For stock trading, read the Golden Kirin analyst research reports — authoritative, professional, timely, comprehensive, helping you uncover potential thematic opportunities! The following content comes from the latest exchange notes of Western Lead fund manager Dong Weiwei, especially his thoughts on AI, which are worth focusing on.
Recently, market performance has been quite poor, and we have systematically sorted out the logic from various aspects. The A-share market is highly complex; doing investment requires not only looking at fundamentals but also considering macro, geopolitics, and chip structure. Many times the market will be in a very chaotic state, especially during this period, and I believe many institutional investors also feel it is somewhat difficult to grasp.
Regarding this round of decline in the A-share market from the end of September to now, if you say there have been great changes in fundamentals, there actually haven't been. US stocks have only fallen recently, but overall they had been rising all along; US Treasury yields have risen, but not in a jumping spike; the domestic economy has been weak for a long time, not suddenly deteriorating recently; policies have also been introduced for real estate, but during this period tech stocks experienced a very significant sell-off, similar to the July adjustment, but at a very fast pace, a rapid decline on top of the weak rebound in August and September. The real reason for the decline is hard to pinpoint precisely, but all information will ultimately be reflected in stock prices, essentially manifesting as an imbalance in trading structure and a collapse in chip structure. Below I will analyze and summarize from multiple dimensions, though the views are for reference only.
First, a core conclusion: in the future, we should appropriately lower our return expectations for the equity market, with opportunities being more structural and phased. Next, I will elaborate. Starting from a review of A-share bull-bear cycles, looking back at A-share history, an upward wave lasts roughly 200 to over 500 trading days, with 400 trading days being approximately the central average of a major upward wave. From the 924 rally to June this year, the duration was also over 400 trading days, close to the average time window of A-share bull market upward waves. In this structural bull market led by AI, the market saw many 10-baggers and even 20-baggers, which is a very large scale.
After entering late September, in this decline, technically, the Shanghai Composite Index's annual line turned from positive to negative, and now most broad-based indices have formed a comprehensive bearish alignment. Short-term moving averages are falling faster, long-term moving averages have also turned, and the adjustment magnitude exceeds a normal adjustment, with the ChiNext and STAR Market indices having fallen more than 30%. Overall judgment: from a cyclical perspective, A-shares may have entered a downward cycle.
The A-share bull-bear cycle is largely determined by the relationship between capital supply and demand: industrial catalysis and policy shifts spawn a round of gains; after stock prices rise overall, increased new share issuance, reductions, and refinancing follow, and after supply and demand balance, combined with some disturbances at the industrial level, the market easily enters a downward phase caused by capital supply-demand imbalance. It is quite difficult for the A-share market to produce companies with decades of sustained upward operating performance, especially private capital. Of course, it should also be noted that compared with before the 924 rally, the state's positioning of the capital market and its willingness to support it are now higher than in the past. With the real estate downturn, high debt-to-financing ratio, weak bank credit willingness, and domestic tech companies needing to compete with the US, the capital market is a very important lever, and our top leadership is very willing to make it work well, not wanting the market to become a bear market and lose its financing function. Although financing is part of the reason for the market decline, once it falls, the regulatory authorities still have a very strong willingness to gradually introduce many direct or indirect measures to maintain the market, including market stabilization mechanisms, counter-cyclical adjustment on the capital demand side, IPO suspensions, or control of reductions and incentives for increased holdings and returns to encourage listed companies to launch these — all are possible means. The July decline could still be classified as normal, but if this round of decline does not stop, the market trend will likely be more volatile or slightly downward.
The future is not completely without opportunities; opportunities are structural and phased. Structural opportunities: we found that during this round of tech plunge, dividend-type, pharmaceutical, real estate, and some thematic sectors performed periodically, just not within the vision and preference of mainstream institutions. Phased opportunities: after falling a lot, there will inevitably be a rebound. Looking back at 2022-2024, even in a major downward cycle, there were strong rebounds in between, like the new energy wave from April to June 2022, which actually rose very significantly, but over a longer period, it was a strong decline. In summary, considering A-shares' own bull-bear cycle, the current capital market positioning, the rapid response of market stabilization mechanisms, and the diversification of subsequent reserve tools, the probability of a direct bull market is not high, and the probability of a complete bear market is also limited, with the overall bias toward a volatile pattern.
In addition, the scale of domestic funds is not small, but currently everyone has low risk appetite and is unwilling to enter the stock market. The long-term investment attractiveness of the real estate market has greatly declined, bank deposit rates are less than 1%, and funds need somewhere to go. If the stock market stabilizes and some stocks fall to value levels, incremental funds will still enter. So my judgment is that for a period going forward, especially if the index falls below 3700 points, one should not be overly pessimistic. Whether we look at a rebound or a long-term downward cycle, it is not appropriate to reduce positions at this level, or to be very bearish.
The macro perspective is divided into two aspects: First, overseas liquidity. The core contradiction overseas is the continuous rise in US Treasury yields, caused by three factors — US fiscal policy, the Middle East war provoked by Trump, and AI — competitively pushing up interest rates. All three factors are competing for capital, and on the financing side they are all competing, causing rates to keep rising. Among these three factors, we need to judge which one is likely to be resolved. If alleviated, it could be positive for the stock market.
1) US fiscal policy is very rigid, related to public welfare, education, and healthcare — about 70% of spending is rigid, and cutting it would affect votes; government departments' discretionary decisions, like military spending, are also hard to reduce, and there is also the need to replace existing debt from the low-rate period, so fiscal pressure is hard to substantially resolve in the short term.
2) The Middle East conflict — the Trump administration has relatively strong room for mediation. Approaching the US midterm elections, if the Nasdaq sees another stampede-like drop of around 2%, I judge Trump has a relatively high probability of making a statement, whether verbal easing or substantive action. The Middle East situation is the link among the major variables most likely to see marginal easing.
3) The backlash from AI itself — North American cloud providers' cash flow has turned negative, with large capital expenditures relying on bond issuance and financing leases, and some emerging cloud companies' financing rates have risen to 15%. Rising risk-free rates directly raise financing costs, while CDS credit spreads are widening, and the market demands higher risk compensation. AI's own revenue growth has also slowed — costs rising on one side, revenue below expectations on the other. The AI sector can only suppress further rate increases through stock price declines, otherwise the contradiction will intensify further.
Therefore, we need to closely watch the Middle East situation and various variables brought by the US midterm elections. The overall macro environment still puts considerable pressure on equities, especially the tech sector. For a decent rebound to occur, US Treasury yields and rate hike expectations need to fall back.
Second, geopolitics. This year China-US relations have been relatively stable overall. We used rare earth controls, and Trump has a stronger businessman attribute, paying less attention to values. The market generally expects Trump may visit China in November, and our side will go to Miami for the G20 summit in December. This year there is a basis for maintaining bilateral relations. But next year's uncertainty is great: if Trump loses the House in the midterms, or even loses both chambers, the Democrats coming to power will value values issues more, and the intensity of tech blockades will likely further increase. If Trump retains both chambers, geopolitically he has already made considerable progress in South America and Greenland. If the Middle East issue is also resolved, the future重心 of the game will further shift to China. In addition, Europe and the US are promoting a mining alliance. If rare earth smelting technology achieves a breakthrough, it will also bring new variables to bilateral relations. Next year China-US relations still face many challenges.
Looking at the domestic macro, real estate has been declining in volume and price for five consecutive years. Compared with global real estate bubble history, the adjustment time and magnitude are already in extreme ranges, and our financial system has withstood this round of pressure. From weekend survey information, Shanghai's home viewing volume and transactions have slightly recovered, but pressure in third- and fourth-tier cities remains great. There is still the possibility of further policy introduction. The probability of a significant further decline in real estate at least this year or next year has somewhat decreased, but constrained by population structure and housing preferences, it is hard to expect a strong prosperity reversal. It is hard to see a strong reversal in domestic demand. The recent rise in some sectors is more about funds shifting out of tech.
Summary: To see a market reversal, the most important trigger is a renewed fall in US rate hike expectations. The current market implies two more rate hikes next year.
Next, industry — focusing on AI. Affected by rising interest rates, the market generally expects next year's AI major players' capital expenditure growth to fall from over 80% this year to around 30%. There are many controversial points: frontier model growth slowing — of course Anthropic's explanation is actively giving up 5-7% of revenue to prevent domestic model distillation, but regardless of the reason, the data indeed did not exceed expectations. At the same time, no particular bottleneck has appeared. Storage price increases eroding profits, and domestic open-source models competitively cutting prices — these negative factors are now being repeatedly examined under a magnifying glass by the market. But from another angle, the near-term industry fundamentals remain relatively solid. Leading model manufacturers are only part of the computing power consumers; a large number of small and medium AI enterprises and training-side demand are not reflected in public data. Computing power rental prices recently hit new highs, confirming current demand remains strong. North American optical communication companies' orders are already scheduled to 2028 and 2029. Domestic optical module manufacturers feedback that 2027 orders are basically locked in, and 2028 is also mostly negotiated. That is to say, the near-term industry prosperity has not significantly deteriorated, and disagreements are mainly concentrated in the long term.
After this round of sharp decline, risk has been released a lot. If the macro level subsequently develops favorably and the chip structure is cleared to some extent, the AI sector has a relatively significant rebound opportunity. If there are further catalysts — whether AI will have further nonlinear changes is uncertain. If our world models, physical AI, and such have major breakthroughs, there could be a major expectation revision. Or even without these, OpenAI and Anthropic's IPO demands are very strong. The previously mentioned RSI recursive improvement — if very obvious data comes out, or capability iteration speed is very fast, the narrative may change again. So the AI industry itself doesn't have major nonlinear changes; it's just that in the short term there are macro trading structure issues.
Sina statement: This message is reproduced from a Sina partner media. Sina publishes this article for the purpose of conveying more information and does not mean it agrees with its views or confirms its description. The article content is for reference only and does not constitute investment advice. Investors operating based on this bear their own risks.
Comments