The daily average token usage has increased by more than 1000 times! The outbreak of China's AI industry

Compared to the 100 billion at the beginning of 2024, it has increased by over 1000 times in two years; Even compared to the 100 trillion yuan at the end of 2025, it has achieved a growth of over 40% in just three months. The exponential surge in token usage marks that China's AI has officially entered an explosive period of large-scale application and value release from the stage of technological exploration.

Simply put, Token is the smallest unit for AI to understand and process information, like the "words and phrases" in human communication. Whether it is asking AI questions, generating copy, writing code, or AI completing data analysis and image recognition, all intelligent interactions will ultimately be broken down into Tokens for computation. Therefore, the daily average amount of Token calls directly reflects the actual usage intensity and application breadth of AI models, and is a core indicator for measuring the activity and industrialization level of the AI industry, just like the "thermometer of the AI industry", accurately reflecting the development pulse of the industry.

From 100 billion to 140 trillion, with a growth of over 1000 times in two years; From 100 trillion to 140 trillion, with a growth rate of over 40% in three months, the explosive increase in token usage is not accidental, but the inevitable result of multiple factors working together. Behind it is the comprehensive maturity of China's AI industry ecosystem.

One of the core driving forces is the deep penetration and qualitative change of AI application scenarios. In the past, AI appeared more as an "entertainment tool", and although applications such as writing poetry and painting attracted attention, it was difficult to form a large-scale and sustainable demand; Nowadays, the real business needs from various industries support a daily average usage of 140 trillion yuan. AI has transformed from a mere embellishment to a core infrastructure for enterprises to reduce costs and increase efficiency. In the fields of industrial manufacturing and warehousing logistics, AI is deeply integrated with robots and sensors, possessing the ability to perform complex tasks; In the field of biomedicine, AI reshapes the drug development process and accelerates the emergence of new drugs; In fields such as film and office, AI generated content and intelligent office tools are widely used, and high-frequency, large-scale paid applications have become the core engine for token consumption growth. At the same time, national level applications such as WeChat and DingTalk have integrated AI capabilities, reaching billions of users and amplifying the scale effect of token consumption through daily calls from a massive number of individual users.

The cost-effectiveness advantage and technological breakthrough of domestic large models provide core support for the surge in demand. Represented by MiniMax, DeepSeek, and other domestic large-scale models, they have reached international standards in coding, intelligent agent, and other scenarios, while API prices are only 1/5-1/10 of similar models in the United States, significantly reducing the usage costs for enterprises and developers, and attracting widespread calls from developers around the world. At the same time, domestic large-scale models continue to make breakthroughs in inference chips, architecture, and system optimization, enabling them to output intelligent services with less computing power and higher energy efficiency. Combined with the application of green power resources and liquid cooling technology in western China, the cost of computing power has been further reduced, increasing the token output per unit of computing power by 60%, laying a technical and cost foundation for large-scale deployment.