Summary of Conference Call Notes Industry Overview - The focus is on the domestic computing power industry chain, which is expected to experience a "high prosperity trend" due to significant changes in supply and demand dynamics [1][2]. Core Insights and Arguments - **Key Turning Point in 2026**: The year 2026 is identified as a pivotal moment for "training + inference" dual resonance, with the continuation of multi-modal scaling laws and the explosive growth of AI applications leading to a steeper-than-expected demand curve [2]. - **Supply Side Improvements**: The supply side benefits from the maturity of domestic chips, the release of wafer production capacity, and the alleviation of overseas chip supply bottlenecks [2]. - **Industry Growth**: This environment is expected to drive sectors such as AIDC (Artificial Intelligence Data Center), cloud and computing services, power equipment, and servers into a cycle of simultaneous volume and price increases, particularly benefiting leading cloud service providers (CSPs) [2]. Market Validation - The company's viewpoints have been progressively validated in the market due to their foresight: - On January 4, a computer annual strategy report was released, indicating a clear AI path for 2026, emphasizing the critical role of computing power and predicting a high expansion rate for domestic computing [3]. - On January 11, an in-depth industry report highlighted the onset of competition for AI entry points and increased capital expenditure (CAPEX) willingness among major companies, signaling an imminent explosion in domestic computing power [3]. - A conference call on January 12 detailed the growth potential and development path of domestic computing [3]. - On January 26, amidst market fluctuations, another call was held to assert the belief in computing power, predicting inflation in domestic cloud services [3]. - On February 8, insights on "AI comic + domestic computing" opportunities were shared, promoting AIDC and cloud service sectors [3]. Related Companies - The following companies are identified as relevant to the discussed trends: Dongyangguang, Cambrian, Haiguang Information, Wangsu Technology, Runze Technology, Xiechuang Data, Huafeng Technology, Shenzhou Digital, Dazhi Technology, Runjian Shares, Kehua Data, SMIC, Huahong Semiconductor, Zhongke Shuguang, Hesheng New Materials, Aofei Data, Youkede, Capital Online, Yunsai Zhili, Ruisheng Intelligent, Inspur Information, Weichai Heavy Industry, and Oulu Tong [3]. Risk Factors - Potential risks include slower-than-expected AI advancement, intensified industry competition, and macroeconomic fluctuations [4].

Market Dynamics - Since the beginning of 2024, A-shares and Hong Kong stocks have alternated in performance, with Hong Kong stocks stabilizing in Q1 driven by the internet sector, followed by new consumption and innovative pharmaceuticals in Q2, leading to a compression of the AH premium to approximately 120 by June 2025 [2] - In July and August, A-shares continued to perform strongly while Hong Kong stocks faced pressure from tightening liquidity and competition in the platform economy [2] Funding Environment - The liquidity situation is improving, with the Hong Kong Monetary Authority passively injecting liquidity in April and May, leading to a temporary drop in HIBOR to near zero; however, by late June, excess liquidity was being withdrawn, and HIBOR rose rapidly to around 4% in August [3] - The Hong Kong dollar has moved away from the 7.85 weak-side guarantee, and the HIBOR-SOFR overnight interest rate spread has returned to a normal range of about 0.36%, indicating that the most stringent phase of the funding environment is likely over [3] Fundamental Outlook - The consensus EPS forecast for the Hang Seng Index for 2025 was revised down from 6.7% in early July to 2.35% by the end of August, primarily due to lowered profit expectations in the platform economy and increased competition in food delivery [4] - However, earnings expectations for sectors such as materials and healthcare within the Hong Kong Stock Connect have been significantly upgraded, and regulatory constraints on unfair competition are expected to reduce price wars in instant retail [4] - With the release of mid-year reports and a shift in outlook for Q4 towards "AI empowerment and efficiency recovery," the internet sector is anticipated to see a rebound in expectations [4] Long-term Framework - The long-term bullish logic for A/H shares is supported by policies and wealth migration, emphasizing a balance between an effective market and proactive government intervention [5] - The dynamic balance aims to stabilize the market while enhancing capital market functions through measures such as mergers and acquisitions, registration system deepening, and attracting long-term capital [5] Structural Changes in Funding - There is a noticeable acceleration in the entry of long-term funds such as social security, insurance, and wealth management into the market, with a clear trend of increased allocation to ETFs and institutional investments [7] - The decline in deposit and wealth management yields has created an "asset shortage" environment, suggesting that both residents and institutions have room to increase their equity allocation [7] Industry and Sector Trends - Emerging sectors such as AI computing chains, semiconductor equipment and materials, military technology, innovative pharmaceuticals, and humanoid robots are advancing from technology to commercialization [8] - This trend is beneficial for platform-based internet companies in AI commercialization as well as for hard technology and its upstream supply [8] External Variables and Capital Inflow - Historically, there is a strong negative correlation between the US dollar index and the Hang Seng Index; if the Federal Reserve enters a rate-cutting cycle in September and the dollar weakens in Q4, the previously high short-selling ratio in Hong Kong stocks may trigger a short-covering rally [9] - The potential for overseas capital to flow back into A/H shares is expected to increase [9] September Outlook - The market may experience fluctuations due to external interest rates and internal expectations, but the tightest phase of the funding environment has passed, and the fundamental narrative of "AI empowerment" is set to unfold [10] - Valuations and risk premiums remain attractive, suggesting that in a "fluctuating-upward" rhythm, sectors such as technology internet (AI), innovative pharmaceuticals, high-dividend stocks, and cyclical leaders with "anti-involution" characteristics are more cost-effective main lines [10] Strategy and Allocation - The strategy focuses on capturing rebound opportunities and the main line of "qualitative change," with a shift from "price wars" to "AI efficiency" in the internet/technology sector [10] - The innovative pharmaceutical sector is viewed positively, with September being a key window for positioning [10] - In the new consumption sector, performance is prioritized, emphasizing differentiation [10] - High-dividend and "anti-involution" sectors are also highlighted, with a focus on selecting companies with stable cash flow and sustainable dividends [10] Valuation Insights - The forecasted PE for the Hang Seng Technology Index is approximately 20.3 times, which is around 30% lower than levels seen since July 2020 [11] - The Hang Seng Index's TTM PE is about 12.3 times, significantly lower than that of the S&P 500, Nikkei, and European stocks [11] - The risk premium of the Hang Seng Index relative to 10-year government bonds is about 6.4%, making it attractive to global capital [11] Core Logic - Following the mid-year reports, the impact of "involution" is weakening, and the narrative for Q4 is shifting towards "AI empowerment," with a focus on commercialization and efficiency [12] - The direction includes AI applications, advertising efficiency improvements, and collaboration in cloud and computing services [12] - The strategy emphasizes holding quality leaders with strong execution capabilities during the concentrated period of academic and medical insurance directory catalysts in Q3 and Q4 [12]