China's Intelligent Computing Capacity Reaches 2.45 Million PFLOPS
China's National Data Bureau announced that the country's total intelligent computing capacity reached 2.45 million PFLOPS (FP16) as of July 2026. Over 85% of this capacity is concentrated within eight national computing hubs and three computing-electricity collaborative development zones. This rapid expansion highlights China's aggressive build-out of AI infrastructure to support large-scale AI model training and deployment. The integration of computing and power grids (computing-electricity collaboration) addresses the massive energy demands of these data centers. Out of the total capacity, 1.45 million PFLOPS has been integrated into the national monitoring and scheduling platform to optimize resource utilization. Additionally, China authorized over 3,400 AI agent patents in 2025, representing a growth rate more than double that of the previous year.
## BACKGROUND
PFLOPS (PetaFLOPS) measures computing speed, representing one quadrillion floating-point operations per second, with FP16 indicating the half-precision floating-point format commonly used in AI. "Computing-electricity collaboration" is a national strategy in China aimed at balancing the high energy consumption of supercomputing centers with green energy grids.