China shifts AI focus to inference and establishes national computing power monitoring

China is shifting its artificial intelligence focus from large-scale model training to the deployment of AI agents and inference services. The China Academy of Information and Communications Technology predicts that inference computing power could account for 80% of China's total computing market by 2029, as the industry moves toward commercializing AI through task-oriented agents that can execute complex workflows.

To support this transition, China has established a national integrated computing power monitoring system. Under the guidance of the Ministry of Industry and Information Technology, the China Computing Power Platform has successfully connected platforms from 31 provinces. This unified network aims to eliminate regional data silos and provide real-time, automated monitoring of computing resources. The platform currently serves over 10,000 registered enterprises and hosts more than 300 large-scale models.

During the 2026 China Computing Power Conference, officials emphasized the need to build a multi-layered architecture spanning hubs, regions, and edges. Future goals include developing high-speed, low-latency transmission channels and promoting the synergy between computing power and electricity to lower costs and application barriers for various industries.