According to a research report from China Galaxy Securities, telecom operators are transitioning from traffic-based operations to a Token value-based business model. Leveraging their extensive computing power layout, nationwide network coverage, massive user base, and enterprise/government channels, they have secured a unique position in both the high-barrier AI market for specific industries and the inclusive AI service market. This shift is expected to open a second growth curve in the AI era, offering a combination of stable dividends and technological growth potential. Key stocks to watch include China Mobile (600941.SH, 00941), China Telecom (601728.SH, 00728), and China Unicom (600050.SH, 00762). The core views of China Galaxy Securities are as follows:
**Shanghai Telecom Launches Token Fee Plan, Creating a New Revenue Stream** China Telecom launched a nationwide Token plan on May 17, 2026, with a minimum cost of 9.9 yuan per month for individuals and families, providing 10 million Tokens. For developers and small-to-medium enterprises, plans start at 39.9 yuan per month. Concurrently, a 17.4 billion yuan procurement for a Token factory was initiated to build infrastructure for token generation, computing power aggregation, model adaptation, and intelligent scheduling. With this move, all three major telecom operators have now introduced Token plans, standardizing the retail of AI services similar to data packages, making Tokens the standardized unit of measurement in the AI era. The daily Token usage in China has surged from 100 billion at the beginning of 2024 to 140 trillion by March 2026, representing growth of over a thousandfold in two years.
**Improving Utilization of Existing Computing Power and AIDC, Enhancing Returns on Heavy Assets** The most immediate benefit of operators introducing Token plans is the transformation of heavy-asset capabilities—such as cloud services, computing power, AIDC, and network resources—into retailable, subscribable, and sustainable AI service revenue units. This restructures revenue streams, improves asset utilization, and enhances Average Revenue Per User (ARPU). Previously, operators monetized their cloud computing capabilities primarily through enterprise projects and IDC leasing, which were more business-to-business (B2B) focused. Token plans, however, for the first time provide low-threshold access to individuals, developers, and small-to-medium enterprises, turning previously high-barrier model invocation and inference services into standardized products accessible to the general public. This significantly lowers the trial threshold for a vast number of ordinary users. After purchase, users can access various mainstream large models via standard APIs, positioning operators to capture future consumption entry points for AI agents and lightweight applications. This means that for the same set of AIDC and computing resources, if usage frequency can be increased and more scenarios accessed through Token plans, the return on unit assets will improve markedly. Following the launch of the large model Coding Plan by China Mobile Cloud, daily Token usage increased by 274% by April this year compared to December 2025, with the number of active clients rising by 704%.
**Massive User Base with Low Customer Acquisition Costs + Integrated Cloud-Network-Edge-End Infrastructure: Telecom Operators as the Strongest Distribution Channel** The advantage of telecom operators lies in their 1.827 billion individual mobile users and approximately 73 million enterprise/government users, giving them control over massive Token consumption entry points. Their strengths in nationwide delivery, billing, enterprise services, and cloud-network integration are very clear. From minutes to data gigabytes, to broadband and bundled plans, operators excel in designing standardized packages, billing settlements, and large-scale operations. Token plans essentially represent the application of their historical capabilities in the AI era. More importantly, operators possess national backbone networks, data centers, edge nodes, dedicated lines, and metropolitan area network resources, enabling cloud-network-edge coordinated scheduling. This grants them advantages in latency, security, and reliability, forming significant barriers. They are well-positioned to better control model entry points and Token pricing power.
**Risk Warnings:** Risks include slower-than-expected adoption of AIGC applications; capital expenditures and investment returns falling short of expectations; and intensifying competition in the computing power industry.
Comments