The Ninth Digital China Summit in Fuzhou has signaled a profound paradigm shift in the nation's artificial intelligence trajectory. No longer content with the novelty of generative dialogue, China’s tech giants and state regulators are pivoting toward 'autonomous execution'—a world where AI agents perform complex tasks rather than just answering questions. This evolution is being quantified through a new economic metric: the Token. By reframing the 'Token' (the smallest unit of data processed by a large model) as a tradable, measurable commodity, Beijing is laying the groundwork for a standardized 'intelligent economy.'
Data released by the National Data Bureau underscores this explosive growth. Daily Token consumption in China reportedly surged from 1 trillion at the start of 2025 to over 100 trillion by the end of that year, reaching 140 trillion by March 2026. Crucially, 2025 marked the first time that data used for AI inference—the actual application and usage of models—surpassed data used for training. This inflection point indicates that China's AI sector has moved out of the laboratory and into the gears of daily industrial and consumer life.
To support this massive demand, the summit highlighted a concerted effort to build a 'sovereign' computing stack. Alibaba showcased its Panmai 920 smart network card and Zhenwu 810E AI chips, designed to power massive clusters of 10,000-plus GPUs while reducing the production cost of Tokens. Meanwhile, China Telecom has spearheaded the 'Token Ecosystem Alliance,' aiming to build a national integrated computing network that treats AI processing power as a utility as accessible and elastic as electricity.
The push toward 'AI Agents' was the most visible trend among platform players. Tencent unveiled its Agent ecosystem, focusing on 'SkillHub'—a community that allows AI to utilize core tools like Tencent Meetings and Maps to execute real-world workflows. Alibaba countered with its Qwen 3.6 Plus model, which boasts a million-token context window, enabling it to digest entire libraries of technical data to perform highly specialized roles in fields like programming and corporate management.
Beyond hardware and software, the establishment of the 'Data Infrastructure Technical Community'—comprised of over 85 entities including state-owned enterprises, universities, and private giants like Baidu and Huawei—points to a structural consolidation. This collective aims to bridge the gap between fragmented data pools and AI applications, ensuring that the next phase of China’s digital transformation is not just about intelligence, but about the sustainable, large-scale automation of its entire economic engine.