As the Philadelphia Semiconductor Index enters a bearish phase: “Kimi Moment” hits Wall Street!
[As the Philadelphia Semiconductor Index enters a bearish phase: “Kimi Moment” hits Wall Street! 】It is worth noting that Friday’s sell-off in U.S. chip stocks also occurred after Chinese artificial intelligence startup Dark Side of the Moon released a new model, Kimi K3. The company said its new Kimi K3 model is comparable to the strongest offerings from OpenAI and Anthropic — largely fueling Wall Street concerns about Silicon Valley’s previous overspending.
This Friday (July 17), the Philadelphia Semiconductor Index, which has attracted much attention in the global technology community, officially fell into a technical bear market. Prior to this, this frenetic bull market set off by memory chips had led the index to double in just three months, but now, the early astonishing gains have suffered a sharp retracement... Market data shows that the Philadelphia Semiconductor Index fell 5.7% overnight. Although it received an influx of funds in late trading and recovered most of its losses, it still closed down 1.6%. The day's decline was still enough for the index to retrace 20% from its historical high at the end of June, reaching the technical threshold indicating that it has entered a bear market. In addition, the index's cumulative decline this week reached nearly 10% - its largest weekly decline since Trump's tariffs triggered a market rout in April 2025. The 30-stock chip benchmark index surged 105% between its March low and last month's high. But today, popular stocks such as Marvell Technology Inc., Arm and Intel have all plunged more than 30% since their respective highs. The Nasdaq Composite Index, which mainly tracks technology stocks, ended down 1.4% on Friday, extending Thursday's 1.5% decline and extending a sell-off that has wiped out tens of billions of dollars in the value of chip and storage stocks. The S&P 500 ended the day down 1%. Obviously, the current momentum of stock prices has seriously overdrawn the expansion speed of fundamentals. Although the fundamentals of the industry are still good and the overall trend is still upward, the previous trend of the index was basically an extreme parabolic market. James Abate, managing director and head of fundamental strategy at Horizon Investments, said when talking about this change. Analysts said the sell-off was driven by renewed concerns about the health of the artificial intelligence sector and rising questions about whether tech giants can still pour trillions of dollars in unbridled capital spending into AI infrastructure in the future. Investors are also weighing the current lofty valuation levels and questioning whether stock prices are rising too high, too fast. "Despite strong fundamentals, market expectations have been well ahead of fundamentals," said Muneeb Muzaffar, senior portfolio manager at Bold Wealth Partners. "P/E ratios already factor in the most optimistic scenarios, which raises the question: Have the best parts of the story been fully priced in by the market? How much premium are investors willing to pay for chip companies' earnings? Is it possible that these expectations will be revised downwards?" As quarterly earnings reports from major technology giants are released, more key information about the AI investment boom will emerge in the coming weeks. Alphabet, the parent company of Google, is one of the first companies to watch, and its earnings will be released after the market close on July 22. Microsoft Corp., Amazon.com Inc. and Meta Platforms Inc. are also scheduled to release earnings next week. "The market is in dire need of seeing clear signals to ensure that the technology giants' AI return on investment (ROI) is sustainable. As long as these expenditures are justified, they will continue to invest, and chip companies will also benefit from it," Muzaffar said. “However, when you overlay valuations, sentiment and capital flows on top of positive fundamentals, the picture becomes a bit complicated.” As one of this year's best-performing stocks, the storage maker's stock price has been hardest hit in this round of decline, as the market is buzzing around "when supply will finally keep up with soaring demand" and "what impact this will have on swelling profits." Micron Technology's stock price has plummeted 30% from its peak, while Western Digital and SanDisk's stock prices have fallen by more than 35%. Samsung Electronics also suffered a share price plunge in early July after the South Korean company reported a 19-fold surge in quarterly profits, but the increase was still not enough to meet investor expectations given that the stock price has risen 150% since the beginning of the year.
TSMC, Nvidia’s core foundry, also failed to escape the same fate this week. Although the company has just raised its full-year capital expenditure and revenue guidance, its stock price still recorded seven consecutive trading days of decline, setting a record for the longest losing streak since July 2022, during which the market value evaporated by 8.8%. Momentum trading suffers backlash Many traders also view the recent losses suffered by popular chip stocks as a sudden reversal in so-called momentum trading. Momentum trading is a strategy in which traders and investors bet that the best-performing stocks will continue to extend their gains. "We are dealing with one of the largest momentum sell-offs on record," said Christian Mueller-Glissmann, head of asset allocation research at Goldman Sachs. "It's been three weeks of a washout." Momentum trading is popular among hedge funds, but it runs the risk of backlash if the market changes direction. A Bloomberg index tracking the performance of momentum strategies has fallen 13% since the broader tech frenzy peaked in June. Analysts say a popular trading strategy among hedge funds and mutual funds this year has been to be long chip stocks while shorting so-called hyperscalers. "The market's optimism about semiconductors was widely sought after ... but now it may be a painful trade for carry traders and mutual funds," Mueller-Glissmann said. In fact, this week's losses came despite strong earnings reports from ASML and TSMC earlier this week. Michael Zigmont, co-head of trading at Visdom Investment Group, said the market had "reacted strongly" to TSMC's forecast of increased capital spending in the coming years, raising concerns about corporate overinvestment. "What this tells us is that even if performance is strong and the outlook is bright, investors may still be choosy about the status quo...perhaps investors are just looking for an excuse to sell certain stocks," Zigmont added. Max Kettner, chief multi-asset strategist at HSBC, said "violent unwinding" in momentum trading could continue if chipmakers fail to meet high profit expectations. 'Kimi moment' alarms Wall Street It is worth noting that Friday’s sell-off in U.S. chip stocks also occurred after Chinese artificial intelligence startup Dark Side of the Moon released a new model, Kimi K3. The company said its new Kimi K3 model is comparable to the strongest offerings from OpenAI and Anthropic — largely fueling Wall Street concerns about Silicon Valley’s previous overspending. Jefferies analyst Matt Ma noted in a note on Friday that the 2.8 trillion-parameter open-weight model performs on par with Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol, and is more advantageous due to its lower price—it costs about 40% less than similar top U.S. models. “Whatever gap once existed between China and the U.S. in cutting-edge AI has now narrowed significantly, and this change happened on a morning when Wall Street was busy convincing itself that the economic benefits of artificial intelligence were unattainable,” Mark Malek, chief investment officer at Siebert Financial, said in emailed comments. Mark Hackett, chief market strategist at Nationwide, said Friday's news from Kimi actually forced investors to quickly unwind a large number of technology stocks, not just chip stocks. It is reported that Kimi K3 is the world's first open source 3 trillion level model, designed for cutting-edge intelligent scenarios such as long-range programming, knowledge work and reasoning. According to Dark Side of the Moon, the research and development focus of K3 is not to simply expand the parameter scale, but to further transform the scale advantage into capability improvement through collaborative optimization of model architecture, training methods, etc.
In the Arena Comprehensive Text Ranking, Kimi K3 currently tops the front-end code rankings with 1,679 points, ahead of Anthropic's Claude Fable 5 (1,631 points) and OpenAI's GPT-5.6 Sol (1,618 points). The ranking is based on approximately 484,000 blind pair-wise tests of 98 models. Given that K3’s predecessor ranked 18th – Dark Side of the Moon therefore achieved a significant leap in performance in a very short period of time. The ability to code AI models is significant—because this is one of the first AI markets with clear business value, measurable output, and the ability to directly automate paid work. This leading model in interface development, capable of running autonomously for hours and at a lower cost than its closest competitors, is competing in one of the most economically valuable areas in the industry. Some on Wall Street also compared this decline to the market turmoil caused by DeepSeek's breakthrough last year. The term "Kimi Moment" was circulating on Wall Street and Silicon Valley almost immediately on Friday - a reference to the DeepSeek shock of early 2025, when investors were forced to reconsider just how much money and computing power it took to build a competitive model. Some industry insiders say the arrival of the "Kimi moment" indicates that the cushion may be crumbling much faster than expected. "The whole game has changed. I expect this will trigger a red alert for some people," predicted AI analyst Kim Isenberg. Rich Privorotsky, head of Goldman Sachs' equity business, also said that for enterprises, governments and developers, a model with near-leading performance, 40% lower cost, and that can be customized or run in-house may be a more attractive option. The very existence of Kimi K3 puts pressure on the U.S. AI lab's pricing power, the massive valuations built around its technological superiority, and its rationale for spending hundreds of billions of dollars building ever-larger data centers. As well-known venture capitalist Chamath Palihapitiya succinctly summed it up: "The math simply doesn't work" - the same purchase of 1 million Tokens with a similar performance difference costs only $0.50 on one side and $56 on the other. In the long run, if the decline in computing power costs allows some companies to deliver the same AI capabilities with less capital, thereby bringing free cash flow in advance, then they will most likely be the winners. In this sense, the downward revision of hardware valuations may simply be the market’s response to lower future computing power costs and a return of value to platforms. (
TSMC, Nvidia’s core foundry, also failed to escape the same fate this week. Although the company has just raised its full-year capital expenditure and revenue guidance, its stock price still recorded seven consecutive trading days of decline, setting a record for the longest losing streak since July 2022, during which the market value evaporated by 8.8%. Momentum trading suffers backlash Many traders also view the recent losses suffered by popular chip stocks as a sudden reversal in so-called momentum trading. Momentum trading is a strategy in which traders and investors bet that the best-performing stocks will continue to extend their gains. "We are dealing with one of the largest momentum sell-offs on record," said Christian Mueller-Glissmann, head of asset allocation research at Goldman Sachs. "It's been three weeks of a washout." Momentum trading is popular among hedge funds, but it runs the risk of backlash if the market changes direction. A Bloomberg index tracking the performance of momentum strategies has fallen 13% since the broader tech frenzy peaked in June. Analysts say a popular trading strategy among hedge funds and mutual funds this year has been to be long chip stocks while shorting so-called hyperscalers. "The market's optimism about semiconductors was widely sought after ... but now it may be a painful trade for carry traders and mutual funds," Mueller-Glissmann said. In fact, this week's losses came despite strong earnings reports from ASML and TSMC earlier this week. Michael Zigmont, co-head of trading at Visdom Investment Group, said the market had "reacted strongly" to TSMC's forecast of increased capital spending in the coming years, raising concerns about corporate overinvestment. "What this tells us is that even if performance is strong and the outlook is bright, investors may still be choosy about the status quo...perhaps investors are just looking for an excuse to sell certain stocks," Zigmont added. Max Kettner, chief multi-asset strategist at HSBC, said "violent unwinding" in momentum trading could continue if chipmakers fail to meet high profit expectations. 'Kimi moment' alarms Wall Street It is worth noting that Friday’s sell-off in U.S. chip stocks also occurred after Chinese artificial intelligence startup Dark Side of the Moon released a new model, Kimi K3. The company said its new Kimi K3 model is comparable to the strongest offerings from OpenAI and Anthropic — largely fueling Wall Street concerns about Silicon Valley’s previous overspending. Jefferies analyst Matt Ma noted in a note on Friday that the 2.8 trillion-parameter open-weight model performs on par with Anthropic’s Fable 5 and OpenAI’s GPT-5.6 Sol, and is more advantageous due to its lower price—it costs about 40% less than similar top U.S. models. “Whatever gap once existed between China and the U.S. in cutting-edge AI has now narrowed significantly, and this change happened on a morning when Wall Street was busy convincing itself that the economic benefits of artificial intelligence were unattainable,” Mark Malek, chief investment officer at Siebert Financial, said in emailed comments. Mark Hackett, chief market strategist at Nationwide, said Friday's news from Kimi actually forced investors to quickly unwind a large number of technology stocks, not just chip stocks. It is reported that Kimi K3 is the world's first open source 3 trillion level model, designed for cutting-edge intelligent scenarios such as long-range programming, knowledge work and reasoning. According to Dark Side of the Moon, the research and development focus of K3 is not to simply expand the parameter scale, but to further transform the scale advantage into capability improvement through collaborative optimization of model architecture, training methods, etc.
In the Arena Comprehensive Text Ranking, Kimi K3 currently tops the front-end code rankings with 1,679 points, ahead of Anthropic's Claude Fable 5 (1,631 points) and OpenAI's GPT-5.6 Sol (1,618 points). The ranking is based on approximately 484,000 blind pair-wise tests of 98 models. Given that K3’s predecessor ranked 18th – Dark Side of the Moon therefore achieved a significant leap in performance in a very short period of time. The ability to code AI models is significant—because this is one of the first AI markets with clear business value, measurable output, and the ability to directly automate paid work. This leading model in interface development, capable of running autonomously for hours and at a lower cost than its closest competitors, is competing in one of the most economically valuable areas in the industry. Some on Wall Street also compared this decline to the market turmoil caused by DeepSeek's breakthrough last year. The term "Kimi Moment" was circulating on Wall Street and Silicon Valley almost immediately on Friday - a reference to the DeepSeek shock of early 2025, when investors were forced to reconsider just how much money and computing power it took to build a competitive model. Some industry insiders say the arrival of the "Kimi moment" indicates that the cushion may be crumbling much faster than expected. "The whole game has changed. I expect this will trigger a red alert for some people," predicted AI analyst Kim Isenberg. Rich Privorotsky, head of Goldman Sachs' equity business, also said that for enterprises, governments and developers, a model with near-leading performance, 40% lower cost, and that can be customized or run in-house may be a more attractive option. The very existence of Kimi K3 puts pressure on the U.S. AI lab's pricing power, the massive valuations built around its technological superiority, and its rationale for spending hundreds of billions of dollars building ever-larger data centers. As well-known venture capitalist Chamath Palihapitiya succinctly summed it up: "The math simply doesn't work" - the same purchase of 1 million Tokens with a similar performance difference costs only $0.50 on one side and $56 on the other. In the long run, if the decline in computing power costs allows some companies to deliver the same AI capabilities with less capital, thereby bringing free cash flow in advance, then they will most likely be the winners. In this sense, the downward revision of hardware valuations may simply be the market’s response to lower future computing power costs and a return of value to platforms. (