Global Demand for Chinese AI Models Surges as Usage and Efficiency Rapidly Expand
By Tanveer Ahmed :
Chinese artificial intelligence models are seeing growing adoption overseas as businesses and individuals increasingly rely on them to improve productivity and solve complex tasks.
Each interaction with an AI system consumes digital resources known as “tokens”, the smallest units of text processed by language models. A token can represent a word, part of a word, punctuation, or even a single character. As the use of AI tools accelerates, demand for these tokens has risen dramatically.
According to industry data, China’s average daily token consumption surpassed 140 trillion in March this year, marking an increase of more than 1,000 times compared with the beginning of 2024.
Yan Yijun, vice president of Chinese AI company MiniMax, said users worldwide are primarily concerned with whether AI systems are intelligent, fast and capable of solving complex problems, while also offering reasonable and sustainable pricing.
“The growing international adoption of Chinese AI models shows that users around the world are increasingly confident in these technologies,” Yan said.
However, balancing performance with affordability remains a major challenge for developers of large AI models. Yan said MiniMax has attempted to address this by improving algorithm efficiency in its “MiniMax M2.5” model, allowing the system to produce responses using fewer tokens while extracting greater value from each one.
The company claims the model can process around 100 tokens per second and can operate for an hour at a cost of roughly one dollar. Estimates suggest that Chinese AI systems can deliver similar performance at around one-tenth of the cost of comparable models developed in the United States.
Experts say the lower costs are not only the result of technological progress but also China’s structural advantages in energy supply and industrial infrastructure.
Li Zhiqing, a professor at Fudan University’s School of Economics, said China benefits from relatively low electricity costs and a well-developed supply chain that supports the rapid expansion of AI computing power.
By December 2025, the number of generative AI users in China had reached 602 million, representing a 141.7 per cent increase compared with the previous year.
The surge in adoption is pushing AI beyond its traditional internet applications and into more complex areas such as office productivity, industrial design and collaborative work. As usage expands, the data generated by these interactions helps refine AI models and enhances their ability to perform increasingly sophisticated tasks.
Energy consumption remains a crucial factor for the global computing power industry, particularly as AI systems require vast amounts of electricity to operate.
Li noted that China holds a competitive edge due to its extensive power infrastructure, including the world’s largest electricity supply network and advanced ultra-high voltage transmission systems. Combined with the growing use of renewable energy, these factors help provide stable and relatively inexpensive computing power.
He added that China’s comprehensive AI supply chain—from chip manufacturing and servers to computing infrastructure, cross-border connectivity and digital payment systems—has further reduced operational costs.
Li said the economic value generated by AI computing can far exceed traditional industries. While one kilowatt-hour of electricity typically produces one to two times its value in conventional manufacturing, the digital economy can multiply that value many times over.
According to Li, China’s energy and manufacturing strengths are increasingly being transformed into digital services that reach global users, meaning the electricity consumed domestically is effectively generating value far beyond its borders.
About Author
