Zhongtai Securities Co., Ltd. has released a research report analyzing the current market dynamics, where the core debate revolves around whether the "AI capital expenditure cycle has peaked." From an industry trend perspective, the AI sector's trajectory remains intact, with recent corrections driven by growing skepticism about the sustainability of AI-related spending. The bearish argument appears to be a double-edged sword: if major companies increase capital expenditure, it dilutes shareholder equity through debt, but if they cut spending, it signals the end of the AI cycle. However, this logic misses a crucial point: debt itself is not the problem; the key factor is whether investments generate positive returns.

Section One: Market Liquidity Recovery and Index Rebound Last Week

Last week, A-share market liquidity showed signs of recovery, with major indexes fluctuating and rebounding. Broad-based indices saw mixed results, with large-cap blue-chip stocks outperforming. The Shanghai Composite Index rose 1.33%, the Shenzhen Component Index increased by 0.49%, and the ChiNext Index gained 1.52%. In contrast, small and mid-cap stocks were weak, with the CSI 1000 index dropping 2.40%. By sector, high-dividend resource sectors like non-ferrous metals, petroleum and petrochemicals, coal, and banking led the gains, acting as safe havens for capital, while building materials, media, and textile apparel sectors declined the most. The core driver of this correction was a risk release following excessive crowding in AI computing direction trades, with the market undergoing a process of position clearing and style rebalancing.

Section Two: How Will Markets and Technology Unfold After the Liquidity Shock?

In the current market, whether the "AI capital expenditure cycle has peaked" is the central point of contention. After a sharp decline in the week before last, last week saw some recovery, with core ETFs experiencing inflows and major indexes showing volume stabilization. This process resembles the liquidity deleveraging seen in early 2024. Looking ahead, liquidity risks have been temporarily mitigated, and the market may see a rebound around the listing of Changxin Technology. From an industry trend perspective, the AI sector's trajectory remains intact. The recent adjustment is primarily driven by market doubts about the sustainability of AI capital expenditure. The bearish logic appears to be a double-edged sword: if major companies increase capital expenditure, it dilutes shareholder equity through debt; if they cut spending, it signals the end of the AI cycle. However, this logic misses a crucial point: debt itself is not the problem; the key factor is whether investments generate positive returns. Currently, the revenue growth of the world's two leading AI companies remains strong, and AI agents are rapidly expanding from code programming into non-coding white-collar work scenarios such as office, finance, legal, and consulting, providing underlying support from downstream demand. Therefore, capital expenditure expansion backed by positive returns is fundamentally different from burning cash without limits. The essence of this decline is liquidity risk from forced liquidation of leveraged funds, not a reversal of the AI industry's fundamentals. Additionally, under political pressure from President Trump, the upside for oil prices is limited. The conflict between the US and Iran continues to escalate, with Brent crude oil prices rising further last week, sparking concerns about oil reaching $100 or more. However, the Trump administration has a clear weakness in this oil price game, as the Strategic Petroleum Reserve has been significantly depleted, facing political constraints from domestic inflation and midterm elections. Within a month, the situation is likely to ease, with current oil prices already at a temporary peak. Looking ahead, with the listing of Changxin Technology next week, the market may usher in a new round of gains. Recently, growth and technology indices such as the CSI 1000, STAR 50, and ChiNext Index have seen large capital inflows, and insurance institutions have made frequent statements, signaling that liquidity risks have been temporarily mitigated. The key verification indicator is whether the small and mid-cap and technology directions, which have suffered the largest declines, will receive sustained capital support. Before the listing of Changxin Technology, the market must maintain a certain level of heat and trading volume, and with leveraged funds already significantly reduced, a short-term rebound is likely. It is important to note that quantitative trading and financial regulation in August may still cause volatility, but due to the relatively thorough clearing of leveraged funds this time, the probability of the market falling below 3,740 and hitting new lows is low.

Section Three: Trends in Active Fund Heavy Positions in Q2 2026

In the second quarter, the concentration of active fund holdings increased significantly, with A-share market cap holdings rising from 2.07 trillion yuan to 2.81 trillion yuan, a quarter-on-quarter increase of 35.25%, while Hong Kong stock holdings declined. The market cap share of the top five heavy industries surged from about 61% in the first quarter to 71%. The electronics sector's share jumped from 18% to 39%, and communications rose from 11% to 15%, while sectors like power equipment and pharmaceutical and biological saw notable declines. Fund holdings evolved into a dual-dominant pattern of electronics and communications. Excluding price factors, the three most heavily increased sectors were building materials, electronics, and mechanical equipment, while communications and computers also saw steady increases, with capital spreading from AI computing to downstream areas like intelligent manufacturing and industrial software. Traditional cyclical and consumer sectors like media, steel, and commercial retail saw significant reductions. The top 20 heavy positions underwent drastic changes, with eight semiconductor companies such as GigaDevice, Three Circles Group, and Shengyi Technology entering the list, while traditional leaders like Zijin Mining and Kweichow Moutai exited, reflecting funds' embrace of domestic substitution and new quality productive forces themes. Hong Kong stock holdings were concentrated in semiconductor targets like SMIC and Hua Hong Semiconductor, as well as Innovent Biologics, while holdings in Tencent and Alibaba were reduced.

Section Four: Investment Suggestions

Liquidity risks may have been temporarily mitigated. Looking ahead, before the listing of Changxin Technology, the market must maintain a certain level of heat and trading volume, and with leveraged funds already significantly reduced, a short-term rebound is likely. First, within the technology sector, priority should be given to domestic semiconductor equipment, storage, and computing power directions. After Changxin's listing, the market may experience a rebound, with the STAR 50 index likely to be the most elastic direction. Second, outside of technology, power equipment (with overseas demand logic), non-ferrous metals (copper and gold, with elasticity opening after oil prices bottom out), and innovative drugs are all worth appropriate attention. Risk warnings include risks of global liquidity tightening beyond expectations, market game complexity beyond expectations, and policy change pace complexity beyond expectations.