Source: Yi Guan Da Shi. Authors: Fang Yi, Zhang Yifei, Tian Kaixuan, Huang Weichi. Core View: The space for the market correction is largely fulfilled, but objectively, the restoration and stabilization of investor confidence will take time. In the short term, the market remains in a consolidation phase, and a rebound is expected around the year-end and early next year. Favor domestic technology, strategic resources, and high-dividend large-cap financials.
Market Outlook: No Excessively Low Expectations, Year-End and Early-Year Strengthening Expected
Since mid-September, we have discussed that under numerous restrictive factors, short-term stock market expectations are not excessively high, remaining in a weak and modestly volatile state. Looking ahead, we believe the market index adjusted to the 3500-3600 range this week, and the downside is largely fulfilled. Going forward, the market index will not have excessively low expectations. We anticipate that over the next 1-2 quarters, the Chinese market may maintain a prolonged sideways consolidation pattern. From a time perspective, after a sharp decline, objectively, the recovery and stabilization of investment confidence in the Chinese market will take some time. We believe a rebound is likely around the year-end and early next year: 1) Along with the adjustment in the Chinese stock market, market-stabilizing forces are simultaneously strengthening, sending a clear signal of managing downside risks and downside space, preventing local issues from spreading into systemic risk, so the market will not have excessively low expectations; 2) Although microstructural obstacles remain in the Chinese market, unlike 2015/2022, this round does not involve an overall deterioration of market microstructure, but is mainly limited to certain heavily-weighted technology stocks. Since the July pullback in tech stocks, their total market capitalization has significantly declined, and their drag on the index has marginally diminished; 3) Objectively, due to the temporary absence of a major growth narrative, incremental capital has entered a wait-and-see and stagnant phase, and risk assessment, especially uncertainties in international relations, has not substantially decreased. After a short-term bottoming, the market will still be in a phase of consolidation and stabilization. However, related factors will also improve around the year-end and early next year, driving the market to strengthen again.
Valuation Risks in Tech Stocks Have Been Largely Released; Awaiting Microstructure Optimization, the Effectiveness of Value Discovery Is Expected to Gradually Return
After rapid adjustments, the stock price correction in electronics/communications this round has approached the level seen in the ChiNext Index during the same period in 2015. Currently, the ChiNext Index valuation has dropped to a historic low since listing (11th percentile), and electronics/communications valuations have fallen to the 50th percentile since 2010. Considering high profit growth and earnings realization, valuations have entered a mid-term high cost-performance range. As tech valuations and stock prices gradually clear, the effectiveness of value discovery is expected to gradually return. However, given that local micro-trading structure issues persist, investments should balance "healthy micro-trading structure" and "sectors and individual stocks with upward earnings revisions." The dominance of small- and mid-cap thematic directions is expected to continue. Importantly, with U.S. cloud providers' free cash flow turning negative and increasing reliance on debt financing constraints, concerns about the sustainability of the U.S. AI narrative and investment remain. In comparison, China's lower interest rates and more accommodative financial conditions are expected to enable domestic AI development and investment to forge a better and independent narrative.
Industry Comparison: Focus on "Healthy Microstructure" + "Upward Earnings Revisions"
1) Emerging Technology: The AI narrative is expanding from单一的 capital expenditure to model penetration improvement and business model formation. Focus on application-side and cybersecurity; compared to overseas, China's AI in 2027 will have an independent development narrative. Recommendations: AI applications (Hong Kong internet/computer/media), cybersecurity (computer), domestic technology (domestic chips/upstream components/materials), etc. 2) Overseas Market Share Continuously Increasing: Pharmaceuticals/agricultural machinery/feed/select chemicals, etc. 3) Strategic Resources: Under complex situations, demand for resource security increases, and diverse risk-free assets need to be supplemented. Recommendations: precious metals/industrial metals/agriculture, etc.; 4) Stable Dividends: Portfolio stability demand increases, balancing low valuation and stable cash flow. Recommendations: power utilities/telecom operators/brokers/banks/real estate leaders, etc. Thematic Recommendations: 1. Embodied Intelligence: Humanoid robot standards implemented, technology ecosystem improved, and mass production of complete machines accelerating. Favor robot bodies/core components/embodied large models. 2. Data Elements: Data rights confirmation and market-oriented reform advancing, AI driving data value release. Favor high-quality datasets/data governance/public data operations. 3. Food Security: Super El Niño intensifies supply disruptions, stable production and supply policies strengthened. Favor seed industry/smart agricultural machinery/efficient agricultural inputs. 4. Stable Dividends: Bank dividend ratios increasing. Favor high-quality banks/leading brokers. Risk Warnings: Overseas economic recession exceeding expectations; global geopolitical uncertainties.
Main Text
1 Market Outlook: Market Adjustment Nearing an End, Sideways Consolidation for the Market Ahead
Market Outlook: No Excessively Low Expectations, Year-End and Early-Year Strengthening Expected. Since mid-September, we have discussed that under numerous restrictive factors, short-term stock market expectations are not excessively high, remaining in a weak and modestly volatile state. Looking ahead, we believe the market index adjusted to the 3500-3600 range this week, and the downside is largely fulfilled. Going forward, the market index will not have excessively low expectations. We anticipate that over the next 1-2 quarters, the Chinese market may maintain a prolonged sideways consolidation pattern. From a time perspective, after a sharp decline, objectively, the recovery and stabilization of investment confidence in the Chinese market will take some time. We believe a rebound is likely around the year-end and early next year: 1) Along with the adjustment in the Chinese stock market, market-stabilizing forces are simultaneously strengthening, sending a clear signal of managing downside risks and downside space, preventing local issues from spreading into systemic risk, so the market will not have excessively low expectations; 2) Although microstructural obstacles remain in the Chinese market, unlike 2015/2022, this round does not involve an overall deterioration of market microstructure, but is mainly limited to certain heavily-weighted technology stocks. Since the July pullback in tech stocks, their total market capitalization has significantly declined, and their drag on the index has marginally diminished; 3) Objectively, due to the temporary absence of a major growth narrative, incremental capital has entered a wait-and-see and stagnant phase, and risk assessment, especially uncertainties in international relations, has not substantially decreased. After a short-term bottoming, the market will still be in a phase of consolidation and stabilization. However, related factors will also improve around the year-end and early next year, driving the market to strengthen again.
2 Valuation Risks in Tech Stocks Have Been Largely Released; Awaiting Microstructure Optimization, the Effectiveness of Value Discovery Is Expected to Gradually Return
Valuation Risks in Tech Stocks Have Been Largely Released; Awaiting Microstructure Optimization, the Effectiveness of Value Discovery Is Expected to Gradually Return. After rapid adjustments, the stock price correction in electronics/communications this round has approached the level seen in the ChiNext Index during the same period in 2015. Currently, the ChiNext Index valuation has dropped to a historic low since listing (11th percentile), and electronics/communications valuations have fallen to the 50th percentile since 2010. Considering high profit growth and earnings realization, valuations have entered a mid-term high cost-performance range. As tech valuations and stock prices gradually clear, the effectiveness of value discovery is expected to gradually return. However, given that local micro-trading structure issues persist, investments should balance "healthy micro-trading structure" and "sectors and individual stocks with upward earnings revisions." The dominance of small- and mid-cap thematic directions is expected to continue. Importantly, with U.S. cloud providers' free cash flow turning negative and increasing reliance on debt financing constraints, concerns about the sustainability of the U.S. AI narrative and investment remain. In comparison, China's lower interest rates and more accommodative financial conditions are expected to enable domestic AI development and investment to forge a better and independent narrative.
3 Industry Comparison: Balanced Industry Allocation, Focus on Sectors and Stocks with "Healthy Microstructure" + "Upward Earnings Revisions"
Industry Comparison: Balanced Industry Allocation, Focus on Sectors and Stocks with "Healthy Microstructure" + "Upward Earnings Revisions." 1) Emerging Technology: Overseas AI capital expenditure's second derivative temporarily slows, but downstream enterprise AI penetration increases, and industry chain profits concentrate toward the software end. Tech sanctions and trade disputes increase, raising the necessity of self-controllability. China's AI has an independent development narrative. Recommendations: Hong Kong internet/computer/media/domestic chips/upstream equipment materials/pharmaceuticals, etc. 2) Strategic Resources: The world needs supplementation of risk-free assets. Under complex international situations, demand for resource security increases. Geopolitical conflicts and AI construction needs will also tighten supply-demand for bulk commodities. Recommendations: non-ferrous metals/agriculture/grid equipment, etc.; 3) Stable Dividends: Portfolio stability demand increases. Domestic risk-free yield declines, improving the cost-performance of dividend assets. Structurally, balance low valuation and stable cash flow. Recommendations: power/telecom operators/real estate leaders/food & beverage/brokers/banks, etc.
AI Credit Risk Does Not Change the Overall Industry Trend, but AI Hardware Rally Faces Significant Challenges. The AI financialization characteristics we previously highlighted are being recognized by the market and evolving into concerns about whether AI financing will default or face circular breaks. We believe the market is currently overly concerned. Neocloud financing risk remains a local issue without a basis for contagion, but it will indeed somewhat affect the pace of upward revision of the industry's second derivative expectations, thereby suppressing hardware price increase expectations. We previously proposed that this AI industry wave has entered a phase of industry financialization. These industrial financial innovations drive rapid growth in AI industry demand from small and medium players. Nvidia's revenue structure shows that since 2026, the marginal contribution to industry demand has shifted from giant investments to new players like NeoCloud, industrial, and sovereign AI. However, recently, Neocloud bonds represented by Coreweave have seen significantly rising yields. Against the backdrop of Anthropic's slowing revenue expectations, the market further worries whether AI project returns can support project cash flow cycles under high U.S. Treasury yields. Coreweave's leverage level continues to rise in 2026, with significantly increased use of off-balance-sheet financing tools. Its DDTL model uses future cloud provider contract cash flows as collateral for financing, so changes in its bond yields amplify market concerns about future AI project cash flow expectations. Compared to the market, we are more optimistic about the AI industry trend but more cautious about the hardware rally: 1) The rise in Neocloud bond yields has not driven up overall AI bond yields. Giants like Google have seen bond yields recently decline along with overall U.S. Treasuries, meaning the bond market is not worried about AI's overall return rate, but about the project returns of the Neocloud + AI Lab combination. This mainly affects the pace of new investment in the AI industry, i.e., the second derivative; 2) A key link in credit risk evolving into a financial crisis is a sharp weakening in collateral prices. The core computing power rental price remains strong. September H100 rental prices continued to rise, meaning AI investment returns remain substantial; 3) However, in Q3, the semiconductor inventory-to-sales ratio in South Korea and Taiwan has shown a slowing trend at high levels. Considering that since Q2, AI capital expenditure has mainly been contributed by interest-rate-sensitive investment entities, high interest rates will significantly affect the second derivative of semiconductor demand, thereby reinforcing expectations of a slowing inventory-to-sales ratio. For cyclical goods, a slowing inventory-to-sales ratio means price increases are difficult to sustain and expectations are hard to further revise upward. The probability of stock price peaks significantly increases.
Within AI Investment, Focus on the Migration of Value from Hardware to Software Within the Industry Chain; Domestic Computing Power Investment Has Stronger Independence. Currently, value within the AI industry chain is shifting toward software. Enterprise AI is entering a commercialization realization cycle of double-click growth in penetration and ARPU. This is driven by both technological iteration and business model changes. In the early AI industry chain, closed-source model vendors, with performance leadership and training capital expenditure, drove hardware segments to become the most direct beneficiaries. However, with the spread of open-source ecosystems, declining model prices, narrowing capability gaps, and continuously falling inference costs, the bargaining power of merely owning models or hardware has marginally weakened. New value is shifting toward the application implementation layer closest to AI commercialization. From a software business model perspective, the industry know-how, private data, customer trust, and delivery and compliance responsibilities accumulated by vertical software constitute clear moats. Enterprises are therefore more willing to pay for compliant, secure task completion rather than merely for token consumption. From a technology iteration perspective, the beginning of the RSI era means AI begins to participate in improving the ability to develop AI itself, so the value importance of the software control layer has significantly increased. Core competition among models is no longer single parameter scale, but system capabilities at the software control layer. In specific enterprise scenario implementation, large models represented by GPT-6 have greatly improved task completion capabilities. However, stronger models do not completely replace software. Instead, they increase demand for middle-layer software, especially governance software related to internal information security and process compliance, and platform software related to workflow and data management. These industries with prominent information-intensive and security-sensitive characteristics are also the fastest directions for enterprise AI penetration improvement. Recommendations: vertical applications and Hong Kong internet leaders/network and information security/data platform software.
On the other hand, importance should be attached to the certainty advantage of domestic computing power growth. On October 16, the General Secretary pointed out during an inspection in Ulanqab: "Implementing the 'East Data West Computing' project and building a national integrated computing power network is an important deployment of the Party Central Committee. We must strengthen scientific planning, element support, and security guarantees to ensure the efficient operation of completed projects and complete subsequent project construction as scheduled." As one of the current "six networks," the computing power network has a clear policy positioning. Since Q3, both fiscal and monetary levels have increased support for the "six networks." We recommend increasing allocation to the domestic computing power chain benefiting from the enhancement of domestic AI capabilities and policy efforts, as well as hardware and material segments with tight supply-demand.
Strategic Resources: Crude Oil Volatility Center Rising, Emerging Demand Share in Bulk Commodity Demand Structure Increasing, Core Resource Leaders Expected to Revalue. The market generally worries about the sustainability of resource price increases, with logic that: 1) Resource price increases are often driven by geopolitical events, so the stability of a high price center is constrained by geopolitical uncertainty; 2) Global non-AI industry prosperity remains weak, suppressing resource demand. But we find: 1) In addition to the rising trend of global geopolitical risk and economic policy uncertainty, developed countries' strategic buffer capacity against crude oil shocks is rapidly declining. In 2026, the U.S. and OECD strategic petroleum reserve scales will significantly decrease, both reaching the lowest levels of the 21st century. The oil price volatility center also marginally rises as developed country crude oil reserves decline. 2) Emerging demand from the AI industry is dominating resource pricing. Since 2023, copper price increases have gradually broken free from high real interest rate suppression. The core reason is that AI industry demand for copper foil and other metal materials is profoundly changing the downstream demand structure of non-ferrous metals and other resource industries. 3) Domestic resource capacity supply in 2027 will continue to be tight. In 2026, new capacity supply (measured by capital expenditure growth and construction-in-progress growth) in A-share mid-to-upstream cyclical industries (coal, steel, chemicals, building materials), agriculture, and pan-domestic consumption industries generally shows a significant decline. Therefore, we expect that in 2027 and even the next 2-3 years, resource capacity supply will continue to be tight or even in shortage. In summary, contrary to market consensus, we believe the strategic value center of bulk commodities still has strong upward momentum, and core resource leader stocks are expected to revalue.
Dividends and Stable Cash Flow: Still an Important Base Position for the Next Stage, but Internal Structure Should Balance Low Valuation and Stable Cash Flow. From a cyclical perspective, during stagflation phases when earnings expectation revisions are blocked and inflation expectations are high, dividend assets typically have relative return advantages. But compared to historical cycles, an important difference now is the structural divergence within dividends. Traditional dividend industries like finance and coal have seen significant declines in dividend yield advantages after rising. Comparing the Dividend Index (high weights in banks, coal, and transport) with the CSI 300 Cash Flow Index (high proportion of consumption, utilities, resources, and manufacturing leaders), the 300 Cash Flow Index dividend yield has significantly improved since August, while the Dividend Index dividend yield overall declined in Q3, and its P/E has reached the 99th percentile of the past decade, no longer having a low valuation advantage. On the other hand, the low-valuation, cash-flow Value 100 Index dividend yield is also significantly higher than the Dividend Index and marginally increasing. This means the next stage's dividend strategy should focus on internal structure. Utilities and some consumption and manufacturing leaders also have allocation opportunities. Historical experience shows that during stagflation phases when dividend strategies prevail, besides finance, power, real estate, food & beverage, telecom operators, rail transit equipment, etc., also have relative advantages. Some traditional industry leaders, after long-term capacity and inventory clearance, may see expectations stabilize in 2027. These tracks are also currently less crowded for institutional allocation. Combined with current industry dividend yields, recommend focusing on: power/telecom operators/white goods/mass consumer goods/commercial vehicles/real estate leaders, etc. In the financial sector, continue to recommend brokers/banks.
4 Thematic Recommendations: Embodied Intelligence/Data Elements/Food Security/Stable Dividends
1. Embodied Intelligence: Policy standards and technology ecosystem improvement, domestic and foreign manufacturers accelerating mass production layout, scenario implementation expected to drive industry chain development. Favor robot bodies and core component industry chains. Investment Advice: Policy standards, technology platforms, and complete machine mass production advancing in synergy, embodied intelligence industrialization expected to accelerate. The first batch of humanoid robot national standards released, covering key links such as perception, decision-making, control, and operation, providing a unified basis for R&D testing and large-scale application. The Ministry of Commerce and other nine departments promote embodied intelligence expansion into commercial logistics, home services, and other scenarios. Nvidia upgraded the Isaac robot development platform, improving model training and deployment ecosystems; Zhiyuan open-sourced a real-scenario embodied reasoning dataset, helping improve robots' complex task execution capabilities. Tesla's Optimus started production and entered the capacity ramp-up phase, with most equipment for a million-unit production line in place. Unitree disclosed cumulative G1 production of approximately 10,000 to 20,000 units and launched the "Yuchuang Plan," promoting coordinated development of the industry chain including robot bodies, core components, and algorithm applications. Recommendations: 1. Industrial-scale delivery driving component demand growth. Favor screws/reducers/dexterous hands/sensors; 2. Model iteration and application scenario expansion. Favor embodied large models/edge AI chips/data training platforms; 3. Domestic manufacturers leading in shipment scale. Favor leading robot body manufacturers with software-hardware synergy and commercialization capabilities.
2. Data Elements: Market-oriented allocation reform continues to advance, AI application deepening and data procurement mechanism improvement accelerate value release. Favor data supply, governance, and circulation operation industry chains. Investment Advice: Data element market-oriented allocation reform continues to advance, with data rights confirmation, circulation, and application systems accelerating improvement. The first batch of data property rights registration certificates has been gradually issued, covering more than ten industries including healthcare and transportation, providing support for data trading, financing guarantees, and equity contribution. Beijing released the "15th Five-Year" digital intelligence development plan, proposing that by 2030 the data industry scale reach over 2.24 trillion yuan, and building 30 national-level and 50 municipal-level trusted data space benchmarks. The National Data Administration held the "Data Elements ×" competition national finals, with 1087 projects shortlisted, promoting demonstration and promotion of industry data application achievements. The National Development and Reform Commission introduced the first batch of 56 comprehensive reform pilot innovative measures for market-oriented allocation of factors. Henan explored public data authorized operation and data asset inclusion in balance sheets. Zhengzhou achieved cumulative data transaction amount exceeding 5.5 billion yuan. The National Data Group was officially established, focusing on data resource development and utilization, high-quality dataset construction, and data infrastructure interconnection, promoting cross-departmental and cross-industry data circulation. Recommendations: 1. Data collection and annotation/high-quality datasets/corpora benefiting from AI data demand growth; 2. Databases/data governance/data middle platforms benefiting from enterprise data development and utilization; 3. Public data operations/trusted data spaces/data security benefiting from data element marketization.
3. Food Security: Policy support plus super El Niño disruption, catalyzing agricultural technology upgrading and stable production and supply demand. Favor biological breeding/smart agricultural machinery/efficient agricultural inputs/grain storage equipment. Investment Advice: Super El Niño intensifies global agricultural production uncertainty. Policy support and disaster prevention and mitigation deployment strengthen stable production and supply expectations. The Agricultural Development Bank of China pre-arranged 200 billion yuan in special credit funds to guarantee autumn grain procurement and grain industry chain funding needs. The National Food and Strategic Reserves Administration promotes whole grain industry development and grain saving and loss reduction, accelerating key technology R&D and grain and oil processing equipment upgrades. The World Meteorological Organization expects this El Niño to peak by end-2026, with the Niño 3.4 sea surface temperature anomaly from November 2026 to January 2027 possibly reaching +3.8°C, exposing some global agricultural production areas to drought, flood, and high temperature risks. The Ministry of Agriculture and Rural Affairs Planting Management Department, etc., deployed cross-seasonal disaster prevention and mitigation measures, promoting drought-resistant and water-saving varieties, water-saving irrigation, "one spray three prevention," and agricultural machinery rapid harvesting and drying technology. Demand for stress-resistant seed industry and efficient agricultural inputs is expected to increase. Recommendations: 1. Seed industry leaders benefiting from seed industry revitalization and biological breeding industrialization; 2. Smart agricultural machinery/agricultural input chains benefiting from agricultural modernization and grain yield improvement; 3. Drying/storage equipment benefiting from grain procurement and storage and grain saving and loss reduction.
4. Stable Dividends: Shareholder return policy strengthening, bank dividend ratios increasing, long-term capital entry and dividend ETF issuance forming catalysts. Favor high-quality joint-stock banks/city commercial banks/state-owned large banks and leading brokers. Investment Advice: Policy support, dividend realization, and investment tool expansion are expected to jointly enhance the allocation attractiveness of dividend assets. The CSRC emphasizes improving listed company shareholder returns, perfecting long-term investment assessment mechanisms, and broadening channels for medium- and long-term capital entry; the central bank proposed establishing non-bank institution liquidity support tools and improving financial market stability mechanisms. Bank dividend intensity increased. The six major banks collectively raised interim dividend ratios to 31% for the first time, with total proposed dividends of approximately 221 billion yuan, further enhancing the stability and predictability of bank shareholder returns. The first batch of 18 active ETFs was collectively issued, among which three dividend-themed products use the CSI Dividend Index as performance benchmark, enriching dividend asset allocation tools. Recommendations: 1. Joint-stock banks/city commercial banks/state-owned large banks benefiting from insurance capital allocation demand, dividend increases, and earnings recovery; 2. Leading brokers benefiting from medium- and long-term capital entry and increased capital market activity.
5 Risk Warnings
Overseas economic recession exceeding expectations. If the global economy slows, it may transmit through external demand to domestic demand, causing economic expectations to decline. Global geopolitical uncertainty. If the Iran-U.S. situation subsequently turns into a long-term full-scale war, it will suppress global risk appetite.
Comments