Snorkel AI Received $350 Million Investment: Company Value Increased to $3.5 Billion

San Francisco-based Snorkel AI closed a $350 million funding round led by Insight Partners and S32. Existing investors Addition, Greylock and Wells Fargo also participated in the round. With this investment, the company's value almost tripled from $1.3 billion in the round in which it raised $100 million in May 2025 to $3.5 billion.

According to information provided by CEO Alex Ratner to Reuters, the intense demand of frontier AI laboratories for increasingly complex training data and simulated environments was the main factor driving growth.

The company's annualized revenue run-rate jumped to $350 million from about $20 million a year ago. This rapid increase is due to the "data-as-a-service" model that was implemented in September 2025. Founded in 2019 by a group of researchers from the Stanford AI laboratory, Snorkel AI sold software in the early years, but today it has focused on offering ready-made data sets and reinforcement learning (RL) environments directly to its customers.

Snorkel has launched an “agent-based data development platform” that combines human experts with thousands of custom AI models and agents. Experts design the scenarios, tasks and evaluation criteria; Artificial intelligence automates most of the quality control processes. The company uses a vast network of tens of thousands of experts from fields such as coding, law and medicine. Ratner stated that they charge for the data products produced, not for human labor, and emphasized that this model protects profit margins while paying higher fees to experts.

Ratner said AI developers are moving away from simple labeling tasks and turning to more difficult and high-risk data:

"100 percent of the data where laboratories will create value will have human input in the foreseeable future. However, to keep up with this complexity, synthetic and automated approaches will have to be used on 100 percent of the data."

Snorkel works with frontier AI labs, hyperscalers, enterprise customers and the US federal government. Coding data is one of the areas in highest demand.

New capital; It will be used to recruit researchers and engineers, expand corporate and government operations, support third-party AI model evaluations, and enter new sectors and data types. While the company prioritizes growth, it aims to reach profitability this year. "Data is becoming rarer, more specialized and harder to find. If you want to train the most advanced, complex and capable models, you now need superior data," said S32 partner Andy Harrison.

After Meta acquired 49 percent of Scale AI for $14.3 billion in June 2025, venture capital investments in startups that provide human-tagged training data accelerated. Rivals such as Mercor and Surge AI also continue to attract investor attention with strong revenue growth.