Snorkel AI Reaches $3.5B Valuation in Series E

Why this matters

The jump from $1.3 billion to $3.5 billion in 17 months tells us where the money in AI infrastructure is actually flowing right now, and it's not toward another foundation model. It's toward the unglamorous problem of getting labs enough good data to train on. Snorkel's pivot from selling labeling software to selling finished datasets is worth watching closely: it suggests that even sophisticated AI buyers would rather pay for an outcome than manage the pipeline themselves.

For founders, that's a signal about where margin sits in this stack, closer to the data layer than many assumed a year ago. For researchers, the hybrid model, human experts plus software rather than a pure marketplace, is a quiet admission that full automation of data curation isn't here yet, despite years of hype about synthetic data replacing human annotation. Insight Partners and S32 leading a round this size, with Wells Fargo staying in as a repeat investor, also says something about who believes data-as-a-service has staying power beyond the current training boom.

Worth tracking whether Snorkel's simulated-environment work becomes the bigger business than datasets themselves.