US data labeling companies that help train American military artificial intelligence are simultaneously earning substantial revenue from China's leading AI laboratories. Collectively, these firms take in roughly $500 million per year from China's six top AI operations — a commercial relationship that sits uneasily alongside Washington's escalating efforts to restrict Chinese access to advanced technology.

The numbers behind the dual-client model

Surge AI reported $1.2 billion in revenue in 2024, built on supplying expert-annotated datasets to both US military branches and Chinese AI laboratories. These annotations are complex, domain-specific datasets used to train sophisticated AI models — work that demands genuine expertise and commands premium pricing.

Mercor has grown even faster, surpassing an annualized revenue run rate of $2 billion by June 2026. Chinese contracts accounted for 2% of its Q2 2026 revenue.

AfterQuery earns at least $50 million in recurring annual revenue from Chinese AI labs alone, while Turing rounds out the group of notable US firms participating in this trade.

On the buyer side, the client list includes ByteDance, Alibaba, Ant Group, and Tencent. The Pentagon has designated Tencent a Chinese military company, meaning a US defense-adjacent data firm is simultaneously servicing an entity that the Department of Defense considers part of China's military apparatus.

Why data annotation matters more than chips

The datasets these American firms produce require English-fluent domain experts, often with advanced degrees, capable of providing the nuanced annotations that teach AI systems to reason, code, and analyze. This is precisely the kind of human expertise that is hardest to replicate domestically in China — which is why Chinese labs are willing to pay American firms hundreds of millions of dollars for it. High-quality human feedback has become a core input for training and refining large language models, and as labs push toward more capable systems, demand for expert annotation — in fields like law, medicine, and software engineering — has grown accordingly.

Companies such as Anthropic also rely on these same annotation services. In practice, the supply chain serving America's most security-conscious AI developers is the same pipeline feeding Chinese competitors.

The regulatory gap

The situation exposes a significant blind spot in US technology export policy. While the Commerce Department's Bureau of Industry and Security has built increasingly sophisticated restrictions around chip exports, chip-making equipment, and even cloud computing access, no equivalent framework governs the sale of annotated training data. Whether lawmakers move to close that gap — by extending export controls or restricting defense-adjacent contractors from serving designated Chinese military companies — is likely to shape both the revenue base of US data firms and the training pipelines available to China's AI labs.