September 22, 2026, (Inside AI) — Snorkel AI, a San Francisco startup that builds training data for artificial intelligence models, has raised $350 million in a funding round that values the company at $3.5 billion, CEO Alex Ratner confirmed. The round was led by Insight Partners and S32, with participation from existing backers Addition, Greylock, and Wells Fargo.
The valuation marks a near tripling from Snorkel's previous round in May 2025, when it raised $100 million at a $1.3 billion valuation. The company also disclosed that its annualized revenue run-rate has crossed $350 million, up from roughly $20 million a year earlier. That growth is tied to its data-as-a-service business, launched in September 2025.
Founded in 2019 by researchers from the Stanford AI Lab, Snorkel initially sold software tools for data labeling. It has since shifted to supplying finished datasets and reinforcement-learning environments directly to frontier AI labs, hyperscalers, enterprises, and the US federal government.
The funding surge reflects a broader shift in how AI developers acquire training data. Simple labeling tasks are no longer enough. Labs now need specialized, high-stakes data to train and evaluate models that handle coding, law, medicine, and other complex domains.
Read: AI startup Mantic raises $25 million for superhuman forecasting
Snorkel's platform pairs human experts with thousands of specialized AI models and agents. Experts design scenarios, tasks, and grading rubrics, while AI automates much of the quality assurance work. The company taps a network of tens of thousands of specialists across fields such as coding, law, and medicine. Ratner said Snorkel sells the resulting data products rather than charging for human labor, a model that lets it pay experts more while preserving margins.
"Our strong view is that 100% of the data that labs will get value out of will have some human input in the foreseeable future," Ratner said. "But 100% of that data will have to use synthetic and automated approaches to keep up with this complexity."
The market for human-annotated training data has transformed since Meta purchased a 49% stake in Scale AI for $14.3 billion in June 2025. That deal validated the sector and spurred investor interest in rivals such as Mercor and Surge AI, which have also reported strong revenue growth.
Andy Harrison, a partner at S32 who co-led the funding, framed the investment as a bet on scarcity. "Data is becoming more rare, more specialized, more difficult to find," Harrison said. "If you want to train the most frontier, complex and capable models, now you need superior data."
Snorkel plans to use the fresh capital to hire researchers and engineers, expand its enterprise and government operations, support third-party AI model evaluations, and push into new industry verticals and data modalities. The company said it expects to reach profitability this year, even as it prioritizes growth.
Coding data remains one of Snorkel's largest areas of demand. The company's agentic data development platform is designed to handle the complexity that comes with training models for software engineering, legal analysis, and medical reasoning. Those domains require nuanced judgment that simple crowd-sourced labeling cannot provide.
Read: OpenAI Seeks $1.5 Trillion Valuation in New Funding Round
The funding round arrives as venture capital continues to flow into startups that supply frontier labs with training data. The Scale AI deal set a benchmark, and competitors have raced to capture similar demand. Mercor and Surge AI have both attracted investor attention, though neither has disclosed a valuation matching Snorkel's new mark.
Snorkel's model of selling data products rather than labor hours differentiates it from traditional annotation firms. By automating quality assurance with AI agents, the company can scale output without proportional increases in human headcount. That structure supports higher pay for experts and healthier margins for the business.
The company's work with the US federal government adds a layer of stability that pure commercial contracts may not provide. Government agencies are increasingly seeking AI systems that can handle sensitive data and complex regulatory requirements. Snorkel's ability to serve that market could open doors to long-term contracts.
Ratner's comments suggest that the human role in AI training will not disappear. Instead, it will evolve. Experts will define the problems and judge the outputs, while AI handles the repetitive work. That hybrid approach may become the standard for frontier model development.
The $3.5 billion valuation places Snorkel among the most valuable private companies in the AI data space. Whether it can sustain its revenue trajectory will depend on continued demand from labs racing to build more capable models. For now, the company's growth story has convinced major investors to double down.
