"Battlefield" Shifts! Call Volume Grows Over 1,000x in Two Years, Token "Computing Power War" Begins

On March 25th, all three major indices opened higher, with CPO movements pushing prices up. The ChiNext Artificial Intelligence ETF (159243) opened nearly 4% higher. Top constituent stocks included Tianfu Communication, which rose over 5%, Capital Online and Runze Technology, both up nearly 5%, with Zhongji Xuchuang and Blue Focus also gaining.

In terms of news, just yesterday, Token received its Chinese name—Ciyuan. The National Data Bureau indicated that at the beginning of 2024, China’s daily Token usage was 100 billion; by the end of 2025, it is expected to surge to 100 trillion; in March this year, it already exceeded 140 trillion, representing over a thousand-fold growth in two years. This signifies that a new business logic based on Token billing is accelerating its development.

Tokens are the measurement units for processing information in large models. As AI large models evolve from “chat tools” to “autonomous agents capable of executing tasks,” the core focus of AI has shifted from one-time model training to continuous, massive Token generation (i.e., inference). As the smallest information unit processed by large models, Tokens have the characteristics of being measurable, priceable, and tradable in the era of intelligence.

Industry experts point out that the new generation AI Agents, represented by OpenClaw, have completed a fundamental transition from “question-and-answer tools” to “practical assistants capable of deployment,” enabling deep integration into office scenarios and bringing disruptive efficiency improvements. Global AI competition has entered a deep phase centered around “Token” as a core metric for industrialization and commercialization. AI Agents, such as those used for “lobster farming,” consume far more Tokens than traditional AI applications, which will directly drive demand for computing power and cloud services. According to Huatai Securities, the scale of Token calls and efficiency are accelerating, which will promote a new cycle of price increases in the computing power leasing market.

A year ago, major internet companies were busy fighting the “food delivery war”; now, the battlefield has shifted to computing power and Tokens. On March 18th, Alibaba Cloud announced price hikes of up to 34% for its AI computing power, storage, and other products. On the same day, Baidu Cloud also announced on its official website a restructuring of prices for AI computing power and storage, stating that these adjustments are to ensure long-term platform stability and service quality. Earlier, on March 11th, Tencent Cloud announced a 463.13% increase in the input price for the Tencent HY2.0 Instruct model.

Data shows that the ChiNext Artificial Intelligence ETF (159243), which tracks the ChiNext AI Index, covers the entire AI industry chain—AI hardware (40%), AI applications (16%), and AI software (44%). The top ten constituents include Zhongji Xuchuang, Xin Yisheng, Tianfu Communication, Blue Focus, Runze Technology, Xiechuang Data, Beijing Junzheng, Kunlun Wanfang, Tonghuashun, and Shenxinfu.

Historically, the ChiNext AI Index has shown stronger resilience. Over the past three years, the index has gained 304.69%; since 2025, it has increased by 132.48%, significantly outperforming similar Sci-Tech Innovation AI indices and the CSI AI Index. In terms of maximum drawdown, all three are comparable. According to Wind, the projected net profit growth rate for the ChiNext AI Index in 2025 could reach 127.36%, notably higher than the ChiNext Composite Index, with industry prosperity expected to further improve.

(Author: Zhang Xiaobo)

【Disclaimer】This article reflects only the author’s personal views and has no relation to Hexun.com. Hexun.com maintains neutrality regarding the statements and opinions expressed in this article and does not guarantee the accuracy, reliability, or completeness of the content. Readers should use this for reference only and bear all responsibilities themselves. Email: news_center@staff.hexun.com