China’s Data Output Surges in 2025, Fueling Large-scale AI Industrial Adoption

The newly released National Data Resources Development Report 2025 reveals robust expansion of China’s data scale and supporting digital infrastructure, laying a solid foundation for the accelerated iteration of artificial intelligence industrial applications. In 2025, the country’s annual data production volume reached 52.26 zettabytes, representing a year-on-year increase of 27.28 per cent, with the growth rate rising by 2.28 percentage points compared with the previous year. China’s total data output accounts for approximately 27.44 per cent of the global volume, maintaining a prominent position in the global data economy.

The report shows that corporate entities dominate domestic data production, contributing nearly 90 per cent of the nation’s incremental data volume. Traditional industrial sectors register steady data growth, with manufacturing, logistics, and software and information technology services recording prominent output increases of 1.27 ZB, 1.22 ZB and 0.92 ZB respectively. Emerging digital sectors demonstrate explosive growth momentum, as embodied intelligence and low-altitude economy achieve year-on-year data growth of 477.78 per cent and 75 per cent. The national data storage capacity also maintains stable expansion, hitting 2.53 ZB in 2025 with a year-on-year growth of 21.05 per cent.

China’s computing power infrastructure continues steady and systematic development. The full implementation of the East Data, West Computing initiative accelerates the formation of an integrated national computing network and steadily elevates intelligent computing supply capacity. By the end of 2025, the scale of national intelligent computing power has reached 1.59 million PFlops. The ongoing generational shift from general computing to intelligent computing strengthens core infrastructure support for continuous AI innovation and industrial deployment across multiple sectors.

Massive high-quality data resources drive AI technology into large-scale industrial application scenarios. In 2025, the total data used for AI training and inference reached 199.48 exabytes, rising 42.86 per cent year on year. Inference data volume stood at 101.34 exabytes, surpassing training data volume for the first time and reflecting the deepening commercial and practical deployment of AI models.

The quality and scale of fundamental AI data resources see remarkable improvement. The country owns more than 110,000 high-quality datasets with a total scale exceeding 908 petabytes, marking respective year-on-year growth of 61.13 per cent and 142.58 per cent. Token calling volume reached approximately 2110 trillion throughout the year, establishing tokens as a new core measurement standard for evaluating artificial intelligence operational capabilities and application maturity.

The coordinated growth of data production, storage and computing power forms a complete industrial support system for digital economy development. The continuous expansion of multi-scenario data resources and the upgrading of intelligent computing infrastructure will further unlock the value of data elements, enabling deeper integration between data resources and artificial intelligence technologies and empowering the high-quality development of the national digital economy.