China to Explore Token-Based Subscriptions and Expand Data Labeling Pilots
At the 2026 China International Big Data Industry Expo, Liu Liehong, Director of the National Data Bureau, announced plans to explore token-based value-added subscriptions and pay-per-effect business models. Additionally, the bureau will launch a new batch of data labeling pilots across 32 cities to support AI development. As global competition in large language models intensifies, establishing standardized token commercialization and data labeling infrastructure is crucial for scaling AI industries. These initiatives will help address bottlenecks in data processing and establish clear economic metrics for AI resource consumption. By June 2026, China's average daily token usage exceeded 500 trillion, driving massive demand for computing power. To support this growth, the China Academy of Information and Communications Technology (CAICT) and 22 organizations have initiated a working group to build a high-quality token service ecosystem.
## BACKGROUND
In AI, a "token" is the basic unit of text processing used by large language models, serving as a key metric for pricing and transaction settlement. Data labeling is the process of tagging raw data to make it usable for training machine learning models. As AI models require vast amounts of structured data, efficient labeling and standardized token pricing are essential for commercial viability.