Snorkel AI raises $350 million in funding led by Insight Partners and S32

Snorkel AI, a company specializing in training data for AI development, announced on September 22, 2026, that it has raised $350 million in a funding round co-led by Insight Partners and S32. The company's valuation is reported to be $3.5 billion, with the funds intended for the expansion of its "data factory" aimed at frontier AI applications.

In addition to the lead investors, new participants in this funding round include March Capital, Blumberg Capital, Allegis Capital, Frontline, Standard, and Third Point Ventures. Existing investors such as Addition, Greylock, Lightspeed, GV, Factory, Prosperity7, Walden Catalyst, and Wells Fargo also contributed to the round.

The company emphasizes that good data is not something to be collected but created. Snorkel AI provides a service that fully manages the creation of data, rather than merely selling tools for customers to use. This involves experts being engaged to construct datasets, benchmarks, evaluations, and the environments themselves, which are then delivered to research teams. They offer non-exclusive datasets and environments known as the "Snorkel Data Series," which are updated quarterly.

Snorkel AI's co-founder and CEO describes the tasks, environments, and grading criteria designed by experts as "Data 2.0," highlighting that simply increasing personnel will not solve the research challenges involved. The approach combines domain expertise with task design, automated checks, calibrated expert reviews, and evaluation environments to transform raw data into measurable signals usable for learning.

For example, in environments referred to as "Enterprise environments," the company simulates actual business workflows with high fidelity to train AI agents. In a recreated environment for insurance underwriting, the company reported an improvement in the correct decision-making rate (pass@1) from 10.9% to 42.0% based on their measurements. This iterative process of evaluating failures and adjusting data and environments accordingly is a key aspect of their operational model.

Reported by BRIDGE.