China data bureau backs token based AI data markets

China’s data regulators are making explicit what many in the industry already understood implicitly: in a world dominated by large models, tokens are not just a technical unit but an economic one. By talking about “token‑based” business models, token trading and paid access to high‑quality datasets, the National Data Bureau is sketching the contours of a domestic market where data and model usage are tightly coupled and monetized under state supervision. That is a different starting point from the largely private, API‑driven data markets in the US and Europe.

From a race‑to‑AGI perspective, this matters because it signals sustained national‑level investment in the inputs that determine frontier capability: curated datasets, standardized data infrastructure and large‑scale compute. If China can make it easier for domestic firms to access regulated, high‑quality data at scale, it narrows one of the key advantages held by US labs. At the same time, a more formal token market could give authorities levers to throttle or prioritize different AI use cases, influencing which sectors get the most model capacity.

The bigger picture is that data policy has moved from a privacy and localization issue to a core instrument of AI industrial strategy. Countries that treat data, tokens and compute as a coherent stack will likely have an edge in shaping how and where near‑AGI systems are trained and deployed.