A recent forecast from research firm IDC suggests that China's AI agent market is shifting into high gear.
The report projects that the number of active AI agents used by Chinese companies will soar to over 350 million by 2031, a clear sign that we're moving from a period of experimentation to one of mass adoption. But what's fueling this incredible growth? It's a powerful combination of three key factors: policy, technology, and infrastructure.
First, the Chinese government is providing a strong tailwind with its 'AI+' action plan. This national strategy sets ambitious goals, aiming for 70% of new devices and agents to be next-generation by 2027, and 90% by 2030. This gives companies a clear signal and the confidence to invest heavily in integrating AI agents into their operations.
Second, the cost and complexity of deploying AI are plummeting. A new technical standard called the Model Context Protocol (MCP) is acting like a universal adapter for AI, making it much easier and cheaper for different AI models and tools to work together. Think of it as the 'USB-C for AI'. At the same time, major cloud providers like Alibaba are aggressively cutting prices for their AI services, making the technology more accessible to businesses of all sizes.
Finally, China is rapidly building up its computing power, or '算力' (suànlì). This is crucial because millions of AI agents running simultaneously require immense computational resources. The expansion is supported by both homegrown chips, like Huawei's Ascend series, and the partial resumption of high-performance chip imports from companies like Nvidia. This dual approach ensures a stable and growing supply of the processing power needed to sustain the AI boom.
While IDC's prediction of a 30-fold average annual increase in token consumption seems astronomical, it's best understood as reflecting explosive growth in the initial years from a very low base. The growth rate is expected to moderate as the market matures. Nevertheless, all signs point to a transformative period where AI agents become integral to how business is done in China.
- AI Agent: An AI program designed to understand goals and take independent actions within a digital environment to achieve them, such as scheduling meetings or processing customer orders.
- Token: The basic unit of data, like a word or part of a word, that an AI model processes. The cost of using an AI is often measured by the number of tokens it consumes.
- MCP (Model Context Protocol): A technical standard that allows different AI models to easily connect with and use various external tools and data sources, reducing integration costs.