CITIC SEC: The Middle East conflict may become a catalyst for a change in market structure.

Source: CITIC SEC Research

The geopolitical impact of the Middle East conflict is significant, but its actual impact on Chinese assets is limited.

From a strategic perspective, predicting such geopolitical issues is extremely difficult. The commodity market may respond impulsively, and the stock market may show localized reactions in the short term, but due to unpredictability, it's hard to become the main variable dominating the market.

The trade war in April reminds us once again that once the long-term reassessment of Chinese assets becomes a consensus, any trading that overreacts to external shock factors tends to have a high error rate upon reflection. However, rising oil prices can temporarily relieve downward pressure on industrial prices, and more importantly, high oil prices will hinder the decline of inflation in the U.S.

Since February this year, the energy component of the U.S. CPI has entered a year-on-year decline, leading the overall CPI year-on-year increase to drop from a peak of 3.0% in January to 2.4% in May. If oil prices surge due to supply factors subsequently, this suppressive effect will inevitably weaken. The inflation risks in the U.S. are heating up again, and conflicts caused by the deportation of illegal immigrants may impact the advancement of the tax reduction bill, while also restricting Trump's room for maneuver on trade issues. If viewed from this perspective, the conflict in the Middle East may not necessarily have a negative impact on Chinese assets.

The sudden change in risk appetite has had the greatest impact on sectors that are at high positions, have high turnover, and strong consensus.

1) The factors supporting the collective rise of micro-cap stocks are showing signs of cracking.

The overall risk of the market is limited under the support of institutions like Central Huijin, but there are no signs of a general improvement in profits for now. Subjective long-position institutions are weak, while the Algo incremental capital is strong and the exposure is tilted towards small micro-cap stocks. These factors have collectively created an environment where micro-cap stocks trend upward while large-cap stocks remain sluggish. However, these factors are beginning to show signs of easing. In an environment where the China-US trade war has entered a steady state without systemic shocks, there is no need for Central Huijin and other institutions to enter the market in large numbers to stabilize it. After the China-US Geneva talks, we estimate that Central Huijin did not continue buying ETFs, and the number of shareholding reduction announcements from listed companies has clearly increased.

In addition, an increasing number of companies are issuing suspension warnings due to abnormal stock price movements. From December last year to April this year, only 2 companies issued suspension warnings. However, since May, at least 5 companies have issued abnormal trading suspension warnings, not including cases of direct trading suspensions. From historical experience, whenever the frequency of abnormal trading suspensions rises, the sentiment for micro-cap and themed speculation tends to cool down. The advantages of Algo products are also明显ly weakening under the condition of deep discounts in index futures. As of June 13, the annualized discount rate of IM's current quarter contracts (after excluding dividends) reached 14.7%, and very few long parts of the Algo strategies have been able to achieve a relatively close to 15% annualized excess return compared to the CSI 1000 Index. The decline in excess advantages of Algo products may lead to a subsequent decrease in the number and issuance scale of filed products, reducing the incremental capital supporting the collective rise of micro-cap stocks.

The deterioration of risk preferences may lead to volatility in such high-position and tightly held sectors.

Last Tuesday, the news of China-US negotiations triggered a huge market震动, primarily stemming from some high-position tightly held sectors, and capital strategies exhibit a high level of consistency, revealing that the currently high-position tightly held sectors are extremely sensitive to external environmental variables.

As of June 13, our investor sentiment indicator, based on Turnover Ratio and closing prices, recorded 54.3 (with a range of 0~100), down 12.8% from the April high of 62.3, while the closing price of the SSE Composite Index has risen to 85.7 in the past year’s percentile. The divergence between the sentiment indicator and market trends further confirms that rational investors are becoming relatively cautious.

Aside from micro-cap stocks, the short-term crowding in some high-growth hot sectors is also relatively high. For instance, the components of Hong Kong stocks' innovative drugs (excluding CXO) contributed 11.5% of the total turnover, marking the highest historical percentile since 2019. The rolling one-year P/S percentile is currently recorded at 83%, also the highest level since 2019, exceeding the level in July 2020. The overall trading volume of A-shares in new consumption and pharmaceutical sectors reached 11.6%, positioning it at the 84.4% percentile since 2019, while the rolling one-year P/S percentile currently stands at 72.1%, with a significant elevation in valuation center expected in 2025.

The weakening of the small-cap and theme rotation model simultaneously means that the logic of returning to trends like AI in the industry will strengthen.

When market attention is highly focused on small-cap stocks and theme rotations, the changes in the AI industry have not stalled.

The usage of AI application tokens among major companies in North America is rapidly increasing, and user activity continues to remain high. Google stated at the I/O 2025 conference that the amount of tokens processed monthly has grown from 9.7 trillion in May 2024 to the current 480 trillion, an increase of about 50 times. According to the statistics from the AI product list in May, OpenAI's ChatGPT and Sora Web had visit counts of 5.68 billion times (+6.8% MoM, same below) and 0.063 billion times (+121.3%) respectively; Google's Gemini and AI Studio had visit counts of 0.54 billion times (+28.9%) and 0.08 billion times (+11.6%).

Behind this, one reason is that improvements in the model's long-text and memory capabilities provide users with a better experience, while another reason is that North American tech giants have begun to shift from a model arms race to a phase of competing for users. Perplexity and OpenAI's search engines have brought significant pressure to traditional giants like Google and Microsoft, prompting these large companies to strengthen the retention of existing users by offering free AI search services, which further boosts inference demand. It is believed that the cycle of 'improved inference computing power → user stickiness → greater inference demand → more inference computing power demand' has already formed in North America, and it has passed the stage where it relies on a major product catalyst to generate a pulse market.

The perception of this issue in the domestic market may lag behind a bit but will not be far off.

The attention towards AI in the domestic market will also return with the cooling of high-positioned sectors. Of course, the most important factor is that hardware companies in the North American AI computing power chain have already demonstrated the first wave of profit effects. From the perspective of relative strength indicators for industry rotation, as of the week of June 13, the optical module and PCB sectors in the AI industry chain have entered a relatively leading stage.

The movement of sectors itself attracts market attention, prompting investors to begin to subtly focus on changes occurring in the AI field. From the perspective of sector evolution logic, the best varieties currently remain in the North American AI supply chain. The logic of domestic AI may wait until domestic inference demand erupts continuously like in North America before it begins to evolve. In terms of market order, it is believed the path will be 'North American AI supply chain → domestic application inference token explosion → domestic computing power → edge devices.'

The continuous and generally low price signals may become a new policy catalyst, but patience is needed.

After the disruptions of the trade war subside, the bottlenecks of the internal cycle will again become the focus for market attention, discussion, and pricing. According to Iceberg Big Data, as of June 8, the average listing price of second-hand houses in first-tier cities in China has seen a consecutive 14-week month-on-month decline, having accumulated a drop of 3.5% since 2025.

Regarding industrial product prices, as of June 13, the Nanhua Industrial Product Price Index has continuously dropped by 7.2% compared to its peak at the end of March, and it is still running at a low level recently. The effects of the previous round of domestic demand-boosting policies have begun to fade, with many regions across the country experiencing a 'rollback' phenomenon in the old-for-new subsidy program. On June 9, the opinions on further guaranteeing and improving people's livelihoods and focusing on solving urgent and difficult issues, issued by the General Office of the Central Committee of the Communist Party of China and the General Office of the State Council, were made public. The release of such documents likely means that the thinking of boosting domestic demand is gradually evolving from simple consumer subsidies to broader areas such as strengthening social security, which is an extremely correct directional approach.

The public discourse around "anti-involution" has recently gained significant traction, and related industries may become very important clues for the second half of the year. It is believed that resolving these issues will not happen overnight, but as long as open discussions are formed and there is a consensus from top to bottom to solve these problems, new targeted policy enhancements will not be far behind.

Diminishing macro fluctuations, returning to industry trends, and being wary of micro volatility remain the core strategic ideas for the next phase.

It is recommended to focus more on the investment logic that returns to industry trends during the semi-annual report season from June to August, avoiding excessive reactions or trades based on macro information, while staying alert to micro style fluctuations. Specific allocation aspects:

1) In the A-shares, it is suggested to pay attention to the AI chain; the computing power supply chain (AI Servers, PCBs, ASIC chips, optical modules, Switches, computing power leasing, etc.) is worth focusing on; the application phase needs continuous tracking and observation, with clear trends in multimodal visual understanding applications. It is recommended to focus on ByteDance ecosystem companies and component and module companies that are growing alongside brand clients, with a focus on product categories such as AI glasses and AI toys.

2) The Hong Kong stocks have entered the bottom-up stock selection phase. Against the backdrop of geopolitical disruptions and sudden changes in environmental variables, Hong Kong stocks serve as a safe haven for global risk assets, and it is recommended to continue balancing the allocation between A-shares and Hong Kong stocks.

3) It is suggested to start paying attention to policy signals again, as a further general decline in price signals may trigger a new round of policy enhancements, but patience is needed.

Risk Factors

The friction in technology, trade, and finance between China and the United States has intensified; the strength and effectiveness of domestic policies, or economic recovery may fall short of expectations; macro liquidity both domestically and internationally may tighten more than expected; conflicts in the Russia-Ukraine and Middle East regions may escalate further; the digestion of real estate inventory in China is not as expected; and the focus of Trump’s policies exceeds expectations.

Editor/rice