After the Shanghai Composite Index falls below 4000 points, what lies ahead? Xingzheng Strategy: Keep a close eye on the July financial report of TSMC and North American cloud manufacturers' Capex.

Industrial released a research report this week stating that the market experienced significant volatility. At the beginning of the week, the market fell under pressure from the "deleveraging" in the US, Japan, and South Korea, but on Thursday, market sentiment followed the global AI industry progress, leading to some recovery. However, on Friday, the AI hardware sector once again experienced a significant pullback, with the Shanghai Composite Index falling below 4000 points, causing some investors to express concerns about the future market, especially in the technology sector. However, at the current time and position, considering recent changes, there are some positive signals accumulating behind the large fluctuations. The adjustments triggered by liquidity and industry narratives in this round may be nearing an end, with some key changes forming a new consensus. With overseas liquidity and sentiment drag, and the global AI industry narrative, which were the core factors affecting the market correction, experiencing key changes and positive signals recently, there is no need to overly worry about the future market. The momentum for a new rally is accumulating and brewing amidst the volatility. Key points from Industrial are as follows: 1. Behind the large fluctuations: Positive signals are accumulating This week saw significant market volatility. At the beginning of the week, under the pressure of "deleveraging" in the US, Japan, and South Korea, the market sentiment followed global progress in the AI industry on Thursday, but on Friday, the AI hardware sector once again experienced a significant pullback, with the Shanghai Composite Index falling below 4000 points, causing some investors to express concerns about the future market, especially in the technology sector. However, at the current time and position, considering recent changes, there are some positive signals accumulating behind the large fluctuations. The corrections triggered by liquidity and industry narratives may be nearing an end, with some key changes forming a new consensus. Firstly, for the overall market, the tracked crowding indicator has once again issued a signal of a short-term sentiment bottom. After a fall due to external factors followed by a chip exchange, the overall crowding in the A-share market has fallen to the bottom range of -1.5 standard deviations. Historically, periods with crowding levels similar to the current level in early December 2025, early April 2026, and mid-June 2026 have all witnessed important bottom layout opportunities. Secondly, with regard to the core factor that triggered this round of correction - AI industry narrative, through recent progress and performance verification of leading companies, the previous concerns have been to some extent corrected and reversed. The core concerns in the market about AI included peaking computational power demand and slow commercialization realization, and the recent progress of leading companies has been enough to address the biggest doubts in the market in recent weeks: Internationally, the willingness of cloud giants to increase computational power remains strong, and the commercialization realization by model giants has exceeded expectations, with the foundation of "rising Capex for cloud giants and rising ROI for model manufacturers" still continuing. 1) Meta announced a continued investment in computational power; Broadcom received orders from Apple; and Micron's expansion plans in the US continue to increase. 2) Anthropic's latest ARR has skyrocketed from over $9 billion at the end of 2025 to over $60 billion, proving to the market that the business model of large models is gradually becoming successful. With the high growth of model manufacturers' ARR, an increase in Capex for core cloud giants can be expected. Similarly, domestic developments have exceeded market expectations, becoming the largest change in the narrative recently. The IPO of Changxin is expected to drive a new upswing in domestic Capex, domestic computational power performance realization has begun, and domestic AI services and applications have started to make profits. 1) After Changxin's IPO, a new round of expansion is expected, boosting orders for domestic equipment, components, and materials, driving domestic semiconductor Capex into a new upswing cycle. With improvements in domestic supply availability and confidence, the relaxation of GPU delivery constraints is expected to accelerate the realization of domestic computational power performance. 2) Some leading companies in the domestic AI and DC chains have significantly surpassed market expectations, proving that with a prosperous domestic computational power demand, the volume of the domestic computational power chain has already started, becoming a direction with significant differences in expectations recently. 3) In Alibaba's latest forward guidance, cloud business revenue continues to grow rapidly, and downstream services and applications of AI in China have already started to realize performance. Therefore, with overseas liquidity and sentiment drag, and the global AI industry narrative, which were core factors affecting the market correction, experiencing key changes and positive signals recently, there is no need to overly worry about the future market. The momentum for a new rally is accumulating and brewing amidst the volatility. 2. What is the market waiting for behind the large fluctuations? However, the biggest feeling in the market recently is still leaning towards rotation and speculation, with the intensity of industry rotation tracking rising to a relatively high level for the year. On one hand, the core focus of the market - AI computational power hardware, although gradually forming a new consensus, is still experiencing significant volatility; on the other hand, some low-level high-performance sectors are undergoing a temporary rebound, but the sustainability of the market is relatively limited. The core of the current market state lies in the fact that AI computational power hardware is undoubtedly the biggest consensus in the market, but funds are rotating and waiting for clearer right-side validation signals in the midst of fluctuations. And these signals are not far from arriving. Key validations that need to be observed going forward: Firstly, potential leading performance catalysts before July 15th. July 15th is the deadline for the disclosure of performance forecasts by A-share listed companies, so it is crucial to observe whether there will be technology leading performance indicators that can further consolidate market consensus and validate the trend. Secondly, Taiwan Semiconductor's announcement of June revenue on July 13th, its Q2 financial report on July 16th, and subsequent guidance. As a upstream player in the computational power industry chain, changes in global demand will be reflected directly in its financial data and guidance. Its order situation, capital expenditure plans, and capacity deployment guidance can help the market judge the sustainability of future global AI computational power demand more clearly. Thirdly, and most importantly, the capital expenditure guidance from the four major North American cloud giants in late July. This is undoubtedly the core focus of the market currently, serving as a right-side signal to validate the sustainability of global demand for computational power and the prosperity of the AI computational power hardware industry. The mainstream view still holds that capital expenditure will continue to rise, and it will be clearer after further communication between the cloud giants and the market. The earliest will be Google's financial report on 22nd July. In the future, as these right-side signals verifying the global computational power landscape, and the sustainability of AI hardware manufacturer performance become clearer one after another, it will help the market reduce volatility, choose directions, and choose structures. 3. Clues worth focusing on in the current interim performance forecasts Before the arrival of the core validations mentioned above, the market may continue to focus on some domestic economic indicators for a supplementary rally. Clues worth paying attention to in the current interim performance forecasts are "expectation gaps" and clues: Demand related to AI is still the most important driver of listed company economic indicators, and the sectors being driven are continuously spreading along the upstream and downstream channels: 1) The domestic AI and DC infrastructure chain (servers, ICT infrastructure, computational power leasing, etc.) benefiting from the expansion of domestic computational power demand has already begun to realize its performance, becoming a direction with the largest margin of change and expectation gap; 2) The expansion driving upstream materials price hikes and releasing performance in AI: MLCC, copper-clad laminates, electronic specialty gases, resins, panels, refrigerants, etc. Other clues worth paying attention to: 1) A large number of lithium battery materials (separators, positive and negative electrode materials, electrolytes, lithium iron phosphate hexafluorophosphate, lithium salts, etc.) companies showing high growth in performance or turning losses into profits; 2) Innovative drugs, ships, securities firms, and coal are currently low-level high-performance clues worth paying attention to. For sectors with strong economic consensus like AI computational power hardware (optical communications, semiconductor industry chain), although the interim performance forecasts already revealed can continue to validate the accelerating economic trend, the market is still waiting for core leading performance indicators, as well as right-side signals from overseas financial reports, guidance, etc. Internally, there is a focus on the configuration cost-effectiveness of the North American computational power chain. On one hand, optoelectronic modules, optical fibers, PCB crowding has fallen to the bottom level, while on the other hand, comparisons between the A-share North American computational power chain and domestic computational power chain leading indicators have accelerated and returned to levels as of June last year. Risk warning Economic data fluctuations, looser than expected policies, lower than expected US Federal Reserve rate cuts, escalating geopolitical tensions, etc.