Industrial: The congestion level of AI has fallen to the bottom. It is advisable to long-term allocate investments in oversold hard technology and optical communication chains.

Industrial has published a research report stating that the current round of significant decline has altered the "differentiation" in terms of stock price position, valuation, and playing cards, but it has not changed the "differentiation" in terms of economic prosperity. Therefore, for those quality hard-tech assets that have already fallen out of a reasonable price-performance ratio, the current situation represents an opportune moment for long-term positioning. Moreover, the congestion level in most core AI sectors has retreated to the bottom, and there has been a considerable degree of overselling compared to overseas markets over the past month. Internally, greater attention needs to be paid to North American leaders in optical communications and computational chains. The main points of Industrial are as follows: Over the past month, we have witnessed the most intense fluctuations in the global market since the current AI bull market began. While the world searches for "reasons" to "de-leverage" amidst negative liquidity feedback and panic, the market can only "exchange space for time," waiting for clearer bottom signals and more defined guidance from the fundamentals. This week undoubtedly serves as an important "watershed." Global stock markets have collectively started a "self-rescue" to curb negative liquidity feedback, with several critical developments both domestically and internationally materializing this week, providing the market with a reason to return to rationality and objective analysis. Thus, when it comes to structural choices and the comparison between AI and non-AI, it ultimately returns to the essence of industry comparison: fundamentals, stock price positions, valuations, and playing cardswhat has been altered after this round of significant decline? What remains unchanged? Once these questions are clarified, the subsequent configuration thinking will become much clearer. 1. What has changed after this significant decline: this year is no longer a "differentiated market" in terms of stock price positions, valuations, and playing cards. Before this round of declines, the biggest feeling in the market was the "differentiation" between AI and non-AI. It is true that the substantial floating profits accumulated earlier and the excessively concentrated playing cards are the main reasons for this global AI adjustment. However, what has changed the most after this round of declines lies precisely in this aspectafter a structural "rebalancing" and digestion of playing cards, the market dynamics in terms of stock price positions, valuations, and playing cards are no longer characterized by "differentiation" this year. Firstly, regarding stock price positions: 1) The high prosperity index's annual returns have turned negative and have underperformed the dividend index: The high prosperity index, which measures the performance of leading stocks in high prosperity industries, has seen annual returns drop from over 60% to negative territory and has underperformed the dividend index. 2) The degree of differentiation in the rise and fall of primary industries from the beginning of the year to now has fallen to the fifth-lowest level since 2010: Prior to the decline (January to June), the standard deviation of the rise and fall of primary industries was 26.6%, marking the fourth-highest since 2010 (only behind 2013, 2015, and 2020); after the decline (January to July), this indicator significantly dropped to 11.5%, the fifth-lowest since 2010 (only higher than 2011, 2016, 2018, and 2022). Secondly, in terms of valuation, most typical technology and consumer leaders currently have a 27-year PE ratio in the range of 10 to 20 times. From a long-term allocation perspective, todays tech leaders are no longer considered expensive. Lastly, in terms of playing cards: 1) Our measure of market sentiment's congestion has clearly shown a "reversal" and "shifts between old and new" roles: The previously low congestion levels in consumer stocks and dividends have risen to high levels, while most technology growth segments have seen congestion levels drop back to historical lows. 2) Following this round of declines, the position pressure in AI sectors should also have been significantly digested: For actively managed public funds, since July, AI-related sectors (including communication equipment, electronic hardware, non-ferrous metals, chemicals, and glass fibers, etc.) have generally seen a pullback of 25% to 50%. Whether due to the drop in stock prices themselves or the active reallocation efforts by fund managers, the allocation ratio in AI-related sectors should have undergone considerable digestion. As evidence, during the two significant TMT declines on July 28 and July 30, the ratio of actively managed funds experiencing positive deviations (actual declines being less than the estimated drop based on Q2 fund holdings) significantly increased, indicating that during this process, active funds made substantial reallocations compared to the Q2 report. For margin financing, the current margin balance for electronics and communications has dropped 23% and 26% from peak levels, exceeding the magnitude of "de-leverage" seen since 2024, and the absolute values have roughly retreated to early May levels, alleviating the playing card pressure from high-volatility funds. Therefore, after this round of adjustments, the "differentiation" between AI and non-AI in terms of stock price positions, valuations, and playing cards has largely converged. When these dimensions of differentiation are "flattened," the core reconsideration of their comparative advantages in allocation should return to judgments about their prosperous benefits. 2. What remains unchanged after this significant decline: the differentiation in economic prosperity What has not changed in this round of significant decline is precisely the differentiation in economic prosperity between AI and non-AI. Regarding this round of AI adjustments, we repeatedly emphasize that the essence is that the negative feedback from liquidity amplifies the volatility and panic, with no substantial changes in the fundamentals; rather, clearer information is needed to re-establish market consensus. With this week's financial results from North American cloud companies and the recent Politburo meeting outcomes, the market should now have clearer judgments regarding the prosperity of AI and non-AI. Firstly, for AI, the pessimistic scenarios previously feared by the market regarding disappointing performance and a slowdown in AI capital expenditures have not occurred. The latest financial reports from North American cloud companies describe an industrial landscape characterized by AI returns significantly exceeding expectations, a continuous increase in backlogged orders leading to supply shortages, and a certainty in future capital expenditures. First, even after substantial upward revisions since Q2, the proportion and magnitude of exceeding performance for technology stocks in the US market in Q2 remain historically high: In Q2 2023, 85% of US technology stocks exceeded EPS estimates, with a median EPS surprise of 5.6%, marking historical highs. Second, hyperscalers have continuously raised their capital expenditure guidance for 2023, with confirmed expansions for 2024: Google, Meta, and Amazon have again revised up their capital expenditure guidance for 2023, with the four major CSPs collectively estimating around $720 billion to $745 billion. With central estimates, the growth rate of capital expenditures for 2023 has increased to approximately +80%. Though they did not provide quantitative guidance for 2024, none indicated any signal of "reduction." Third, cloud companies report AI returns that consistently exceed market expectations, with a sustained sharp increase in the scale of backlogged orders, ensuring the continuity of AI capital expenditures: Google, Microsoft, and Amazons cloud revenues have all significantly outperformed market expectations, with previously invested capital expenditures translating into tangible AI returns. More importantly, the scale of backlogged orders for these three cloud businesses continues to surge, with growth rates dramatically exceeding cloud revenue growth, exacerbating supply-demand imbalances and boosting visibility for future growth, while also ensuring the sustainability of subsequent AI capital expenditures. Secondly, for non-AI, the focus will still be on domestic policies in the second half of the year. The latest political bureau meeting conveyed a policy thought for the second half of the year that is marginally more aggressive, though ensuring the effective use of existing resources remains a priority. "Counter-cyclical adjustments" and "prompt planning for new policies" are more proactive compared to April. If there is increasing downward pressure on fundamentals in Q3 and Q4, new policies may provide support. However, ensuring effective use of existing resources remains a priority, whether in monetary or fiscal policies, thus suggesting that the necessity for substantial increases in either may be weak. Consequently, the differentiation in fundamentals between AI and non-AI might still persist into the second half of this year, but this round of declines has brought their valuations back to the same level. A simple comparison of PE-G ratios indicates that most typical technology and consumer leaders currently have a 27-year PE in the range of 10 to 20 times, while the respective expected growth rates are over 30% and around 10%. Once the market stabilizes, it will reassess the allocation value of technology leaders based on this. 3. Current allocation strategy This round of significant declines has changed the "differentiation" in stock price positions, valuations, and playing card dimensions, but has not altered the "differentiation" in terms of economic prosperity. Therefore, for quality hard-tech assets that have fallen out of a reasonable price-performance ratio, the current time represents an opportune moment for long-term allocation. Moreover, the congestion levels in most core AI sectors have retreated to the bottom, and there has been a substantial degree of overselling over the last month compared to overseas markets. Internally, increased attention should be directed towards North American leaders in optical communications and computational chains. Additionally, from the perspective of marginal changes in economic prosperity and recent catalysts, we currently recommend focusing on: Downstream AI: Pay attention to the changes in the narrative of the AI mainline moving overseas and the numerous recent catalysts in AI downstream (software, applications); Upstream AI materials: Focus on the recovery of oversold AI upstream materials (new materials, small metals & energy metals, glass fibers, plastics); Advanced manufacturing: Shipbuilding, battery storage, innovative pharmaceuticals, power grids; Cyclical alpha: Since July, there have been upward revisions to profit expectations in chemicals, textile manufacturing, non-financial sectors, industrial metals, and beer.