Snorkel AI triples its value to 3.5 billion dollars

Snorkel AI, a company that supplies the data used to train artificial intelligence models, has raised 350 million dollars in new funding at a valuation of 3.5 billion dollars, nearly three times what it was worth last year.

The Series E funding round, announced on 22 September, was led by the investment firms Insight and S32. Other investors included Third Point, March, Blumberg, Allegis, Standard VC and Frontline, alongside existing backers including Lightspeed, Greylock, GV and Wells Fargo.

Selling data

Snorkel originally sold software to help companies label data automatically. Last year it changed direction and began supplying customers with finished datasets, a service it calls data-as-a-service.

The company said that the data service, launched nearly a year ago, had grown more than 18-fold and passed 375 million dollars in annualised revenue in the week of the announcement.

Why data matters

Modern AI models learn from huge amounts of data. As models become more capable, the companies building them increasingly want high-quality, specialised data created or checked by experts, for example in medicine, law, finance or coding, to teach their systems to handle difficult tasks.

Supplying that data has become a fast-growing business, with several companies competing to provide it to the major AI developers.

A soaring valuation

The new valuation is nearly three times the 1.3 billion dollars at which Snorkel was valued when it raised 100 million dollars in May 2025. It reflects how much investors are willing to pay for companies that supply the building blocks of AI.

A booming market

Investment in AI companies has continued at a rapid pace in 2026, with large sums going to model developers, chip makers and the companies that supply data and infrastructure. Snorkel said it would use the money to scale up what it describes as a data factory for frontier AI.

Why training data matters

AI models learn from examples, and the quality of those examples has a big effect on how well a model performs. As companies try to build AI for specialist tasks in areas such as law, finance and medicine, demand has grown for carefully checked data created or labelled by experts.