League of Nations Minsheng Securities: Token “inflation” is the core main line of the current AI industry market

The Zhitong Finance App learned that Guolian Minsheng Securities released a research report saying that the average domestic daily calling volume has increased more than a thousand times in two years. Under the gap between supply and demand, the price of computing power leasing and cloud services at home and abroad continues to rise, while C-side tiered payments have been implemented to verify the commercial value of tokens. The top-level system design establishes a closed loop of the entire chain from data supply to model development to scenario implementation, clarifying the core position of Token as a value carrier and providing definitive support for the development of the industry. Token “inflation” is the core main line of the AI industry in this round. Non-linear growth on the demand side is driving the revaluation of computing power, large models, and application of the entire industry chain, focusing on leading companies with clear technical barriers and sufficient commercialization verification.

The main views of Guolian Minsheng Securities are as follows:

Under the “inflation” trend of Token, the AI computing power industry is ushering in a new growth cycle where demand, architecture and domestic alternatives resonate

Token “inflation”: The average daily domestic call volume has increased more than a thousand times in two years. Under the gap between supply and demand, the price of computing power leasing and cloud services at home and abroad continues to rise, while C-side tiered payments have been implemented to verify the commercial value of tokens.

Industrial conflicts have moved from insufficient total GPU capacity to release system-level efficiency: in the agent scenario, the high proportion of CPU processing delays is a core bottleneck limiting the release of GPU efficiency. CPUs have jumped from traditional supporting roles to core variables in AI infrastructure, and efficiency improvement technology paths such as supernodes and storage optimization have been implemented simultaneously.

Domestic computing power ushered in a historic development window for modular computing collaboration: the deep adaptation of large models and domestic computing power breaks the dependence of overseas technology paths, and the performance of core manufacturers is expected to maintain rapid growth. At the same time, the Token operation model upgrades computing power trading to intelligent service transactions, and business model restructuring opens up room for long-term growth.

The big model industry is ushering in a three-driven growth phase of business model verification, rapid growth in demand space, and implementation of domestic substitution

Overseas benchmarks have followed the commercialization path of advanced intelligence: Anthropic achieved 10-fold annualized ARR growth, exceeding trillion US dollars in corresponding valuations, and verified the industrial logic that “the upper end of model intelligence directly determines pricing power”. The trend of rising volume and price of high-end models is clear.

AI agents drive the growth in demand for tokens: consumers are shifting from humans to machines, and the upper limit of demand is no longer limited by population size; resident agents, represented by OpenClaw, push model calling from a single round of question-and-answer to long-link multi-tool execution, switching industry value measurement units to tokens. Operators and major Internet companies have launched hierarchical packages one after another, and the business model has been upgraded from computing power trading to intelligent service trading.

The importance of sovereign AI is highlighted: Restricting the export of cutting-edge models highlights the importance of sovereign AI. Large domestic models have entered a new stage of development with local breakthroughs and large-scale verification.

The AI application industry is maturing at an accelerated pace along the path of system construction, vertical implementation, and model innovation

The top-level system design establishes a closed loop of the entire chain from data supply to model development to scenario implementation, clarifying the core position of Token as a value carrier, providing definitive support for industry development; the vertical industry model has moved from technical verification to a closed commercial loop stage, and the cooperation model has moved from traditional project system to value sharing. Leading manufacturers have taken the lead in implementing a subscription business model combined with token pricing, and the replication path is clear; the AI agent-driven native business model has also completed large-scale verification of the first scenario, and the payment scenario has already run through the intelligent capacity to The entire chain of transaction completion is closed, and a basic trust mechanism has been established for the comprehensive development of the smart economy.

Aspect of the target

We recommend leading companies such as Zhi Spectrum (02513), Cambrian (688256.SH), Inspur Information (000977.SZ), Haiguang Information (688041.SH), and Central Control Technology (688777.SH).

Risk Alerts

Technology iteration falls short of expected risks; commercialization falls short of anticipated risks; industry competition increases risks.