AI Training Data Startups Reach Total Revenue of $8.5 Billion
According to Deedy's statistics, over 50 startups sold data and RL environments to large AI labs by July 2026, driving advancements in AI behind the scenes, with total revenue of approximately $8.5 billion and a combined valuation of about $100 billion. More than 75% of the market share is dominated by four companies: Scale, Surge, Mercor, and Handshake. This ecosystem has become a key infrastructure supporting the training of cutting-edge models. Source: Public Information
ABAB AI Insight
Deedy has long observed AI infrastructure and previously analyzed the data supply chain. Companies like Scale, which started with early data labeling, have grown into multi-billion dollar revenue giants, reflecting a centralization trend similar to the RWA or synthetic data wave from 2023 to 2025. The four leading companies have concentrated capital and talent resources through large-scale collection, professional labeling, and high-fidelity RL environments, forming an oligopoly. Their motivation lies in the rigid demand for high-quality data in post-training of large models, driving the entire industry's valuation inflation. Similar to the early dominance of AWS in cloud services or Nvidia's position in chip design, the AI training data field is currently transitioning from fragmented supply to concentration on a few platforms, with the remaining 50+ companies competing for the leftover market. Essentially, this represents a concentration of capital: data is a core production factor for AI, and leading players absorb most of the revenue through network effects and technological barriers. The mechanism is that economies of scale and proprietary dataset advantages create a moat, accelerating AI progress while raising the entry barriers for smaller players. ABAB News · Cognitive Law 1. Data oligopolies determine the AI ceiling, with 75% market share serving as a barrier. 2. Behind-the-scenes infrastructure, with a front-end valuation of $100 billion. 3. Those selling shovels profit immensely, while gold diggers rely on the supply chain.