China's official data has confirmed an astonishing explosion in artificial intelligence (AI) usage across the country.
This surge is best measured by a new key metric: AI tokens, which essentially count the units of data processed by AI models. According to China's National Data Administration (NDA), daily token consumption grew from just 1 trillion in early 2025 to 100 trillion by the end of the year, and then accelerated even further to over 140 trillion per day by March 2026. This isn't just a number; it represents a massive economic surface area. At conservative valuations, this volume of activity could represent a market for AI services worth between $2.1 billion and $21.1 billion annually.
So, what's driving this incredible growth? A major factor is a coordinated push from the government. First, China has been building a massive national computing infrastructure through its 'East-Data-West-Compute' project. This initiative creates large-scale data centers in western regions where energy is cheaper and connects them to the data-heavy eastern cities, effectively building a superhighway for AI computation. Second, the government has actively reformed its data policies to create "data-element markets," making it easier to produce and trade high-quality datasets—the essential fuel for training and running powerful AI models.
Alongside this government-led infrastructure and policy push, technological self-reliance and market competition have played a crucial role. First, in the face of U.S. export controls on advanced chips from companies like Nvidia, China accelerated the development of its own domestic AI accelerators, such as Huawei's Ascend series. These homegrown chips are now powerful enough to support massive-scale AI operations. Second, a fierce price war has erupted among major Chinese tech companies like Baidu, Alibaba, and DeepSeek. They are drastically cutting the prices of their AI model services, making it incredibly cheap for developers and businesses to use AI, which in turn drives up token consumption.
In essence, this token explosion is no accident. It's the visible result of a multi-pronged national strategy that combines top-down infrastructure planning, supportive data policies, a determined push for technological independence, and fierce bottom-up market competition. China is not just building AI models; it is building a complete, self-reinforcing ecosystem where massive AI usage is becoming the new normal.
- Glossary
- AI Token: A basic unit of data (like a word or part of a word) that an AI model processes. Counting tokens is a way to measure how much an AI is being used.
- East-Data-West-Compute: A Chinese national strategy to build a unified network of large data centers, mostly in the energy-rich western regions, to process data coming from the more populated eastern regions.
- Inference: The process of using a trained AI model to make predictions or generate outputs based on new, unseen data. This is what happens when you ask an AI a question.