Snorkel AI Raises $350M Series E, Valuation Triples to $3.5 Billion
News Summary
Snorkel AI, a startup that builds curated training data for artificial intelligence systems, announced on September 22, 2026 (Eastern Time) that it has raised $350 million in a Series E funding round at a $3.5 billion valuation — nearly tripling the $1.3 billion valuation it held just 17 months earlier. The round underscores how quickly demand for high-quality, expert-curated training data has grown as AI labs push toward more capable and specialized models.
The Funding Round
The Series E was led by Insight Partners and S32, the venture arm connected to Alphabet's growth ecosystem, with additional participation from Third Point, March, Blumberg, Allegis, Standard VC, and Frontline. Existing backers also returned for the round, including Addition, Lightspeed, Greylock, GV, P7, Wells Fargo, Walden Catalyst Ventures, and Factory. The jump from a $1.3 billion valuation at the company's $100 million Series D just over a year ago to $3.5 billion today reflects a rapid re-rating of the AI data infrastructure sector as a whole.
From Data Labeling Software to Data-as-a-Service
Founded seven years ago by CEO Alex Ratner and a team that grew out of Stanford University's AI lab, Snorkel launched commercially in 2019 as a software platform for automating data labeling — the process of tagging raw data so machine learning models can learn from it. Roughly a year ago, the company pivoted to a "data-as-a-service" model, shifting from selling labeling tools to directly producing finished, high-quality datasets and simulated environments that AI labs and enterprises can plug straight into their model training pipelines.
That pivot appears to have paid off. Snorkel says the data-as-a-service business has grown 18-fold over the past twelve months and crossed a $375 million annualized revenue run rate in the week of the funding announcement.
How the Agentic Data Platform Works
At the center of Snorkel's pitch is what the company calls its Agentic Data Platform, which pairs human subject-matter experts with specialized AI agents rather than relying on either humans or general-purpose language models alone. According to Ratner, the agents can take an expert's initial "sketch" of a task and expand it into a full environment or complete set of data instances, while also handling quality control and routing work to the right specialist.
Snorkel reports that these specialized agents speed up quality-control work by more than 50% and lift review accuracy by more than 15 points compared with human reviewers working with standard large language models. The company also claims its system delivers more than twice the accuracy of non-specialized frontier LLM baselines on data-quality tasks. Ratner has described the approach as a "recursive self-improvement" loop, in which human feedback continuously retrains and sharpens the AI agents that assist with data creation.
Notably, Snorkel accounts for payments made to its network of human experts as cost of goods sold rather than folding them into headline revenue figures, a structuring choice that distinguishes its reported financials from some peers in the space.
Customers and Market Position
Snorkel says its customer base now spans frontier AI labs, hyperscale cloud providers, so-called "neolabs," vertical AI companies, large enterprises, and U.S. government agencies. The breadth of that customer list illustrates how the need for curated, domain-specific training data has spread well beyond the handful of companies building the largest general-purpose models.
The funding round also arrives amid a broader boom in AI training-data businesses. Other companies in the space have reported similarly explosive growth: data-labeling and expert-network company Mercor has reported reaching $2 billion in gross annualized revenue, while workforce data platform Handshake has said it surpassed $1 billion in the category, according to industry reporting. Investors appear to be betting that as AI labs exhaust easily available public data, demand for expert-generated, verified, and synthetic training data will keep expanding rapidly.
Use of Funds
Snorkel says it plans to use the new capital to expand its engineering, research, and go-to-market teams, grow its Open Benchmarks Grants program — a roughly $3 million commitment supporting the creation of open evaluation datasets for the AI research community — and invest further in specialized AI models. The company also says a portion of its research effort will focus specifically on data development for AI alignment and safety, an area that has drawn increasing attention as models are deployed in higher-stakes settings.