Training data provider Snorkel AI raises $350M at $3.5B valuation

Artificial intelligence training data provider Snorkel AI has raised $350 million in funding, bringing its valuation to $3.5 billion. Insight Partners and S32 led the Series E deal, joined by Alphabet Inc. 's GV startup fund and over half a dozen other investors. Founded in 2019 by researchers from Stanford AI Lab, Snorkel AI's first product was Snorkel Flow, a software platform that automated the process of creating supervised learning datasets.

Supervised learning involves training neural networks using labeled datasets comprised of prompts and correct answers. Snorkel Flow reduced the time-consuming task of creating these datasets using statistical methods developed by the company's Stanford founders, addressing accuracy issues of earlier automation approaches. In 2022, Snorkel AI shifted its focus from selling software to providing ready-to-use training datasets and expanded its offerings to include reinforcement learning, a more complex AI training approach.

Reinforcement learning datasets contain unanswered questions, requiring AI models to figure out answers without human assistance. The company relies on tens of thousands of human experts to generate reinforcement learning training tasks and provides customers with technical assets for AI training runs. Snorkel AI develops AI evaluation rubrics and improves them based on human reviewer feedback.

In 2023, the company reported an 18-fold growth in revenue, crossing an annualized run rate of $375 million. The newly raised funds will be used to hire more engineers, invest in AI safety initiatives, and support the development of open-source model evaluation benchmarks.