According to the latest data from the National Data Administration, by 2025, the total amount of data used for training and inference in China's artificial intelligence field is expected to reach 199.48 EB (exabytes, 1 EB = 1024 PB = 1,048,576 TB), an increase of 42.86% year-on-year. Among them, the volume of inference data will exceed that of training data for the first time, reaching 101.34 EB. This growth trend shows the breadth and importance of artificial intelligence technology in various industries.

Additionally, the data generated by China's system software and artificial intelligence in 2025 is expected to reach 26.92 ZB (zettabytes, 1 ZB = 1024 EB), which is the first time this volume has exceeded that of traditional Internet of Things sensing data, marking the rapid development of emerging technologies and the explosive growth of data generation. The National Data Administration also predicts that future demand for inference computing power will be in a ratio of 3:1 or even higher compared to training computing power.

To meet the demand for urban computing power, the National Data Administration plans to focus on scenarios with low latency, high reliability, and high security to adapt to the changing digital economy environment. According to the data, the daily word token call volume across the country is expected to increase from over 10 trillion at the beginning of 2025 to 100 trillion by the end of the year, showing significant exponential growth, with the total word token call volume throughout the year expected to reach about 21,100 trillion.

As artificial intelligence technology continues to advance and its application scope expands, future data processing and computing power demands will show more diversified and complex trends. The National Data Administration will continue to monitor industry dynamics, promote the formulation of relevant data standards and technological innovation, and promote the sustainable development of the artificial intelligence industry.

Key Points:   

🌟 By 2025, the total amount of artificial intelligence data in China will reach 199.48 EB, up 42.86% year-on-year.  

📈 The volume of inference data will exceed that of training data for the first time, reaching 101.34 EB.  

🏙️ The National Data Administration will focus on urban computing power demand, promoting the application of low-latency and high-reliability technologies.