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

300Following

690followers

Feed

SK Hynix has dropped more than 30% since October, but the Hong Kong stock 7709 has plummeted nearly 60%. This is why I have always emphasized that ordinary investors should avoid leveraged ETFs if possible. Many people think buying leveraged ETFs only amplifies gains, but when facing continuous declines, they realize the risks of these products are far more severe than imagined. 7709 is a daily leveraged product tracking SK Hynix under Southern East Ying, currently using a maximum 2x dynamic leverage mechanism. On October 2, 7709 closed at HKD 42.20, but by October 20 it had fallen to around HKD 17, a nearly 60% drop in just half a month. During the same period, SK Hynix’s Korean stock fell from 1,841,000 KRW to about 1,223,000 KRW, a drop of about 33.6%. Even more dramatic, on October 15, 7709 plunged 24.31% in a single day, losing nearly a quarter of its principal in one day. Many don’t understand why 7709 could fall so much when SK Hynix itself hasn’t halved in value. Because leveraged ETFs don’t simply multiply the stock’s long-term returns by two. They track daily price changes and adjust leverage exposure according to market fluctuations. Continuous declines compress principal, and in choppy markets, even if the underlying stock eventually returns to its starting point, the leveraged ETF can still incur losses due to daily compounding effects. Here’s a simple example. A stock falls from 100 to 90, a 10% drop, then rises 11.11% back to 100. But if a daily fixed 2x leveraged product starts at 100, it falls 20% to 80 on day one, then rises 22.22% on day two, ending at only 97.78. The underlying stock breaks even, but the leveraged product loses 2.22%. This doesn’t even account for management fees, financing costs, or tracking errors. Looking at 7709’s actual case: if it falls from HKD 42.20 to HKD 17.36, that’s a cumulative loss of about 58.9%. To return to the original price, it needs to rise about 143%. The underlying stock dropped 33.6%, so to break even it needs to rise about 50.6%. The difficulty of recovering losses is not comparable. What’s worse, products like 7709 don’t automatically gain better long-term value just because the price has fallen sharply. Even if you are bullish on SK Hynix’s HBM demand in the coming years, multiple rounds of severe volatility in between can cause the final investment returns to lag far behind simply holding the original stock. It’s important to distinguish: buying leveraged ETFs usually doesn’t require margin calls like margin accounts, but that doesn’t mean your principal is safe. The product price can keep falling, and in extreme cases, investors may lose most of their invested capital. Leveraged ETFs are not completely useless. For traders who can strictly control position size, holding time, and risk exposure, they can be used as short-term tools. But treating them as ordinary ETFs for long-term holding, especially in highly volatile sectors like semiconductors, significantly amplifies risk. Long-term growth in the semiconductor industry is different from short-term stock price increases. Even if SK Hynix, $MU, and $NVDA enter an uptrend again, it doesn’t mean investors who bought leveraged products at high prices can easily break even. Investing is not about who earns the most during uptrends, but about who can preserve enough principal after wrong judgments and extreme volatility to participate in the next opportunity.

Yesterday, the US semiconductor stocks rebounded sharply, but I still insist that this rally is only a technical correction and should not be taken as the market having bottomed out. The performance of semiconductors in South Korea, Japan, and Taiwan today further indicates that the selling pressure on the global AI industry chain has not ended. $SOXX $MU $SNDK $NVDA $QQQ On October 19, the Philadelphia Semiconductor Index rose 1.61%, ending a streak of eight consecutive trading days of decline. $MU rose about 3%, $SNDK rose 5.5%, and $NVDA rose 1.4%. However, one detail is worth noting: the Philadelphia Semiconductor Index reached an intraday high of 11,551 points but ultimately closed at 11,316 points, significantly below the opening level. Even though the overall close was higher that day, there was still considerable selling pressure at the intraday highs. By October 20, the Asia-Pacific markets did not continue the US stock rebound. South Korea's KOSPI closed down 3.73%, SK Hynix fell 4.23%, and Samsung Electronics dropped 2.98%. Japan's Kioxia fell about 3.2%, and Taiwan's TSMC dropped about 0.6%. While US memory stocks collectively rose yesterday, Asia's leading memory stocks came under pressure again today. The US stock rebound the previous night did not receive sustained support from the major semiconductor markets in Asia. US pre-market also showed divergence. As of early morning October 20 Eastern Time, Nasdaq 100 futures weakened slightly, and S&P 500 futures were basically flat. Compared to the general tech stock gains in yesterday's pre-market, the market is clearly more cautious today. However, the slight fluctuations in futures alone are not enough to confirm that US stocks will continue to decline. Why does this rebound still lack conviction? Because the issues that caused the semiconductor decline earlier have not been resolved. AI infrastructure financing costs remain high, and the market is beginning to reassess whether massive capital expenditures can deliver sufficient profits. Meanwhile, Brent crude remains near $110, and the US 10-year Treasury yield stays above 5%, with the high interest rate environment continuing to pressure growth stock valuations. More critically, market breadth is a concern. Although Nasdaq rose 1.11% on October 19, the number of declining stocks on the New York Stock Exchange and Nasdaq still exceeded the number of advancing stocks. The index rebound mainly relied on some heavyweight tech stocks rather than a broad recovery in market risk appetite. The semiconductor sector's eight consecutive days of decline were followed by a short-term short-covering and oversold rebound, which is completely normal. But a true trend reversal requires continuous gains, breaking through key resistance, pullbacks with support, and a gradual strengthening of the global semiconductor industry chain. These conditions have not yet been fully confirmed. Next, attention should be paid to whether $SOXX can regain and hold above its short-term moving averages, and whether $MU and $SNDK will break below previous lows again. If Korean memory stocks continue to be under pressure and US memory stocks fail to sustain their rebound, the risk of a secondary decline will increase. Of course, if US stocks can absorb the selling pressure from Asia-Pacific, continuously reclaim key resistance levels, and the number of advancing stocks increases significantly, then the short-term cautious view should be revised. Long-term optimism about AI infrastructure and the memory industry does not mean one must be fully invested in the short term. After this October decline, the market needs improvements in earnings expectations, credit conditions, and capital support collectively, not just a single day’s rally. Yesterday’s gains do not prove a market reversal, and today’s Asia-Pacific decline alone does not prove that the next sharp drop is inevitable. The current price structure is more consistent with a technical rebound after a decline, and a trend reversal still requires more evidence.

Wall Street has just released a rather astonishing set of data: the entire S&P 500 index is expected to see nearly 30% earnings growth in the third quarter, with a significant portion coming from a few semiconductor companies. $NVDA, $MU, and $AVGO are supporting the earnings expectations for U.S. stocks, but this also means that if the growth of a few core companies falls short of expectations, the entire market could face a revaluation. According to Goldman Sachs data cited by Axios on October 14, $MU, $NVDA, and $AVGO are expected to contribute nearly 40% of the S&P 500's third-quarter earnings growth. This means that the profit increase of the index composed of 500 listed companies is highly concentrated in three semiconductor companies. More notably, the latest FactSet statistics on October 19 show that the overall earnings of the S&P 500 in the third quarter are expected to grow 29% year-over-year, but if $MU and $NVDA are excluded, the earnings growth rate of the remaining companies drops to 20%. The divergence within the tech seven giants is even more apparent. FactSet expects MAG7's third-quarter earnings growth to be 33.2%, but excluding $NVDA and $META, the remaining five companies' earnings growth rate is only 7.5%. In other words, even among the world's strongest tech companies, the number of firms truly bearing the main earnings growth burden is fewer than imagined. This explains why the entire U.S. stock market was so sensitive when semiconductors continuously declined in October. $MU benefits from rising prices of HBM, DRAM, and storage; $NVDA benefits from AI accelerators and computing infrastructure expansion; $AVGO benefits from custom AI chips and high-speed networking businesses. These companies' profitability is indeed improving, but the market had already priced in quite high growth expectations in advance. Stock price increases require not just companies making money, but actual earnings consistently exceeding market expectations. When a company's valuation already includes several years of rapid growth, even if quarterly profits grow 50% year-over-year, the stock price can still plunge if management lowers future guidance. However, concentrated earnings do not mean other U.S. companies are in decline. FactSet expects the 493 companies excluding MAG7 to still see 27.7% earnings growth in the third quarter, indicating that profit expansion is not entirely limited to the AI sector. In fact, in the fourth quarter, the market expects these 493 companies' earnings growth rate to possibly surpass MAG7 for the first time in nearly four years. This is the most important change to watch in the next phase. If semiconductors continue to maintain high growth while industries like industrials, finance, and consumer sectors start contributing more profits, U.S. stocks have a chance to break free from overreliance on a few tech companies. But if earnings expectations for $MU, $NVDA, and $AVGO begin to be revised downward and other sectors cannot take over growth, the S&P 500 may face dual pressure on earnings expectations and valuation. The semiconductor adjustment in October has already shown that Wall Street does not disbelieve the future demand for AI, but is beginning to reassess how high these demands can support valuations. When the earnings performance of a few companies is enough to influence the growth expectations of the entire index, studying semiconductors is no longer just studying an industry, but studying the core earnings pillar of the entire U.S. stock market.

The semiconductor sector just went through a fierce sell-off in October, yet Jensen Huang is preparing to take the stage personally on October 21. $NVDA, $MU, $AVGO, $TSM, and the entire AI industry chain may迎来 one of the most important technical catalysts of the week. NVIDIA has confirmed that the 2026 GTC Berlin conference will be held from October 21 to 22, with Jensen Huang delivering the keynote speech on the 21st from 11 a.m. to 1 p.m. German time. Notably, this GTC is not just focused on GPU performance; AI factories, robotics, industrial AI, inference computing, and European sovereign AI are all included in the core agenda. Even more interestingly, on October 19, Jensen Huang attended an investment summit in Berlin alongside German Chancellor Scholz. The German government and NVIDIA signed a cooperation intention document to promote industrial AI development, focusing on Physical AI, which means integrating AI into robots, automobiles, and real industrial production systems. Why is this important? Over the past two years, the market’s valuation of AI has mainly been based on large tech companies continuously purchasing GPUs and expanding data centers. $MSFT, $GOOG, $META, and $AMZN have been increasing capital expenditures, directly driving profit growth for companies like $NVDA, $TSM, and $MU. However, this growth model has its issues. If AI investment remains concentrated in a few cloud computing giants for the long term, the market will sooner or later question whether capital expenditures can be sustained and when massive investments will generate sufficient cash returns. Industrial AI may offer an alternative growth path. Germany has a large automotive, machinery, and industrial automation system. If AI truly enters factories, robotics, autonomous driving, and industrial simulation, future computing power demand will no longer rely solely on chatbots and large language model training. For the industry chain, $NVDA benefits from GPUs, software platforms, and robotics ecosystems; $TSM is responsible for advanced chip manufacturing; $MU and $SKHY provide high-bandwidth memory; and $AVGO may gain incremental opportunities through network connectivity and custom chip demands. However, the implementation of industrial AI does not necessarily mean all these companies will simultaneously receive new orders. Whether this conference can become a catalyst for a semiconductor rebound depends on whether Jensen Huang can present business progress that exceeds market expectations. If it’s just showcasing robotics and promoting AI application scenarios, with actual orders and revenue realization still unclear, the speech’s hype may not sustain a continuous stock price rise. Conversely, if there are clear new customers, large-scale computing deployment plans, and verifiable commercialization progress, the market has reason to reassess AI’s demand potential over the next few years. Semiconductors have already experienced significant valuation compression in October. What the market currently lacks is not another AI concept but new evidence proving that future profits will continue to grow. October 21 is worth watching, but one tech conference should not be equated with a market reversal. Even if Jensen Huang brings good news, if $SOXX cannot continuously recover from previous declines and $NVDA does not form a new upward trend, short-term gains may still only be event-driven rebounds.