Why We Invested in Snorkel AI: $350M Series E at a $3.5B Valuation

Snorkel AI is building the agentic data factory for the increasingly sophisticated data and environments behind frontier AI.

By Jimmy Zhu and Jared Katzen

AI models are becoming more capable and the data required to train, evaluate and improve them is becoming more sophisticated. As AI systems move beyond generating answers to reasoning and taking action, developers increasingly need expert data, realistic environments and rigorous evaluations to test performance and identify where systems still fall short.

That shift is central to our investment in Snorkel AI, the frontier AI data lab that Blumberg Capital is investing in as part of its $350M Series E at a $3.5B valuation. The round was led by Insight Partners and S32. Blumberg Capital joins new investors March Capital, Allegis Capital, Frontline, Standard, One Prime Capital, Third Point Ventures and D.E. Shaw Ventures, alongside existing investors including Addition, Greylock, Lightspeed, GV, Factory, BlackRock, Prosperity7, Walden Catalyst and Wells Fargo.

The financing follows a period of rapid growth for Snorkel. The company has surpassed $375M in annual run-rate recurring revenue, up more than 10x in nine months. Snorkel launched its expert Data-as-a-Service offering in September 2025 and now works with leading AI labs and enterprises on the data behind frontier model evaluation and training.

The opportunity looks very different today than it did during the earlier stages of machine learning. Snorkel describes that earlier period as “Data 1.0,” when data development largely meant labeling relatively simple tasks at scale. Today’s frontier and agentic AI systems require expert agentic tasks, environments and rubrics that can take highly qualified people hours or even days to construct. Designing them well is increasingly research work, where quality and complexity determine value.

As models become more capable, differentiation increasingly shifts to the specialized, high-difficulty data needed to improve performance on problems where a merely plausible answer is not enough. The challenge is no longer simply sourcing more data. It is determining what data a model needs, developing the expert workflows to produce it and evaluating whether it actually improves performance.

Frontier and agentic AI are changing what is required of the data behind them. Snorkel has spent years developing the research and technology to build expert data and environments for increasingly complex AI systems. Snorkel calls this shift Data 2.0, and its agentic data factory is designed around it. The company combines human expertise with research and technology to develop the data and environments used to evaluate and improve advanced AI systems.

Our conviction starts with the team.

Co-founder and CEO Alex Ratner and co-founder Chris Ré built Snorkel out of the Stanford AI Lab, where their work on weak supervision and data-centric AI helped pioneer the field. That research foundation continues today, with more than 250 peer-reviewed papers cited more than 25,000 times.

What differentiated Snorkel for us was the depth of the team’s work in data development and how relevant that work has become to the problems AI developers are confronting today. The team has deep technical credibility and is applying that expertise to a clear and growing commercial need.

A growing market for more complex AI data

We have seen enterprise AI adoption move beyond simply purchasing subscriptions to standard AI services. More sophisticated organizations are increasingly looking at how they can customize AI for their own workflows, performance requirements, costs and data environments. That evolution increases the need for specialized data and evaluation.

As AI moves into more specialized and demanding applications, training and evaluation increasingly require domain expertise and environments that reflect the complexity of real-world work. This creates opportunities beyond frontier AI labs and into industries where accuracy and specialized knowledge matter.

Snorkel is already extending its capabilities into longer-running, complex work in areas including healthcare, law and software engineering. With the new funding, the company plans to expand the capacity of its agentic data factory, accelerate investment in vertical and enterprise AI, extend its research and technology into new domains and modalities, and deepen its investment in open research, including its Open Benchmarks Grants initiative.

We believe Snorkel is well positioned as specialized data and evaluation become increasingly important to both frontier and enterprise AI. Its combination of deep research, technology and expert human knowledge is difficult to replicate and increasingly relevant as AI systems take on more complex work.

Congratulations to Alex, Chris and the entire Snorkel team. We look forward to partnering with them as they continue to build the data and environments behind frontier and agentic AI.

Hear directly from Alex in his interview with TBPN.