Zhongjin: The narrative of the AI bubble being neutral, actively positioning in the industry in the short to medium term to seize opportunities with lower expectations.

CICC released a research report stating that the CAPEX guidance of major overseas technology giants in 2026 has been significantly raised, approaching or exceeding their operating cash flow, and the pressure of depreciation and amortization on cloud profit margins is gradually accumulating. However, with the acceleration of revenue growth, self-developed GPU chips, and price increases of cloud products, it is expected to improve the ROI of cloud investment. Currently, AI is driving the accelerated growth of cloud business in terms of orders and revenue year-on-year. In the short to medium term, as long as the main business operations of giants like Meta and Google do not undergo significant changes, their balance sheets remain healthy and the sustainability of their capital expenditures is not a short-term concern. Faced with a relatively complex situation, one can prioritize seizing undervalued core businesses and opportunities with expected differences in AI layout. CICC's main points are as follows: AI investment: The CAPEX guidance of major overseas technology giants in 2026 has been significantly raised, approaching or exceeding their operating cash flow, and the pressure of depreciation and amortization on cloud profit margins is gradually accumulating. However, positive factors are also seen: 1) Currently, the financing channels of major overseas technology giants are smooth, with Meta, Google, Amazon all conducting financing of several hundred billion US dollars, and the financing rates are low, with high investor acceptance of bond issues; 2) Cloud vendors' revenue growth, self-developed GPU chips, and price increases of cloud products are expected to improve the ROI of cloud investments. There is still a 1-2 year window for iterative model capabilities, exploration of AI commercialization realization, and the development of value-added services for AI cloud, thereby enhancing the visibility of investment returns. AI benefits: 1) Model layer: Since 4Q25, overseas models have improved in agent execution and control capabilities, multimodal capabilities, and long-text capabilities, with agents gradually penetrating cloud and core company businesses; 2) Business layer: AI continues to improve the logic of advertising and e-commerce, driving the improvement of advertising volume and price and GMV growth of e-commerce; 3) Cloud computing: AI drives the year-on-year acceleration of orders and revenue growth in cloud business. The narrative of the AI bubble theory is still neutral: AI technological advancements are indeed creating incremental revenue and user value in multiple dimensions such as programming, chatbots, traditional business efficiency improvement, and cloud computing. However, the competition-driven, prisoner's dilemma-style Capex investment of US tech giants does pose a long-term risk of overinvestment. However, in the short to medium term, as long as the main business operations of giants like Meta and Google do not undergo significant changes, their balance sheets remain healthy and the sustainability of their capital expenditures is not a short-term concern. Faced with a relatively complex situation, one can prioritize seizing undervalued core businesses and opportunities with expected differences in AI layout. Investment recommendations: It is recommended to focus on Meta (META.US) with greater expected differences, stable barriers in social business, faster progress in AI models and other fields by 2026, and potential for reduced losses in VR business; recommend Amazon.com, Inc. (AMZN.US), with a solid retail base and continuous release of profit margins, with potential to catch up in AI ecology and AWS in the medium to long term; Google (GOOGL.US) is recommended in the medium to long term, with a comprehensive AI layout in hardware, models, cloud, applications, and other fields blossoming, but short-term attention may be needed for risks of overheated sentiment. Risk factors: Global macroeconomic uncertainties; political risks of GEO Group Inc; AI business development falling short of expectations; regulatory factors.