CITIC SEC notes that the strengthening of the K-shaped market divergence has been driven by prevailing narratives. Recent fluctuations and convergence as these narratives shift back are to be expected. This week has seen marginal changes in two key narratives. One relates to assumptions about the global monetary environment; the market is beginning to more comprehensively assess the Federal Reserve's policy stance rather than presuming a hawkish rate-hiking path. The narrative around monetary tightening and a strong US dollar is shifting, which has helped repair negative sentiment towards non-AI sectors. The other change concerns the AI industry trend. News related to META sparked significant controversy, reflecting the market's current low tolerance for any negative AI industry information and highlighting the downstream need for more diverse commercial monetization models to support more aggressive upstream investment expectations. Additionally, we anticipate a significant easing of outflow pressure from A-share broad-based ETFs, representing the most crucial marginal change in market liquidity. The correction in the interest rate hike narrative, combined with these liquidity changes, is expected to catalyze a recovery in some non-AI sectors with solid earnings support.
Shifting Narratives Behind the K-Shaped Divergence
Market volatility is within expectations. Over the past few weeks, we have repeatedly explained how the global K-shaped market divergence has been reinforced by narratives. Firstly, a strong US dollar and interest rate hike expectations themselves have a greater impact on non-AI sectors, exacerbating the K-shaped split driven by diverging business fundamentals. Secondly, relatively weak domestic demand data in China has deepened investor bias against non-AI sectors, leading to a narrative of opposition between "carbon-based" and "silicon-based" sectors—a discussion rarely seen overseas. This has caused A-share non-AI sectors to significantly underperform their overseas counterparts, even for companies operating with global exposure. This is particularly true for cyclical manufacturing sectors with solid fundamentals and innovative drug sectors with sustained positive industry trends. Thirdly, beyond benefiting from the extended景气 of the North American AI supply chain, China's tech sector has also priced in a valuation premium for domestic supply chain autonomy and controllability. Since June, it has significantly outperformed overseas AI sectors. Setting aside narrative-driven and emotional opposition, the market should inherently have a more diverse pricing system and包容的审美视角. When a market is temporarily dominated more by narratives and sentiment, short-term波动 due to narrative shifts is normal. The long-term bull market logic and the primary线索 of technology will not change because of this.
Marginal Changes in Two Narratives
Monetary Assumptions and Industry Trends
1) Monetary Assumptions: The market is beginning to more comprehensively审视 the Federal Reserve's policy orientation rather than presuming a hawkish rate-hiking path. Recent remarks by Waller exhibited characteristics of a "hawkish stance不变, with marginal wording turning dovish." He stated that US inflation risks have eased over the past four weeks and that the supply expansion brought by AI could profoundly change economic operations. However, whether this leads to inflation or deflation should be judged by the central bank based on data. The market narrative that AI infrastructure is crowding out energy and resources, leading to inflation in the "carbon-based" world and triggering rate hikes, currently lacks sufficient data verification. Waller had previously indicated a greater focus on the trimmed-mean PCE, which剔除异常价格扰动, compared to core PCE. The year-on-year trimmed-mean PCE for May was 2.4%, significantly lower than core PCE, indicating that underlying inflationary pressures remain relatively温和. Subsequently released June non-farm payroll data further verified a cooling labor market, leading the market to reprice the Fed's policy path, with US Treasury yields and the US dollar index同步回落. Since February, the market began repricing hawkish expectations, the US dollar index rebounded, and K-shaped divergence across markets intensified. AI hardware demand, due to its high景气 and strong order visibility, has relatively limited sensitivity to monetary tightening. In contrast, non-AI sectors have been more suppressed by high interest rates, a strong dollar, and risk aversion. Recent marginal changes in the走势 of the dollar and gold show that the market's extreme担忧 about liquidity tightening has somewhat eased. As the strong dollar narrative cools and overseas liquidity expectations marginally improve, valuations for previously suppressed non-AI sectors are expected to gradually修复, and market style may shift from extreme AI concentration towards a more均衡 structure.
2) Industry Trends: META's move sparked significant controversy, reflecting the market's lack of tolerance for negative AI industry information. We believe META's shift towards IaaS itself has little substantive impact on AI computing investment; opening IaaS services might even be to更持续地稳住 CAPEX. However, the capital market gave META's stock price a positive short-term定价. A greater concern is mutual learning among CSP providers. In this matter, the risk is actually可控. Google/Amazon/MSFT already have IaaS businesses, so it's not about模仿 META, and currently, stock price is not the primary short-term矛盾 for these companies. The more critical issue is that the current market热度 is so high that it is extremely sensitive to any negative information about AI. This itself implies that前期 market differentiation was too severe, and the行情需要适当的均衡. Recent争议 over token costs are also heating up among North American AI application companies. The CEO of Palantir's questioning of the business models of frontier AI labs has sparked热议. This means that even within the AI sector itself, the imbalance in economic profit distribution between upstream and downstream is triggering industry反思. Hardware manufacturers represented by storage and optical communication have明显跑赢 the Mag7 this year. If calculated based on six-month cumulative excess returns, the figure currently reaches 199%. Balancing and理顺利益分配 may be crucial for重启 the next stage of the market.
Important Marginal Change in Fund Flows
Outflows from broad-based ETFs are expected to slow significantly in the second half. Over the past three weeks, four major沪深300 ETFs saw net redemptions of 42.2 billion yuan, 90.4 billion yuan, and 58.5 billion yuan respectively, while other major broad-based ETFs did not see significant net redemptions. We expect the selling pressure on broad-based ETFs to ease substantially in the second half. On one hand, based on calculations using actual shareholding data as of the end of 2025, as of July 3, 2026, large funds hold approximately 200-300 billion yuan in remaining沪深300 ETF holdings. If the current net redemption pace continues, it is预计 to end within a week, leaving little potential selling规模. On the other hand, according to market consensus forecast data, the projected 2026E net profit growth rate for all A-shares / all A-shares excluding financials and petroleum & petrochemicals is about 9% / 13.5% respectively. An annual index increase of around 10% could be considered consistent with a平稳慢牛节奏. The necessity for下半场控温 has also decreased. In the first half of this year, broad-based ETFs累计净赎回 1857.7 billion yuan, while sector/theme ETFs累计净申购 253.1 billion yuan, with tech ETFs净申购 88.9 billion yuan. For non-AI sectors lacking active fund inflows, this相当于 being in a净减量 market of罕见规模. Conversely, if ETF net outflow pressure减小 in the second half, the资金压力 on A-share non-AI sectors will also明显缓解.
Focus on the Recovery of Some Non-AI Sectors as the Rate Hike Narrative Shifts and Market Liquidity Improves
In the first half, the强势美元 and Fed rate hike narrative had a stronger抑制 effect on non-AI sectors, thereby reinforcing the K-shaped market split. If this narrative扭转, it will conversely促使 the K-shaped divergence to收敛. Looking ahead to the second half, as net outflows from broad-based ETFs are expected to slow, the资金压制作用 previously exerted on non-AI sectors may明显缓解, which is also conducive to the修复 of some non-AI sectors with solid fundamentals. Regarding具体配置, we still recommend adhering to an AI + Energy/Chemicals structure. On the AI side, our preference is shifting towards sectors with强供给约束 and low估值缩圈. From an industry chain perspective, we更推荐偏中下游的品种, such as cloud providers, storage, gas turbines, diesel generator sets, etc. On the Energy/Chemicals side, within电新, we are看好电解液及添加剂,隔膜等品种 for their业绩兑现. In化工, a下降 in oil price中枢 and波动率 brings补库开工 demand, and peaking macro liquidity expectations could be subsequent潜在的节奏点. Currently, we are更看好一些成本下探空间大,相对刚需且估值较低的品种, like制冷剂,磷化工,氨纶,染料,大炼化, etc. In有色, we recommend算力金属 like tin and copper, which have some AI exposure but whose valuations have been temporarily压制 by the macro加息叙事. Additionally, we continue to suggest增配低估值的券商, as current瑕疵 like流动性压制 may begin to gradually消退 in the second half, with interim earnings previews also serving as a催化. Simultaneously, we看好创新药 for its持续向上产业趋势 and彻底持仓出清, expecting its修复.
Comments