framing matters because the bottleneck in frontier AI development has visibly moved: as compute and architectures commoditize, the differentiated input is high-quality, task-specific human and expert data, delivered with evaluation loops attached.
Snorkel's growth figures suggest the market agrees. An 18x expansion in roughly a year, attached to a platform that counts frontier labs and U.S. government agencies among its partners, is the kind of traction investors have been pricing into the broader AI data supply chain all year.
What Expert Data-as-a-Service Actually Sells
Snorkel's platform addresses a problem that has shifted up the AI stack. In the early era of large language models, scale came from scraping the open web. In 2026, the marginal gains come from expert knowledge: physicians reviewing clinical summaries, lawyers annotating case reasoning, engineers writing and judging code with domain nuance. That labor is expensive, slow, and hard to manage — which is exactly the gap Snorkel's Expert Data-as-a-Service and Snorkel Evaluate products were built to fill.
The company's offering on its AI Data Development Platform centers on evaluating and tuning specialized AI systems: enterprises bring a model and a task, and Snorkel supplies the expert reviewers, programmatic labeling workflows, and measurement loops needed to know whether the model is actually improving. The government agency partnerships matter here too — public-sector AI procurement demands documentation, auditability, and quality control that ad hoc labeling vendors have historically struggled to provide.
What the $350 Million Will Fund
The company positions itself as building the "frontier lab for AI data," and the new capital gives it runway to expand engineering, research, and go-to-market efforts across that platform. Expect competition to intensify: the specialized data layer has become one of the most contested parts of the AI stack, with hyperscalers, model labs, and independent vendors all racing to own the expert-feedback loops that frontier training runs increasingly depend on.
The Bigger Picture: Venture Capital's AI Concentration
The Series E lands against a backdrop in which the overwhelming majority of venture dollars are flowing to AI companies, with a small set of model builders and their suppliers absorbing most of it. Data infrastructure sits squarely inside that privileged circle — every frontier model release, enterprise deployment, and agentic product launch consumes more of exactly what Snorkel sells.
For Snorkel, the challenge over the next eighteen months will be converting a skyrocketing run rate into durable, diversified revenue before the market for expert data consolidates or its largest customers — the frontier labs themselves — vertically deeper into in-house data operations. For the industry, the round is one more confirmation that in 2026, the money is following the data.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
