China's AI inference data surpasses training data, marking a shift to industrial application

According to a report by the China Academy of Information and Communications Technology (CCID), China's artificial intelligence inference data volume has for the first time exceeded training data volume, signaling a transition from model development to large-scale industrial application. As of June 2026, the daily average token invocation volume surpassed 175 trillion, representing over 1,000 times growth compared to early 2024. The iteration cycle for top-tier models has also been reduced from months to weeks or even days. The report highlights the progress of the 'AI + Manufacturing' initiative, with over 500 advanced intelligent factories established across 31 provinces, covering more than 80% of manufacturing sectors. Over 1,000 application scenarios have been implemented in 20 industries, including steel and automotive. Additionally, the China Internet Network Information Center's report indicates that as of the first half of 2026, the user base of generative AI in China exceeded 700 million, with a penetration rate surpassing 50%.