2024年至今，市场对AI的期待一直很高，因其应用落地国内外均低于预期，历经这一曲折，市场对AI的预期，已由大模型、大模型应用落地，直接切向了各计算平台、硬件云或端侧的真实落地推进，即市场不再偏好用哪个大模型(自有或外部)、大模型的参数迭代、应用方的资源禀赋(AI公司、大模型厂商、众多行业龙头)，而是更偏好真实推进的利润、收入、市占率/用户数等强检验指标。The market's expectations for AI have shifted from the implementation of large models and large model applications, directly targeting the real landing and promotion of various computing platforms, hardware clouds, or end-side.
According to the smart financial APP, Guotou Securities released a research report stating that the market's expectations for AI have shifted from the large model to the landing and promotion of various computing platforms. The essence of AIAgent is to perform the 'execution' function in the landing scenarios and applications in the digital world, manifested as machines independently completing tasks based on specific goals and self-iterating. The areas with fewer and smaller screens are most in need of AIAgent. AIAgent, as a new round of technological innovation, can achieve cost reduction and efficiency improvement by participating in marketing, management, production, and other major processes, reshaping organizational structure and rebuilding competitive advantages in the AI era. Successful iterations are expected to hedge against the downward valuation and bring commercial model and valuation improvements to existing companies.
The market's expectations for AI have shifted from the large model to the landing and promotion of various computing platforms.
From 2024 to the present, the market's expectations for AI have been high, as its applications both domestically and internationally have fallen below expectations. After this twist and turn, the market's expectations for AI have shifted from the implementation of large models and large model applications, directly targeting the real landing and promotion of various computing platforms, hardware clouds, or end-side. The market no longer favors which large model (internal or external) to use, large model parameter iterations, or application resources endowment (AI companies, large model manufacturers, and many industry leaders), but rather favors strong verification indicators of real promotion, such as profit, income, market share/users, etc.
The essence of AIAgent is to perform the 'execution' function in the landing scenarios and applications in the digital world.
Depending on the different scenarios of execution, it is divided into two paths: the physical world and the digital world, deducing that the ultimate form of AI may be AIAgent and embodied intelligence. The bank believes the essence of AIAgent is to perform the 'execution' function in the landing scenarios and applications in the digital world, manifested as machines independently completing tasks based on specific goals and self-iterating. Products that have already appeared include Auto-GPT, BabyAGI, AgentGPT, MicrosoftJarvis, ChaosGPT, etc., but AIAgent cannot directly execute in the physical world; it acts more as a human assistant, thus requiring interaction with people, and needs to work with smartphones, MR, AIPiN, and other smart terminals to display information and accept commands.
The areas with fewer and smaller screens are most in need of AIAgent.
The bank believes that as a personal assistant, the most basic function of AIAgent is to interact with humans, understand intentions, and complete execution on the ground. The difference between it and traditional software is that it does not require human intervention, so it does not rely on large screens, mice, keyboards, and other interaction devices at the interaction level. Therefore, in many scenarios, the less and smaller the demand for interaction devices such as screens, the higher the dependence on AIAgent, and it is expected that its progress speed will be faster.
As a new round of technological innovation, AIAgent is expected to bring improvements in business models and valuations to existing companies.
For companies like SenseTime and Fourth Paradigm, leveraging AIAgent can smoothly expand or switch from traditional B2B business models to B2C, expanding business scope, optimizing financial statements (increasing gross margins); for large model manufacturers, leveraging AIAgent, they can expand their business scope to various ends, synchronously exploring the practical applications of large models in the virtual and physical worlds; for various industries, especially leading companies, leveraging AIAgent can reduce costs and improve efficiency, reshape organizational structures, and rebuild competitive advantages in the AI era. Successful iterations can hedge against downgrades in valuations.
Investment advice: AI high-elasticity targets are expected to recover in the fourth quarter. Key focuses: Fourth Paradigm (06682), SenseTime-W (00020), Mango Excellent Media (300413.SZ), China Film Co., Ltd. (600977.SH), Guangdong Advertising Group (002400.SZ), Time Publishing and Media (600551.SH), Anhui Xinhua Media (601801.SH), Jiangsu Phoenix Publishing&Media Corporation (601928.SH), Beijing Jetsen Technology (300182.SZ), Offcn Education Technology (002607.SZ), Guomai Culture (301052.SZ), wasu media holding (000156.SZ), China Publishing & Media Holdings (601949.SH), etc.
Risk warning: Changes in underlying logic, operational management governance, and volatile market preferences.
