美股投资young

美股科技投资者 ai狂热者分享每日快讯. all in ai

299Following

662followers

Feed

The semiconductor sector has already fallen this much; who can still save the market in October? I believe the earnings reports in the next two weeks may determine whether $SOXX will see a decent rebound or continue to break down further. $QQQ $SPY $NVDA $MU $SNDK As of Thursday, $SOXX has declined for 8 consecutive trading days and remained under pressure during Friday's session. Even more alarming, TSMC and ASML delivered strong earnings reports but still couldn't stop the sector's decline. When the most core semiconductor manufacturing companies cannot reverse market sentiment, I think it's unrealistic to continue expecting chip companies to save the market with a strong EPS. But the real highlight of October is yet to come. On October 21, $META, $LRCX, and $TXN will release earnings simultaneously. Meta's AI capital expenditure and actual returns will directly impact market confidence in data center investments; Lam Research can verify whether orders for advanced processes and storage equipment remain strong. The same week also includes $WDC, followed by $STX. These companies will help us judge whether demand in the semiconductor and storage supply chains has deteriorated. October 28 is the day I consider most critical. $MSFT and $GOOG will report earnings, and the Federal Reserve will announce its interest rate decision on the same day. Microsoft's Azure growth, Google's cloud expansion, TPU investments, and the free cash flow performance of both companies could directly influence the market's assessment of AI business models. On October 29, $AMZN and $SNDK will report. Amazon's AWS growth and future capital expenditures will further validate AI infrastructure demand; SanDisk will need to prove how long profitability growth in the storage industry can continue through NAND prices, margins, and next quarter guidance. This round of semiconductor sell-off can no longer be simply attributed to weak chip demand. From recent market behavior, I tend to believe that after the short squeeze ended in September, a large number of existing AI bulls are reducing risk exposure, and the market is simultaneously digesting previously overextended valuations. High U.S. Treasury yields, oil prices and inflation pressures, midterm election uncertainties, and massive AI financing demands are all suppressing the willingness of funds to re-enter. Therefore, these upcoming earnings reports are very critical. If Microsoft, Google, Meta, and Amazon all prove that AI demand remains strong, and revenue and operating cash flow can gradually catch up with capital expenditures, combined with a decline in Treasury yields, I believe semiconductors have a real chance for a strong valuation recovery. But if these companies continue to announce significant increases in capital expenditures while free cash flow remains under pressure, the market may become even more worried about AI investment returns and future financing pressures. There is an even more dangerous outcome: all earnings beat expectations, yet semiconductors still do not rise. If even the most important tech giants cannot reverse market risk appetite, it means the current selling pressure is far from over. I remain long-term bullish on AI but cautious about the rest of October's market. After $SOXX's extreme consecutive declines, a technical rebound could happen anytime, but I will not easily assume the bottom has formed before sustained buying returns to the market. For those with heavy positions, the most important thing ahead is not to bet on which company can save the market, but to ensure they can withstand a second round of declines if earnings disappoint. On October 21, watch AI capital expenditures; on the 28th, watch cloud computing and the Fed; on the 29th, watch AWS and storage. These days may be the most important short-term verification window for semiconductor trends. If these earnings at the end of the month still cannot save $SOXX, I will be even more cautious about the subsequent market.

During this October tech stock crash, I find myself increasingly envious of those holding large amounts of cash. For investors who were fully invested chasing highs before, this might be the most painful phase. But for those with ample cash, no leverage pressure, and a willingness to hold for three to five years, I believe the market is gradually carving out a very valuable golden opportunity. $SOXX $QQQ $NVDA $MU $SNDK I have previously emphasized that this adjustment looks more like a continuous withdrawal of existing AI bull funds rather than a full-scale short squeeze. After the short squeeze rally in September ended, the market began to digest the previously overextended valuations. Meanwhile, high US Treasury yields, rising oil prices, the approaching US midterm elections, and concerns about AI capital expenditure returns are all forcing investors to reduce risk exposure. This has created a very noteworthy phenomenon: many companies’ stock prices have plummeted consecutively, but their core competitiveness has not deteriorated to the same extent in just a few days. $TSM still holds the world’s leading advanced process manufacturing capabilities, $ASML maintains very high lithography technology barriers, $GOOG’s search, cloud computing, and AI ecosystem have not suddenly lost value due to price adjustments, and $AVGO’s long-term position in custom chips and network infrastructure remains fundamentally unchanged. Of course, a price drop does not mean all companies are cheap. But if a company’s profitability over the next three to five years has not significantly worsened, yet its stock price has been heavily adjusted due to macro panic and capital withdrawal, then the long-term expected returns may start to improve. That is why I believe cash is especially important at this stage. Those fully invested are thinking every day about when the rebound will come, whether to cut losses, and if they can bear the losses. Those using leverage have no choice; once margin requirements are hit, they may be forced to sell at the worst possible time. But those holding large amounts of cash are completely different. If prices fall today, they can observe; if they fall tomorrow, they can buy in batches. If the market really experiences a second round of systemic decline, they still have funds to continue deploying. You don’t need to guess the bottom, nor do you have the pressure to profit by a certain date. Especially when buying the underlying stocks directly, rather than betting on a rebound through high-leverage options. Underlying stocks have no expiration date that can cause them to go to zero, nor do they lose time value directly due to short-term implied volatility drops. As long as the company’s profitability, competitive advantages, and balance sheet are not permanently damaged, investors have the conditions to wait for business results to gradually materialize. Moreover, I believe the truly valuable opportunities ahead are not limited to semiconductors. Advanced manufacturing can be studied through $TSM, $ASML; AI computing and custom chips can focus on $NVDA, $AVGO; cloud computing platforms include $GOOG, $MSFT; the storage supply chain has $MU, $SNDK; optical communications include $LITE, $COHR; power and data center infrastructure include $VRT, $ETN. These companies do not share the same risks. Storage and optical communications have more obvious profit cycles, software platforms usually have stronger revenue continuity, and power equipment is influenced by orders, capacity, and project cycles. High-quality companies spread across different business models are more suitable for long-term holding than simply betting on a single highly volatile stock. But I would not recommend going all-in at this point. The October decline has been severe, but the systemic risks of $QQQ and $SPY have not been fully released. If US Treasury yields continue to rise, oil prices and inflation pressures fail to ease, or AI capital expenditures start to truly impact corporate profit forecasts, even lower prices could still come. Therefore, I prefer to build positions gradually. Establish a base position first, then increase investment based on further valuation declines, changes in corporate profit expectations, and market stabilization. Always keep enough cash to withstand the next round of adjustments. I think many investors’ biggest mistake is always trying to recover losses from previous months within a few days during a crash. But those with sufficient funds don’t need to be so anxious. Assuming AI infrastructure continues to expand over the next few years, quality semiconductor companies’ profits keep growing, and cloud computing and enterprise software commercialization steadily improve, then stocks bought at reasonable prices now have the chance to benefit from both corporate profit growth and future market risk appetite recovery. Of course, this process may take one, two, or even more years, and not all stocks will return to previous highs. But for companies with true long-term competitive advantages and continuous free cash flow generation, time can become an investor’s friend. I even feel that after this October adjustment, many will re-recognize the importance of cash and position management. Holding cash during the market’s wildest times may mean missing some upside. But when the market truly panics, cash means choice. You can pick companies, pick valuations, wait for better entry points, and even accept further declines after buying. I still believe this October’s adjustment may not be over, and short-term rebounds do not necessarily mean the market bottom. But that does not stop me from seriously looking for companies worth holding for the next three to five years. For those with ample cash, continued market declines are not necessarily all bad. If you can buy batches of truly quality companies at reasonable or even undervalued prices, avoid leverage, control position size, keep enough cash for worse scenarios, and then let corporate profit growth gradually materialize, I believe this is a much more comfortable investment approach than frequent short-term trading. The most comfortable investment state may be having cash when the market falls and having positions when the market rises. No need to guess daily ups and downs, no need to constantly battle the market. Buy truly valuable assets and leave the rest to business operations and time. I believe this is the greatest value of holding ample cash during this October tech stock crash.

In the sharp decline this October, I have instead started seriously screening for the most worthwhile companies to buy over the next two to three years. If the market continues to fall, I believe many truly competitive long-term enterprises are gradually entering price ranges worth reevaluating. $SOXX has fallen for eight consecutive trading days, with semiconductors experiencing an extremely rare continuous sell-off in recent years. Meanwhile, the US 10-year Treasury yield hovers around 5%, international oil prices remain high, and with the November midterm elections approaching, the market is beginning to reassess the capital expenditure and profitability of the entire AI industry chain. I have previously emphasized that this decline, from a market perspective, looks more like the crowded AI long positions actively reducing exposure rather than shorts fully taking over. The September rally included obvious chasing and valuation expansion components. As the short squeeze gradually ends, the market begins to reassess how much these companies can actually earn in the future and what valuations are appropriate in a high-interest-rate environment. Therefore, I remain cautious about a possible second wave of decline. But from a long-term investment perspective, every time the market sells off tech stocks due to macro pressure, it creates a batch of quality companies worth tracking. October may be creating such an opportunity. I have divided the companies currently most worth researching into seven sectors. The focus is on enterprises with genuine technological barriers, profit realization capabilities, and long-term industry positions, rather than simply looking for stocks with the largest declines this year. 1. Advanced Semiconductor Manufacturing: $TSM, $ASML I believe this sector is one of the directions with the highest long-term certainty. TSMC just released its Q3 2026 financial report, with revenue of $46.72 billion, a year-over-year increase of 41.2%; net profit of NT$763.1 billion, up 68.7%; and a gross margin of 66.4%. Advanced processes contributed 79% of wafer revenue. A company at the core of global semiconductor manufacturing maintaining such a high gross margin and profit growth at this massive revenue scale is truly worth attention. In the future, whether it’s Nvidia’s GPUs, Broadcom’s custom ASICs, or Apple and other companies’ advanced chips, all require advanced wafer manufacturing capabilities. Market shares of different chip design companies may change, but high-end manufacturing capacity and yield cannot be replicated quickly. TSMC’s advantages lie not only in manufacturing processes but also in advanced packaging, customer co-design, economies of scale, and long-term accumulated yield experience. $ASML is positioned even further upstream. The company’s Q3 revenue reached €11.7 billion, with a gross margin of 57.5% and net profit around €5 billion. More importantly, management revealed that some customers have already started discussing equipment orders for 2029 and even 2030. The extremely high R&D difficulty, supply chain complexity, and customer certification costs of advanced lithography equipment give ASML a very special industry position. Together, these two companies form the important foundation of advanced semiconductor manufacturing. Of course, they face risks. TSMC faces geopolitical issues, overseas expansion costs, and customer concentration; ASML is affected by export controls, customer capital expenditure cycles, and equipment order fluctuations. But if such companies fall sharply with the semiconductor sector due to macro panic while advanced process demand and customer orders do not deteriorate simultaneously, I would view this adjustment as a very worthwhile long-term opportunity. 2. AI Computing Power and Custom Chips: $NVDA, $AVGO I focus most on these two companies in this sector. $NVDA remains one of the core companies in AI computing infrastructure. In Q2 FY2027, the company’s revenue reached $96.2 billion, up 106% year-over-year, with data center business revenue at $89 billion, up 117%, and an overall gross margin of 75%. This scale of revenue and profitability shows Nvidia is no longer just a GPU seller. The CUDA software ecosystem, network interconnect, complete systems, development tools, and long-term customer usage habits together form its competitive moat. Short-term stock price may experience sharp adjustments due to valuation compression, but as long as global AI computing demand continues to grow, Nvidia still holds a very strong commercial position. However, I do not believe it can enjoy unlimited valuation premiums. Hyperscale cloud providers are increasing self-developed chip investments, and chip performance improvements may also change customer purchasing rhythms. $AVGO is another company I personally value highly. Broadcom’s Q3 FY2026 revenue was $29.6 billion, up 86%, with free cash flow reaching $13.7 billion. Unlike companies relying solely on AI chip revenue growth, Broadcom also has custom ASICs, network switching chips, and infrastructure software businesses. It can participate in hyperscale cloud providers’ self-developed AI chips and benefit from data center expansion through network interconnect devices, while VMware-related software provides another cash flow source. This business model makes me pay close attention. Of course, customer concentration in custom chip business, competition among cloud providers, and future product margin changes require ongoing observation. My judgment is that if the October adjustment continues to deepen, $NVDA and $AVGO will be the two AI core companies I am most willing to repeatedly study valuation and profit forecasts for. 3. Storage Industry Chain: $MU, $SNDK, $WDC, $STX Storage is a sector I have long followed and is the most elastic in this adjustment, while also requiring extra attention to cyclical risks. $MU Micron just released Q4 FY2026 results, with quarterly revenue of $54.23 billion and operating cash flow of $43.97 billion. These figures indicate that AI servers, HBM, and DRAM demand have pushed Micron’s profitability to very high levels. Micron’s advantage lies in its combination of HBM, server DRAM, and other storage products, enabling participation in AI training, inference, and traditional data center storage demand growth simultaneously. But I must remind that the storage industry is highly cyclical. Current extremely high margins cannot simply be treated as normal margins for the next five years. If supply-demand reverses and DRAM prices fall, even if the company remains profitable, the market may preemptively lower valuations. $SNDK is even more aggressive. The company’s Q4 FY2026 revenue was $8.97 billion, up 372%, but about two-thirds of the quarter-over-quarter growth came from price increases and one-third from volume growth. This precisely illustrates its opportunities and risks. If NAND supply-demand tightness continues, Sandisk’s profit elasticity could be astonishing. But once the price cycle reverses, profit compression could be equally severe. Therefore, I think $MU should be studied from the perspective of HBM, DRAM competitive landscape, and long-term customer demand, while $SNDK requires more focus on NAND prices, inventory, capacity deployment, and order structure. $WDC and $STX are two other storage companies worth attention. They mainly benefit from large-capacity mechanical hard drive demand in data centers, differing from Micron’s HBM and DRAM business. As AI-generated data volume increases, cloud providers still need large, cost-controlled long-term storage capacity. Especially $STX’s HAMR technology path and $WDC’s product competitiveness in enterprise large-capacity hard drives are worth continued observation. But the storage sector should not only be judged by this year’s profit growth but also whether current prices already reflect a better supply-demand environment for the next one to two years. I believe October may indeed create opportunities for storage stocks, but these stocks may not bottom on the same day. 4. Optical Communication and AI Networks: $LITE, $COHR, $ANET, $MRVL Optical communication remains a direction I consider very worthy of attention in the coming years. As AI data centers expand, the issue is no longer just how much computing power a single GPU has, but how to achieve efficient data transmission among numerous chips. With the development of 800G, 1.6T, and higher-speed networks, the importance of optical modules, lasers, optical chips, and high-speed switching equipment is increasing. $LITE is a company very worth watching here. Lumentum achieved about $1.01 billion revenue in Q4 FY2026, with a non-GAAP gross margin of 50.4%, and next quarter’s revenue guidance further increased. Its value lies not only in traditional optical modules but also in indium phosphide-related photonic devices, laser technology, and other high-speed optical interconnect solutions. $COHR has a broader photonics technology and manufacturing capability. Q4 FY2026 revenue was $2.05 billion, up 34%, with GAAP gross margin of 38.5%. As data centers gradually migrate from traditional copper interconnects to more high-speed optical connections, Coherent’s product portfolio has strong long-term research value. In contrast, $ANET operates a different business model. It does not just sell optical devices but directly provides data centers with high-speed Ethernet switching systems, network architecture, and related software. The company’s Q2 2026 revenue surpassed $3 billion for the first time, indicating cloud data center and AI network demand is converting into actual revenue. $MRVL is also worth studying, especially its custom chips, high-speed interconnect, and data center semiconductor business. However, this sector faces a very real issue: many companies have already been priced with high growth expectations. If order fulfillment is slower than expected or future expansion brings supply pressure, stock prices may experience significant valuation compression. I pay particular attention to opportunities for $LITE and $COHR after deep adjustments but will not assume valuations are cheap just because they have fallen consecutively. As for $AAOI, I will continue to observe, but its operating scale, profit realization, and stock volatility risks differ significantly from the more mature companies mentioned earlier. 5. Cloud Computing and AI Platforms: $GOOG, $MSFT, $AMZN If the market really continues to fall, I believe this sector may offer more balanced long-term opportunities than some highly elastic hardware stocks. $GOOG is one of the companies I am most willing to study long-term. Alphabet’s Q2 total revenue was $119.8 billion, up 24%, with Google Cloud revenue at $24.8 billion, up 82%. Many focus only on Gemini but overlook Google’s extremely large commercial ecosystem. Google Search, YouTube, Google Cloud, Android, Chrome, Workspace, and self-developed TPU together form the company’s competitive advantages. Even if AI models themselves gradually become commoditized, Google still has search distribution, advertising clients, cloud customers, and large existing business revenue. I especially value the synergy between Google Cloud and TPU. But it must be noted that AI is changing the competitive environment of search products, capital expenditure is rapidly expanding, and profit growth does not necessarily mean free cash flow improves simultaneously. Also, Alphabet’s recent partial book profits are affected by equity investment valuation gains, so non-recurring income should not be fully regarded as core operating profit when analyzing valuation. $MSFT is the core company in enterprise software and cloud computing. Q4 FY2026 revenue was $90 billion, up 18%, with operating profit of $40.6 billion. Azure, Microsoft 365, GitHub, enterprise databases, and security products form a large enterprise customer ecosystem. For Microsoft, AI’s greatest long-term value may not only be selling computing power but embedding AI functions into software products that enterprises are already willing to pay for continuously. However, its current stock price is relatively strong and does not mean October has necessarily provided a large enough valuation discount. $AMZN is also worth watching. Q2 AWS revenue was $42.2 billion, up 37%, with operating profit of $16.6 billion. Amazon has cloud computing, advertising, e-commerce logistics, and AI infrastructure businesses. But Amazon is also one of the companies with the most obvious AI capital expenditure expansion. Its free cash flow in the last 12 months has turned negative, and Q2 net profit was affected by large investment valuation gains. This means when researching Amazon, one cannot only focus on EPS growth but must seriously assess whether AWS revenue growth can support the expanding capital investment. If the market continues to sell off large tech companies due to AI capital expenditure fears, I will prioritize studying companies that can continuously generate operating cash flow from mature businesses. 6. Data Center Power, Cooling, and Energy Equipment: $VRT, $ETN, $GEV This is a direction I think is very easily overlooked in AI investment but is becoming increasingly important. Future large data centers face constraints not only in high-performance chip supply but also in power capacity, cooling efficiency, transformers, distribution systems, and grid connection speed. $VRT mainly provides data center power and thermal management infrastructure. In Q2 2026, the company’s revenue was $3.27 billion, up 24%, with adjusted operating margin of 22.6% and adjusted free cash flow of $925 million. Its products cover UPS, power distribution, liquid cooling, and critical data center infrastructure. As AI rack power density rises, the importance of cooling and power systems will continue to increase. $ETN is a more mature power management company. Eaton’s Q2 revenue grew 21% year-over-year, with electrical business backlog up 43%. Its value comes not only from AI data centers but also commercial buildings, power distribution, and industrial electrification. $GEV is also a direction worth studying. The company’s Q2 orders reached $24.2 billion, backlog about $176 billion, benefiting from demand growth in power generation equipment and grid infrastructure. With more data center construction, natural gas power equipment, transmission and distribution systems, and grid upgrades may face long-term demand. My attitude toward these three companies is clear: the industry direction is worth long-term optimism, but the market has already given some power equipment companies high valuations. Only when stock prices have fully adjusted and order and profit expectations remain stable can they be considered truly attractive opportunities. 7. Cybersecurity and Enterprise Software: $CRWD, $PANW This sector’s business model is completely different from semiconductors. Semiconductors are affected by capital expenditure cycles, inventory, and product prices, while mature enterprise security software relies more on subscriptions, customer renewals, and product expansion for revenue. $CRWD achieved $1.47 billion revenue in Q2 FY2027, up 26%, with annual recurring revenue of $5.84 billion, up 25%. Its Falcon platform covers endpoints, cloud environments, and various security scenarios. If enterprises further deploy AI Agents in the future, identity management, endpoint protection, cloud security, and automated threat response may become more important. $PANW has a product portfolio including network security, cloud security, and security operations. Q4 FY2026 revenue was $3.41 billion, up 34%, with next-generation security business annual recurring revenue reaching $9.1 billion, up 63%. Both companies have strong customer bases and recurring revenue but competition, acquisition integration, product security incidents, and stock-based compensation impacts cannot be ignored. Especially if a cybersecurity company experiences a serious product vulnerability, the market needs to reassess the impact on customer trust and renewals. Not every plunge caused by negative events should be seen as a buying opportunity. If the October tech stock adjustment continues to spread, causing valuation compression in quality cybersecurity companies, I think they are worth adding to the long-term watchlist. Finally, about my buying approach. I believe the stocks truly worth researching in this adjustment can be divided into three tiers. The first tier includes companies with global leading positions, extremely high competitive barriers, and mature profitable models. $TSM, $ASML, $NVDA, $AVGO, $GOOG. These companies occupy important positions in advanced manufacturing, semiconductor equipment, AI computing, custom chips, and large internet platforms. Their long-term value comes from real business, not just AI themes. The second tier includes companies with noticeable profit growth and cyclical elasticity. $MU, $SNDK, $LITE, $COHR, $ANET. These companies may achieve very high profit growth when industry demand expands continuously, but valuations are more sensitive to market expectations, supply-demand, and customer capital expenditure. The third tier includes long-term opportunities brought by AI infrastructure expansion. $VRT, $ETN, $GEV, $CRWD, $PANW. Their business models differ from semiconductors, helping investors avoid concentrating all funds in the same industry risk. Of course, I would never assume a company is cheap just because it’s on the list. Many semiconductor companies have risen astonishingly this year, and even with recent double-digit pullbacks, they may still be more expensive than last year. Especially in storage, companies must be evaluated using normalized cyclical profits rather than simply calculating P/E based on current extremely high profitability. What I truly look forward to now is the market giving these quality companies more reasonable prices. If the 10-year US Treasury yield remains high, oil prices do not fall, and corporate financing costs rise further, this adjustment may be far from over. Even a sharp short-term rebound in $SOXX cannot directly indicate a mid-term trend reversal. Therefore, I prefer to hold cash, build watchlists of quality companies in advance, and gradually establish long-term positions as macro risks ease, valuations normalize, and market structure improves, rather than investing all funds in the first phase of continuous decline. If the October decline continues to deepen, I will focus on three things: whether company profit forecasts truly worsen, whether valuations relative to normalized future profits are attractive enough, and whether the market begins to show sustained buying. If stock prices fall 20% but profit forecasts do not deteriorate and the company’s long-term competitive position remains unchanged, this situation is worth serious study. If stock prices fall 20% but prior profit expectations are simultaneously cut by 30%, it may still be expensive. This October adjustment may be painful for investors who chased AI stocks at high prices. But for investors with cash, able to endure volatility, and willing to observe companies over two to three years or longer, the market actively compressing tech stock valuations may not be a bad thing. I still believe AI will long-term change global computing infrastructure, enterprise software, and data center capital allocation. But how much value this industry can create in the future and how much investors are willing to pay today are two completely different questions. I will not accept any price just because AI is promising long-term. Nor will I deny companies with genuine technological barriers, sustained profitability, and long-term customer demand just because the market has fallen sharply. If October truly creates a golden opportunity, I hope to buy companies that can still expand competitive advantages over the next five years, not speculative stocks that rise the most in the next rebound. The list of good companies can be determined in advance. Truly worthwhile buying prices must be given by the market.

$SOXX has declined for 8 consecutive trading days as of October 15, an extremely rare trend in semiconductor history. I believe the likelihood of a sharp rebound is increasing, but I must emphasize that this rebound is likely the last good opportunity for bulls to reduce positions. $QQQ $SPY $MU $SNDK Observing the recent decline's market structure, I am increasingly inclined to think that the main selling pressure in semiconductors currently comes from existing long holders actively exiting, rather than a large-scale concentrated short attack. There is a fundamental difference between the two. During the July correction, hedge funds significantly increased their short exposure to AI-related stocks, and the large short covering demand provided momentum for the subsequent rebound. But the situation in October is completely different. JPMorgan's latest research points out that the market currently lacks the massive short base seen in July, and bottom-fishing buying is also clearly insufficient. This means that even if the market becomes extremely oversold, it may not replicate the rapid recovery seen in July. I believe this correction should be understood more from macro and valuation perspectives. The U.S. 10-year Treasury yield remains near 5%, and high interest rates are continuously raising financing costs for AI infrastructure; oil prices stay above $100 per barrel, further increasing inflation and monetary policy uncertainty; with the U.S. midterm elections approaching in November, the market still needs to digest future fiscal and policy environment changes. The coexistence of these factors means investors are no longer willing to pay the previously high valuations for long-term growth. Another easily overlooked issue is that the September rally itself had already priced in a lot of optimistic expectations in advance. I think the previous rise included obvious short covering, chasing capital, and valuation expansion, and once the short squeeze ended, the market returned to focusing on profit realization, capital expenditure returns, and financing costs—these real issues. The current decline largely reflects the valuation contraction from prior overexpansion. Even if TSMC and ASML report good earnings, it is difficult to change the capital flow direction of the entire sector in the short term. More worryingly, semiconductors have already fallen quite severely, but the adjustments in $QQQ and $SPY have not fully caught up. If macro pressures further transmit to other tech giants, semiconductors may face a second round of systemic selling pressure. Currently, I do not see sufficient evidence that this risk has been fully priced in by the market. Of course, the extreme 8-day decline also means the short-term trading structure is becoming fragile. As long as selling eases slightly, combined with some short-term short covering, there is a chance for a few days of sharp rebound. The problem is, if existing funds continue to exit and there is a lack of new capital to take over, such a rebound is likely just a technical correction within a downtrend. So my stance is clear: I remain long-term bullish on AI and semiconductors, but I will not conclude the market has bottomed just because of consecutive sharp declines in the short term. If a sharp rebound does occur next, but volume, capital flow, and overall market structure do not improve simultaneously, I am more inclined to use the rebound to reduce positions. In July, we saw rapid capital inflows after the decline; now, we see buying hesitation increasing and existing funds gradually withdrawing. In this environment, a rebound is possible, but it should not be mistaken for risk having been eliminated. For heavily positioned bulls, the next rebound may be more important than the next bottom.

$SOXX has already fallen for 8 consecutive trading days, and the semiconductor sector is experiencing an extremely rare continuous sell-off in history. I now actually believe that a rather fierce short squeeze rebound is very likely to occur next. But my attitude toward this rebound is very clear: this may be the last relatively decent opportunity for bulls to reduce their positions. $QQQ $SPY $MU $SNDK What does 8 consecutive days of decline mean? It means that a large number of bulls holding semiconductors have been forced to reduce their positions, and trend-following shorts have been continuously accumulating floating profits. The market is a two-way trade; shorts cannot hold their positions forever. When stock prices enter an extremely oversold area, as long as there is a catalyst that can improve sentiment even a little, profit-taking shorts will collectively cover, and short-term funds will rush to catch the rebound, potentially creating an extremely fierce rise in a very short time. It doesn't even require real fundamental improvement; just relief in selling pressure can cause a rebound lasting several trading days. BTIG has studied similar extreme semiconductor market conditions in history. In September 2002, $SOXX rebounded about 10% within 3 trading days, then continued to fall 33%; in October 2008, under similar circumstances, it rebounded about 20% in 3 days, then dropped 38%. I don't think history will necessarily repeat itself, but these cases precisely prove one thing: after extreme declines, violent rebounds can completely occur, and the larger the rebound, the easier it is to mistakenly believe the market has bottomed. This October's market situation is clearly different from July's. July's decline quickly attracted funds back in, but this time, even with strong earnings from TSMC and ASML, semiconductors continue to be under pressure. This indicates that current market concerns about AI capital expenditure, financing costs, and valuations have exceeded the extent that short-term earnings improvements can offset. Especially with U.S. Treasury yields remaining high and AI infrastructure financing risks not yet fully digested, the market lacks sufficient evidence to prove that sustained buying has returned. I even think that the semiconductor decline is already quite severe, but the overall pressure on $QQQ and $SPY has not yet fully reached the level of semiconductors. If subsequent risks spread from AI hardware to other tech giants, even if semiconductors rebound first, they may be dragged down again by a second round of market adjustments. Therefore, I do tend to believe that after continuous extreme declines, the probability of a rapid rebound triggered by short covering is rising. But I will not immediately change my mid-term market view just because $SOXX suddenly surges one day. If the rebound is not accompanied by sustained volume increase, higher lows, and overall relative strength improvement in the sector, I would be more inclined to proactively reduce risk exposure rather than chase the highs. I remain long-term bullish on AI and semiconductors, but now it is necessary to distinguish between good companies and good entry points. If there really is a fierce short squeeze next, I would rather treat it as a rare opportunity to retreat. Because more dangerous than missing a rebound is misjudging a rebound within a downtrend as the start of a new uptrend.

A batch of semiconductor stocks dropped 20% in two weeks, while consumer staples rose 7.6%. This divergence in the US stock market has become large enough to change the overall portfolio returns. If your watchlist is full of storage, optical modules, servers, and cloud computing power, what you've seen recently is almost continuous sharp declines. But looking at the entire market, you'll find that healthcare, consumer staples, energy, and utilities are all rising. In the same US stock market, investors with different holdings are experiencing completely different market conditions. Let's clarify the extent of the declines first. The following are all calculated based on closing prices from October 2 to October 15, excluding dividends. Among major semiconductor stocks, $ARM fell 20.4%, $AMAT fell 20.4%, $LRCX fell 20.2%, and both $LSCC and $INTC fell about 20.1%. From chip architecture licensing to manufacturing equipment, then to FPGA and processors, multiple segments have seen about a 20% pullback. The breadth of this sell-off is hard to explain solely by competition news about any single chip. Within storage, $SNDK fell 18.9% and $MU fell 13.4% over the same period. They indeed fell painfully, but Micron is not the worst performer among these mainstream semiconductor stocks. Conversely, $NVDA fell 7.1% and $AVGO fell 3.0%, showing significantly more resilience. Grouping all stocks simply as "AI concept" ignores the huge differences in each company's profit structure, starting valuation, and market expectations. Looking along the industry chain, losses in optical communications are even more striking: $AAOI fell 27.6%, $COHR fell 25.6%, and $LITE fell 19.9%. The first two have already lost more than a quarter of their value during this period. The previously favored high growth and high elasticity also amplify portfolio losses when risk appetite weakens. After a 25% drop, a roughly 33.3% rise is needed to return to the original level, so a daily rebound of a few percentage points may look strong but might only recover a small portion of previous losses. On the other hand, during the same period, consumer staples ETF $XLP rose 7.6%, healthcare ETF $XLV rose 6.6%, energy ETF $XLE rose 6.0%, and utilities ETF $XLU rose 4.6%. Materials ETF $XLB rose 3.5%, and financials ETF $XLF rose 3.3%. For comparison, semiconductor ETF $SOXX fell 12.6%. The gap in returns between consumer staples and semiconductors widened by about 20.2 percentage points in just around two weeks. These numbers indicate that the current market still has strong sector rotation characteristics. AI hardware continues to be under pressure, while other sectors are still achieving higher pricing. Demand for consumer staples and healthcare is less related to tech companies' expansion budgets; energy depends more on oil and gas supply-demand and profitability; financials require analysis of bank earnings and credit quality. The logic behind these sectors' rises differs, yet they collectively provide a buffer for the index. My understanding is that the market is re-evaluating how fast AI capital expenditures can sustain growth and when these investments will convert into cash returns. Equipment, storage, optical communications, and cloud computing may seem like different businesses but ultimately may rely on a few customers' investment budgets. Although there are many stocks, their revenue drivers are highly correlated, so concentration risk remains. This is why I previously started focusing on healthcare and traditional consumer sectors: the portfolio needs assets that can profit from different demand drivers. However, how long this rotation can support the index still needs observation. If semiconductors gradually stabilize and sectors like consumer staples and healthcare can maintain their gains, the market has a chance to digest adjustments through internal rotation. If these strong sectors also start to decline continuously while equipment, storage, and optical communications keep hitting new lows, the previous buffer in the market will thin, increasing the risk of a second phase of catch-up declines. I still do not rule out a sharp rebound in semiconductors, but mid-term trend improvement requires broader stabilization in equipment, storage, and optical communications. A rally in one or two stocks is not enough to cover the weakness across the entire industry chain. What is worth being cautious about now is that semiconductors have already suffered heavy losses, while the broader index is still supported by other sectors. Whether subsequent risks can be absorbed depends crucially on whether this layer of support can continue.

$TSM's net profit grew nearly 69% year-over-year, $ASML delivered strong earnings, yet the entire semiconductor sector continues to plunge consecutively. The strangest aspect of this market is that even the core industry leaders cannot drive a sector rebound. $SOXX $QQQ $NVDA $MU TSMC's Q3 revenue was about $46.7 billion, with net profit reaching 763.1 billion New Taiwan dollars and a gross margin of 66.4%. ASML also benefits from demand for advanced process equipment, and management maintains a positive outlook on orders for the coming years. These two companies are almost at the core of global advanced semiconductor manufacturing. If even their performance cannot change the market trend, then the recent decline cannot be simply explained by fundamentals. The entire market is undergoing a clear capital reallocation. As of the week ending October 14, U.S. equity funds still saw net inflows overall, but technology sector funds experienced net outflows of about $5.2 billion. Money hasn't completely left U.S. stocks; it has just exited the previously most crowded tech trades. This is why I have repeatedly emphasized that the real major drop in October may not have arrived yet. Currently, $SPY still has support from financials, healthcare, and other sectors, but semiconductor stocks like $SOXX, $MU, and $NVDA have been continuously sold off. The mere withdrawal of existing capital from the tech sector has already caused such a rare consecutive decline. If other sectors also start to lose support later, the entire U.S. stock market could enter a second phase of risk release. I believe it is necessary to distinguish between July and October now. July was characterized by crowded trades quickly unwinding, followed by rapid capital inflows that formed a strong rebound. But after extreme overselling in October, the market still lacks sustained buying. Positive news cannot drive prices up, and brief rebounds are repeatedly sold off. This pattern is closer to the capital structure seen in March and parts of 2022. The market does not necessarily need a sharp decline in corporate earnings to crash. Suppose a company's EPS grows 20%, but the market compresses its PE from 40x to 25x; the stock price can still fall significantly. The market is now recalculating the return on AI capital expenditures, not just focusing on how much next quarter's revenue can grow. I still believe semiconductors have the conditions for a strong oversold rebound, but there is currently insufficient evidence to prove the mid-term downtrend has ended. Especially with $TSM and $ASML reporting strong earnings, yet the sector still cannot stabilize, this failure of positive news to take effect requires high vigilance. My judgment remains unchanged: continue to control positions for the rest of October and do not rush to go all-in on bottom fishing. The most dangerous point now is not that semiconductors have fallen so much, but that semiconductors have fallen this far and the entire U.S. stock market may not have truly started its catch-up decline yet.

Some bond yields of $CRWV have already broken through 15%, and Wall Street is beginning to reassess the financing risks of the entire AI data center sector. I believe this issue is far more alarming than the continuous plunge in semiconductor stocks. $NVDA $NBIS $AVGO A batch of bonds issued by CoreWeave in June this year had a yield of about 9.675% at issuance, but now the secondary market yield has risen to around 15%. Although this does not mean that the actual interest on all the company's debt has become 15%, it at least indicates that the credit market is demanding higher risk compensation. I have previously emphasized that the real major drop in October may not have arrived yet. Currently, market attention is focused on the decline in semiconductor stock prices, but the financing issues behind AI infrastructure may be the key to the subsequent spread of risk. Over the past two years, the entire AI industry chain has relied on massive capital expenditures to drive growth. GPUs, servers, liquid cooling, power, and data center construction—all these segments require continuous funding. $NVDA can sell GPUs to customers, but computing power rental companies like $CRWV need to first bear substantial equipment investments and then recover costs through rental income over the coming years. Once financing costs continue to rise and project returns decline, some companies may be forced to slow down expansion. This is where my real concern lies. Now, with semiconductors continuously plunging and bulls failing to form effective support, the credit market is starting to show pressure again. If downstream computing power companies cut capital expenditures, the forward growth expectations of upstream $NVDA, $AVGO, and $MU could also be affected. Of course, there is currently no evidence that AI demand has collapsed, and the financial strength of large cloud providers is far stronger than that of highly leveraged computing power rental companies. But AI demand growth and AI investments generating sufficient profits are two completely different matters. I still believe that the risks in October may not have been fully released yet. Currently, semiconductors mainly reflect the withdrawal of existing funds, but if credit spreads widen further and financing conditions tighten, risks could spread from highly indebted AI companies to the entire tech sector. It is especially necessary to be cautious after $QQQ continues to weaken and the relatively resilient $SPY also begins to come under pressure. The real danger is not that AI companies temporarily face more expensive financing, but that the market suddenly realizes some AI expansion plans relying on high leverage can no longer proceed at the original cost. For the rest of October, I still lean towards a defensive stance. Whether a real major drop will come is still uncertain, but the warnings from the credit market cannot be ignored.