China's AI ambitions: Beijing is seeking not only to build powerful AI models, but also to become a major global supplier of the data used to train them. China's National Data Administration has outlined a plan to build large, high-quality datasets and make them available internationally. The objective is partly technological—improving Chinese AI capabilities—but also geopolitical, because whoever supplies important training data can gain influence over what future AI systems know and how they understand the world.
The motivation is closely connected to the language and cultural imbalance in existing AI training data. Much of the world's high-quality digital material is in English, which means many AI models have been trained predominantly on information reflecting Western perspectives. Chinese officials and researchers see this as a strategic weakness. By producing large datasets containing Chinese-language material, scientific information, images, videos and professionally labeled examples, China could strengthen its own models while also increasing the amount of Chinese-origin information incorporated into AI systems used around the world.
This effort goes beyond simply collecting more internet data. China is reportedly investing in expert data annotation, turning raw information into carefully labeled datasets that can be particularly valuable for advanced AI systems. State-backed datasets such as WanJuan are part of this broader ecosystem. The potential influence is significant: if international AI developers increasingly rely on Chinese datasets, the values, assumptions and information-selection choices embedded in those datasets could indirectly affect AI systems used outside China. That creates a new dimension to the geopolitical AI competition—control over data can matter almost as much as control over chips and models.
The broader takeaway is that the AI race is expanding from models and computing power to information itself. China already has a large consumer AI ecosystem—ByteDance's Doubao, for example, surpassed 100 million daily active users during China's 2026 Lunar New Year period—giving Chinese companies access to enormous volumes of real-world interaction data. If Beijing succeeds in becoming a major international supplier of high-quality AI training data, it could gain influence over the foundations on which future AI systems are built.
