Market trends from late September to early October are expected to see subdued trading activity, influenced by the Mid-Autumn Festival and National Day holidays, barring any unexpected macro or industry developments. October will likely center on Q3 earnings season, with the AI supply chain's high prosperity expected to continue. November may shift toward thematic trading as Q3 reporting closes and institutions enter their year-end ranking period. December calls for positioning in the year-end rally based on policy signals from the Central Political Bureau and Central Economic Work Conference meetings.
Market structure in Q4 is likely to feature more alpha opportunities in individual stocks or sub-sectors, significantly increasing trading difficulty compared to the first half. Key upside risks include a "924-style" policy shift domestically and breakthrough progress in the AI industry chain. Downside risks include an unexpected escalation of overseas geopolitical conflicts, tighter-than-expected global liquidity, rising global recession expectations, a reversal in AI industry trends, systemic risks in US equities, and underwhelming domestic growth-support policies.
During Q3, the global AI industry chain experienced significant volatility. As of September 17's close, all major A-share indices posted losses for the quarter. AI upstream hardware sectors such as electronics, communications, and building materials corrected notably, while financials, consumer, and stable sectors outperformed growth styles. Looking toward Q4, can the AI industry chain, supported by strong fundamentals, make a comeback? If not, will market style shift, and will pro-cyclical sectors have an opportunity for a cyclical recovery?
The core view is that policymakers are showing increasing acceptance of economic growth reaching the lower end of the Two Sessions target range. In Q4, policymakers are likely to continue the policy tone set at the late-July Political Bureau meeting, accelerating the implementation of existing policies while selectively adding targeted incremental measures to support the economy. With the Fed resuming rate hikes at its September meeting, greater attention in Q4 should be paid to global geopolitical developments, overseas macro liquidity conditions (especially the possibility of consecutive Fed hikes), and trends in global growth industries, particularly the AI chain. For the A-share market, the continuation of a K-shaped divergence in macroeconomics and mid-level corporate profits suggests a lack of momentum for trendline index gains.
In terms of industry trends, whether the AI application segment achieves breakthrough progress, whether CSP giants' massive capital expenditures can be sustained after Fed rate hikes, and when computing power investments can truly convert into profits remain the most critical factors affecting AI industry chain investment. Q3 Review: High Congestion Triggers Sharp Correction in AI Hardware The global AI industry chain saw intense volatility in Q3. The K-shaped divergence in the global economy in H1 led funds to cluster in AI hardware. However, the contradiction between rising capital expenditures and business models that haven't yet closed the loop, combined with high congestion and leverage, triggered a deep correction in global AI infrastructure trading in July. As Microsoft, Amazon, and other CSP giants reported better-than-expected results, the global AI chain rebounded to varying degrees in August. During this rebound, the Nasdaq outperformed the Philadelphia Semiconductor Index, indicating the trading logic based on BOM price increases had shifted.
Entering September, global stock markets moved sideways amid geopolitical instability, rising long-end US Treasury yields, and heightened expectations of Fed rate hikes. As noted in our semi-annual report, the comprehensive rise in BOM caused a notable divergence between upstream hardware stocks benefiting from AI infrastructure and downstream CSP giants. While CSP giants continue to increase capex, they face both rising expenses and potential token price wars, with revenues struggling to cover costs and ROIC uncertainty rising. The fundamental divergence led global AI trades to cluster in upstream hardware, creating high congestion and leverage in some markets like Korea. Events such as Apple's price increases and class-action lawsuits against Samsung, SK Hynix, and Micron in the US made traders realize the combined price hikes across upstream materials were far beyond midstream and downstream capacity to absorb, contradicting the long-term trend of declining token prices. This sparked concerns that the semiconductor industry's aggressive investment cycle was becoming increasingly difficult to justify.
Consequently, July saw a correction of the overextended pricing chain trades globally, with high congestion and leverage causing sharp selloffs in AI hardware stocks represented by memory. August brought a technical rebound, but indices like Korea's KOSPI, the Philadelphia Semiconductor Index, and A-share ChiNext and STAR 50 failed to recover July losses. Despite macro disruptions like rising long-end yields and geopolitical uncertainty, the essence is that AI development remains on the left side of the J-curve. With business models not yet closed, growing capex pressures CSP giants' cash flows, forcing external financing. Examples include BlackRock's Meta data center bond issuance with a 7.53% yield and only 1.6x subscription coverage, Blackstone's QTS Realty Trust issuing near-junk-level debt for Microsoft's data center, Alibaba's HK$80 billion share placement for AI investments, and Zhipu's completion of a $5 billion financing round. These instances show the market remains skeptical whether such massive capital investments can yield sufficient returns before new breakthroughs in the AI chain.
By September, A-shares underperformed overseas markets. While external factors like geopolitical tensions and central bank rate hikes contributed, domestic factors were the root cause of reduced risk appetite. H2 domestic macro policy has focused on support without stimulus, accelerating implementation of existing measures, making a repeat of the "924" and "926" moments in 2024 difficult. At the capital market level, ETF inflows were not significant after mid-July announcements of A-share holdings increases, while IPOs, lock-up expirations, and major shareholder reductions affected micro-liquidity. These internal factors led to marginal increases in external influence on A-shares, reinforcing the market's tendency to follow declines rather than gains. As of September 17's close, all major A-share indices were lower for Q3.
Trend Outlook: Indices Lack Momentum for Trendline Rises From the latest policy signals, policymakers show increasing acceptance of economic growth reaching the lower end of the target range. Q4 policy will likely continue the tone from the late-July Political Bureau meeting, accelerating existing policy implementation while adding targeted incremental measures. The policy focus remains on supply-side issues like technological self-reliance, manufacturing self-sufficiency, and energy security, rather than broad-based demand-side stimulus. If geopolitical tensions remain unresolved and oil prices stay elevated, prompting further monetary tightening, China's export growth could face challenges, warranting attention to domestic demand countermeasures.
Given these conditions, the K-shaped divergence in earnings between emerging and traditional economies is expected to persist in Q4. For A-shares, this means a lack of momentum for trendline index gains. With the Fed resuming rate hikes in September, Q4 requires heightened awareness of geopolitical evolution, overseas liquidity conditions, and global growth industry trends, especially AI. Trading is expected to be subdued from late September to early October due to holidays, unless unexpected macro or industry developments occur. October will focus on Q3 earnings season, with AI chain prosperity expected to continue. November may shift to thematic trading as institutions enter ranking periods. December positioning for year-end rallies will depend on policy signals from the Central Political Bureau and Central Economic Work Conference.
Macroeconomy: Policy Fine-Tuning for Support, Full-Scale Shift Unlikely Despite increased domestic growth pressure in H2, the July Political Bureau meeting signaled focus on accelerating existing policy implementation, with emphasis on supply-side initiatives like technological self-reliance, manufacturing upgrades, and industrial chain security, while providing support without stimulus to domestic demand and consumption. A series of People's Daily articles from August 22-25 reinforced this message, suggesting policymakers acknowledge growth stability pressures but don't feel urgent about them. The H1 GDP growth rate of 4.7% is considered consistent with current factor supply, technological progress, and institutional innovation conditions. Risk resolution for real estate, local debt, and small financial institutions may create contractionary effects and growth costs, but these short-term costs are viewed as trade-offs for long-term stability and sustainable development.
The policy stance suggests a high threshold for full policy shifts, with current thinking focused on converting existing fiscal resources into tangible work faster. As long as growth remains in a reasonable range, policymakers are willing to accept certain growth costs in exchange for further clearing of local debt, real estate, and financial risks. The late-August release of a notice on improving commercial housing sales systems, promoting a shift from presale to completed-property sales, will have mixed effects. While it reduces buyer risks and may aid inventory clearance and price stabilization, it will slow developers' capital turnover, reduce land purchases and new construction, and potentially pressure local fiscal revenues and employment. The policy framework is set, with actual impact dependent on local implementation details.
Policies tightening social security contribution base verification, offshore income taxation for mid-to-high income groups, and retrospective reviews of high-tech enterprise tax benefits may add pro-cyclical pressure on the real economy. While Q2 GDP growth of 4.3% fell below the Two Sessions target range of 4.5%-5%, policymakers consider this acceptable, maintaining a fine-tuning approach rather than a comprehensive "924-style" policy shift. For Q4, expect continued implementation of existing policies with selective incremental measures like accelerating construction in new power grids, computing networks, and urban underground pipelines. The focus remains on supply-side policies rather than broad demand-side stimulus. Global macro liquidity has also been tightening, with the ECB and Bank of Japan raising rates, and the Fed's September 17 rate hike. The Fed chair's hawkish stance raised concerns about a new rate hike cycle potentially triggering recession. If geopolitical tensions persist and oil prices stay elevated, further monetary tightening could raise global recession expectations, challenging China's export growth and requiring attention to domestic demand offsets.
Corporate Earnings: K-Shaped Divergence Persists Between Traditional and Emerging Sectors From interim report data, all-A-share net profit attributable to shareholders grew 19.4% cumulatively in H1, up 10.1% from Q1, while non-financial A-shares grew 21.52%, up 5.71%. By Shenwan primary industry classification, electronics, non-ferrous metals, and non-bank financials performed strongly due to AI prosperity, supply-demand imbalances, and capital market conditions, while pro-cyclical sectors tied to traditional economy generally underperformed. Non-bank financials, electronics, and steel showed the most improvement from Q1, while military, agriculture, and computer sectors declined. Structural factors including domestic AI investment, global energy transition, and Chinese companies' overseas expansion suggest the earnings divergence between emerging and traditional economies will continue in H2.
Despite noticeable earnings improvements, particularly for the STAR 50, after Q3's global AI chain volatility and domestic non-market disruptions, trader risk appetite has declined notably, with index valuation premiums disappearing. The equity risk premium for all A-shares sits above the decade average. While dividend yields retain advantages over bonds in an "asset shortage" environment, full recovery of risk appetite requires policy shifts or new industrial chain breakthroughs.
Structural Outlook: Alpha Phase for Individual Stocks as Drivers Weaken After Q3's dramatic swings, traders are asking whether the AI chain can rebound in Q4 or whether style shifts toward pro-cyclical rebounds will occur. Unless application-side breakthroughs happen, AI infrastructure is unlikely to repeat H1's comprehensive beta rally. Combined with the support-without-stimulus policy stance preventing convergence of fundamental divergence between emerging and traditional economies, Q4 is expected to feature more alpha opportunities in individual stocks or sub-sectors, with trading difficulty substantially higher than H1.
AI Industry Chain: Broad Hardware Rally Over, Awaiting Application-Side Progress From the four major US CSP giants' earnings, AI remains on the left side of the J-curve. While AI cloud businesses outperformed expectations, capex revisions were primarily due to BOM price increases, especially memory costs. Free cash flow missed across three of four CSP giants, with Microsoft facing significant FCF challenges after accounting changes. With cash flow pressure, CSP giants rely on external financing, facing higher costs after September rate hikes. If rate hikes trigger recession expectations, CSP traditional business cash flows would also face pressure, impacting AI capex capacity and computing power demand. NVIDIA's late-August earnings beat expectations, but its role as AI chain credit provider shows growing risk exposure as revenues rise.
Traders are watching whether AI applications can replace coding as the new narrative, reenacting the Anthropic Claude moment, or whether a model's capability and pricing can successfully penetrate multiple industries. Given the declining token price trend, AI hardware is unlikely to repeat H1's broad beta rally without application-side breakthroughs, instead showing more individual stock or sub-sector alpha. The "shovel-selling" trade based on BOM price increases ultimately converts into concerns about CSP cash flow deterioration and forced capex reductions. Memory prices remain high despite supply-demand imbalances, difficult for both consumer electronics and the AI chain itself to absorb. Catalysts for memory stocks have shifted from earnings revisions driven by DRAM and NAND price increases to LTAs, buybacks, and broader capital returns. Recent calls by Anthropic's CEO for AI companies to slow frontier model development, supported by other industry leaders, mark notable industry sentiment shifts. For Q4, application-side breakthroughs, CSP capex sustainability after Fed hikes, and timing of computing power investment profit conversion remain critical factors.
Pro-Cyclical Sectors: Limited Macro Stimulus, Style Shift Awaiting Timing The current policy stance of support without stimulus means large-scale style shifts require better timing. Interim report data shows traditional consumer sectors continue to underperform due to macro-level K-shaped divergence. While non-bank financials and real estate improved from Q1, sustainability faces pressure in H2. The late-August housing sales system reform notice significantly changes real estate development business models. Under the new completed-property sales system, developers must self-finance land and construction costs until near-completion, extending cash turnover cycles from approximately 2 years to 3-4 years. This reduces funds available for land purchases, subsequently reducing future saleable resources, shrinking overall industry supportable sales, land purchases, new construction, and investment scales. Unless significant home price increases or land price declines emerge, developers' ROE face pressure from lower asset turnover.
However, the policy's impact on home prices leans positive. Reduced delivery risk improves new home demand, while reduced land purchases and new construction shrink future supply. Essentially a supply-side reform, whether home prices truly stabilize depends on demand-side recovery. The new policy may bring medium-to-long-term valuation de-rating pressures for real estate, with industry concentration rising but surviving developers not necessarily receiving higher valuations given weakening scale, ROE, and asset turnover. The early-September announcement of an additional 300 billion yuan special government bond issuance to supplement core tier-1 capital for major financial institutions continues the "924" policy rather than representing new measures. This scale falls below market expectations and extends beyond large state-owned banks to insurers and policy financial institutions, suggesting regulators favor more rational credit extension by large commercial banks. Financial data shows corporate bonds, non-financial corporate equities, and government bonds accounted for 50.3% of new social financing in the first eight months, historically exceeding RMB loan share for the first time. With credit slowing and quality improving, bank operations are transitioning from scale expansion toward quality, efficiency, and shareholder returns, potentially enhancing dividend attributes.
Core Conclusions: Awaiting New Macro or Industry Catalysts First, policymakers show increasing acceptance of growth reaching the lower end of the Two Sessions target range. Q4 policy will likely continue the late-July Political Bureau meeting tone, accelerating policy implementation with selective targeted incremental measures. Policy focus remains on supply-side issues including technological self-reliance, manufacturing self-sufficiency, and energy security rather than broad demand-side stimulus. If geopolitical tensions persist and oil prices stay elevated, forcing further monetary tightening, rising global recession expectations could challenge China's exports, warranting attention to domestic demand offsets. The earnings divergence between emerging and traditional economies is expected to continue. With the Fed resuming hikes, Q4 requires heightened attention to geopolitical developments, overseas liquidity conditions, and global growth industry trends, especially AI.
Second, the K-shaped divergence in macroeconomics and corporate profits means A-share indices lack trendline momentum. Trading will likely be subdued from late September to early October due to holidays unless unexpected macro or industry developments occur. October's Q3 earnings season will see AI chain prosperity expected to continue. November may shift to thematic trading as institutions enter ranking periods. December positioning for year-end rallies depends on policy signals from the Central Political Bureau and Central Economic Work Conference.
Third, traders are watching whether AI applications can replace coding as the new narrative or achieve breakthrough progress, which remains critical for AI chain investment. Unless application-side achievements occur, AI infrastructure is unlikely to repeat H1's broad beta rally given the token price decline trend. The support-without-stimulus policy stance prevents convergence of fundamental divergence between emerging and traditional economies. Q4 is expected to feature more alpha opportunities in individual stocks or sub-sectors, with trading difficulty substantially higher than H1.
Fourth, for stock index futures operations, internal factors remain the key determinant for A-share market direction despite external disturbances. If domestic policy shifts comprehensively (such as increased fiscal stimulus focused on consumption and welfare) or AI industry makes notable breakthroughs, the market could break its stalemate and stabilize. Conversely, absent further catalysts, the market will continue its range-bound pattern. Upside risks include "924-style" domestic policy shifts and AI chain breakthroughs. Downside risks include escalating overseas geopolitical conflicts, unexpected tightening of overseas liquidity, rising global recession expectations, reversal in AI chain prosperity, systemic US equity risks, and underwhelming domestic growth-support policies.