last night! U.S. chip stocks collectively plummeted! Storage concept stocks, sudden changes in late trading!
Last night and this morning, the three major U.S. stock indexes fell across the board. The Nasdaq and S&P 500 both fell by more than 1%. Chip stocks collectively fell. The Philadelphia Semiconductor Index closed down 1.6%. The latest point has retraced 20% from the record high, entering a technical bear market. U.S. storage concept stocks rebounded sharply during the session, but their stock prices fell sharply in late trading. Some analysts pointed out that the market's doubts about the return on AI (artificial intelligence) capital expenditures and the sustainability of chip demand have rekindled, leading to continued sell-offs in the highly valued and crowded AI sector, and some active fund managers have begun to reduce exposure to related assets. The three major U.S. stock indexes closed down collectively On July 17, Eastern Time, the three major U.S. stock indexes fell collectively. As of the close, the Dow fell 0.77%, the Nasdaq fell 1.4%, and the S&P 500 fell 1.01%. Large technology stocks generally fell, with SpaceX falling more than 5%, and the 13th Starship test flight triggered an abort procedure less than a second before launch; Nvidia, Google, Meta, and Tesla all fell more than 2%, Amazon and Microsoft fell more than 1%; Apple rose 0.14%. U.S. chip stocks fell sharply across the board, with the Philadelphia Semiconductor Index falling more than 5% at the beginning of the session and finally closing down 1.63%. Among component stocks, Applied Materials fell more than 5%, TSMC ADR, Texas Instruments, Lam Group fell more than 2%, and Intel fell 2%. Storage concept stocks rebounded sharply during the session, but their stock prices fell back in late trading. As of the close, SK Hynix rose by 1.13%, and once rose by more than 9% during the session; Micron Technology fell by 0.5%, and once rose by more than 5% during the session; SanDisk fell by more than 3%, and once rose by more than 6% during the session; Seagate Technology rose by more than 5%, and Western Digital rose by more than 2%. As the market begins to reassess the investment boom in the field of AI, some investors are beginning to bet that the growth of AI spending may slow down. Analysis shows that some active fund managers have begun to reduce their exposure to related assets. Goldman Sachs TMT trading experts pointed out that the anxiety of US stock investors has increased significantly in the past 48 to 72 hours, and anxiety surrounding the continued correction of AI and semiconductors is intensifying. Judging from the options market, major semiconductor ETFs have experienced extreme bearish demand this week, with the put-call volume ratio exceeding 3 times, much higher than the historical average. Nomura Charlie McElligott said that all effective strategies this year are running in the opposite direction, and any further deterioration may trigger further selling, with bulls forced to close their positions and shorts will increase their positions. Judging from the recent market performance of U.S. stocks, the technology sector of the U.S. stock market is experiencing the most obvious internal style switch this year - funds are shifting from the most crowded AI infrastructure transactions to platform companies with more stable cash flows. The second-quarter earnings season for U.S. stocks is still in its early stages, with 49 S&P 500 index companies having announced their results. According to LSEG data, 90% of the companies' profits exceeded market expectations. Ryan Detrick, chief market strategist at Carson Group, pointed out that the earnings season has just begun, but it is currently performing very strongly. The banking sector has made a very good start to the earnings season, and more industries and companies will announce results in the next few weeks. On July 17, Eastern Time, the New York Times reported that Meta was in preliminary negotiations with Anthropic PBC and planned to lease the computing power of its data center to the latter. If this cooperation is reached, the two-year contract value may reach up to 10 billion US dollars. Boosted by this news, Meta's share price quickly narrowed its intraday decline, eventually closing down 2.8%. It once plunged nearly 6% during the intraday session. According to reports, the design of the terms of this potential deal shows that both parties are leaving a way out for uncertainty. The flexible arrangement of monthly payment and early exit means that US$10 billion is only a theoretical maximum - the actual constraints are far lower than the superficial figure. The deal structure is similar to the agreement Anthropic signed with SpaceX in May this year - the latter was a three-year deal with a total amount of $45 billion. The intensive emergence of two very large lease negotiations in the short term means that the computing power procurement needs of leading AI model companies are still in the explosive expansion stage.
For Meta, low-barrier clauses reduce the negotiation resistance for the first major customer transaction. For Anthropic, this is equivalent to locking an option in the computing power market - retaining the right to use resources, but being able to adjust the position at any time. According to previous reports, Meta is discussing building a new cloud computing business and may sell its excess AI computing power directly to external customers. Meta CEO Zuckerberg previously stated that the company currently uses almost all of its computing power for internal projects, but the high market demand for computing power is prompting him to reconsider leasing some of its computing power to external parties. At the same time, Anthropic has been actively purchasing external computing resources. In May this year, the company announced an agreement with SpaceX to pay nearly $45 billion over the next three years to obtain computing power in its multiple data centers. In the same month, Anthropic also signed a computing power cooperation agreement worth US$1.8 billion with cloud service provider Akamai.
For Meta, low-barrier clauses reduce the negotiation resistance for the first major customer transaction. For Anthropic, this is equivalent to locking an option in the computing power market - retaining the right to use resources, but being able to adjust the position at any time. According to previous reports, Meta is discussing building a new cloud computing business and may sell its excess AI computing power directly to external customers. Meta CEO Zuckerberg previously stated that the company currently uses almost all of its computing power for internal projects, but the high market demand for computing power is prompting him to reconsider leasing some of its computing power to external parties. At the same time, Anthropic has been actively purchasing external computing resources. In May this year, the company announced an agreement with SpaceX to pay nearly $45 billion over the next three years to obtain computing power in its multiple data centers. In the same month, Anthropic also signed a computing power cooperation agreement worth US$1.8 billion with cloud service provider Akamai.