#美股AI概念股普涨
On July 9 local time, U.S. AI-related stocks surged across the board. The Nasdaq Composite jumped 1.3% to 26,206.89; the S&P 500 rose 0.81% to 7,543.64; the Dow Jones Industrial Average gained 0.27% to 52,487.41. The Philadelphia Semiconductor Index surged 3.06% in a single day, becoming the strongest main theme of the day.
At the individual stock level, storage chips and optical communications led the market: SanDisk jumped 7.59%, Micron Technology rose 4.52%, Western Digital climbed 5.04%, and Seagate Technology gained 3.50%. In optical communications, Lumentum surged 11.13%, Applied Optoelectronics rose 6.79%, and Corning increased 4.54%. Among semiconductors, ARM rose 9.20%, AMD climbed 5.67%, and both Marvell Technology and ON Semiconductor rose by more than 4%.
Most large-cap tech stocks moved higher: Meta closed up 4.70%, Tesla rose 3.17%, Amazon gained 1.40%, Apple increased 0.90%, and Microsoft rose 0.27%. Nvidia and Google, however, each ended slightly lower, down 0.66% and 0.84%, respectively.
Three driving forces: why did it break out now?
First driving force: Micron’s super investment plan reignites the storage market. Micron Technology officially announced that it will raise its U.S. investment plan by 15k USD, from $200 billion to $250 billion by 2035, to address the surge in demand for memory chips driven by the expansion of AI infrastructure. The company’s goal is to bring 40% of DRAM capacity to be implemented in the United States. BofA believes that by 2027, global cloud and AI infrastructure spending could reach about $1.5 trillion, with 35% to 40% flowing into memory-related segments.
Second driving force: SK hynix’s massive IPO creates a siphon effect. The dust has settled on SK hynix’s U.S. stock ADR issuance. The guidance price was set at $149 per share, about a 3.10% premium to Korean shares, raising roughly $26.5 billion. It is expected to become the largest-scale IPO by a foreign company in U.S. stock market history. The offering received more than seven times the oversubscription by institutional investors, highlighting global capital’s fierce pursuit of AI storage core assets.
Third driving force: Meta’s narrative reverses, and Zuckerberg publicly denies “compute oversupply.” Earlier, the market had interpreted a potential move by Meta to rent out compute externally as a signal of “compute oversupply.” In interviews, Zuckerberg flatly denied it, emphasizing that no company in the industry feels it has excess computing resources, and that Meta has already put all available computing power to use. Meanwhile, Meta announced it will build its first Canadian AI data center in Alberta, Canada, investing more than 13 billion CAD (about $9.1 billion), with a scale of about 1GW. It also launched Muse Spark 1.1 and the Meta Model API, for the first time bringing its own large-scale model to a paid API. The market is starting to interpret Meta’s AI spending as a move toward monetization—not just burning money.
In addition, international oil prices fell significantly—WTI crude dropped about 2% to around $72, while Brent crude slipped to around $76—easing inflation worries and loosening risk appetite.
Under the surface of broad gains, internal market differentiation cannot be ignored. Nvidia and Google fell against the trend, suggesting that capital is not indiscriminately buying all AI assets; instead, it is making refined selections. Hardware areas that directly benefit from the expansion of compute infrastructure—such as storage and optical communications—are the first choice.
More importantly, in the S&P 500 Information Technology sector, more than 60% of stocks are down over 20% from their 52-week highs, which by common definition places them in “bear market” territory. This data reveals a key reality: the prosperity of the AI trade is highly concentrated in a small number of top names, while many tech stocks have already fallen behind.
At the same time, Morgan Stanley’s chief strategist Michael Wilson pointed out that the growth momentum in semiconductor stocks is fading, and capital is rotating toward “hyperscale AI cloud service providers” that have lagged relatively in performance. This means capital rotation in the AI space has already begun.
The AI broad-based rally on July 9 is, in essence, a second confirmation of the AI investment thesis—Micron’s $250 billion investment, SK hynix’s oversubscription, and Meta’s commercialization pivot. The three signals together point to one conclusion: capital expenditures on the AI hardware side have not yet peaked, and the memory shortfall is still real.
But even amid the feast, there are concerns: after SK hynix officially listed on Nasdaq on Friday, has near-term sentiment already been fully priced in? Does Nvidia and Google’s down move against the trend imply that the compute leaders are losing pricing power? The AI trade is moving from a phase of “broad-based gains” into a new stage of “refined pricing”—what to buy “matters more than” whether to buy.
On July 9 local time, U.S. AI-related stocks surged across the board. The Nasdaq Composite jumped 1.3% to 26,206.89; the S&P 500 rose 0.81% to 7,543.64; the Dow Jones Industrial Average gained 0.27% to 52,487.41. The Philadelphia Semiconductor Index surged 3.06% in a single day, becoming the strongest main theme of the day.
At the individual stock level, storage chips and optical communications led the market: SanDisk jumped 7.59%, Micron Technology rose 4.52%, Western Digital climbed 5.04%, and Seagate Technology gained 3.50%. In optical communications, Lumentum surged 11.13%, Applied Optoelectronics rose 6.79%, and Corning increased 4.54%. Among semiconductors, ARM rose 9.20%, AMD climbed 5.67%, and both Marvell Technology and ON Semiconductor rose by more than 4%.
Most large-cap tech stocks moved higher: Meta closed up 4.70%, Tesla rose 3.17%, Amazon gained 1.40%, Apple increased 0.90%, and Microsoft rose 0.27%. Nvidia and Google, however, each ended slightly lower, down 0.66% and 0.84%, respectively.
Three driving forces: why did it break out now?
First driving force: Micron’s super investment plan reignites the storage market. Micron Technology officially announced that it will raise its U.S. investment plan by 15k USD, from $200 billion to $250 billion by 2035, to address the surge in demand for memory chips driven by the expansion of AI infrastructure. The company’s goal is to bring 40% of DRAM capacity to be implemented in the United States. BofA believes that by 2027, global cloud and AI infrastructure spending could reach about $1.5 trillion, with 35% to 40% flowing into memory-related segments.
Second driving force: SK hynix’s massive IPO creates a siphon effect. The dust has settled on SK hynix’s U.S. stock ADR issuance. The guidance price was set at $149 per share, about a 3.10% premium to Korean shares, raising roughly $26.5 billion. It is expected to become the largest-scale IPO by a foreign company in U.S. stock market history. The offering received more than seven times the oversubscription by institutional investors, highlighting global capital’s fierce pursuit of AI storage core assets.
Third driving force: Meta’s narrative reverses, and Zuckerberg publicly denies “compute oversupply.” Earlier, the market had interpreted a potential move by Meta to rent out compute externally as a signal of “compute oversupply.” In interviews, Zuckerberg flatly denied it, emphasizing that no company in the industry feels it has excess computing resources, and that Meta has already put all available computing power to use. Meanwhile, Meta announced it will build its first Canadian AI data center in Alberta, Canada, investing more than 13 billion CAD (about $9.1 billion), with a scale of about 1GW. It also launched Muse Spark 1.1 and the Meta Model API, for the first time bringing its own large-scale model to a paid API. The market is starting to interpret Meta’s AI spending as a move toward monetization—not just burning money.
In addition, international oil prices fell significantly—WTI crude dropped about 2% to around $72, while Brent crude slipped to around $76—easing inflation worries and loosening risk appetite.
Under the surface of broad gains, internal market differentiation cannot be ignored. Nvidia and Google fell against the trend, suggesting that capital is not indiscriminately buying all AI assets; instead, it is making refined selections. Hardware areas that directly benefit from the expansion of compute infrastructure—such as storage and optical communications—are the first choice.
More importantly, in the S&P 500 Information Technology sector, more than 60% of stocks are down over 20% from their 52-week highs, which by common definition places them in “bear market” territory. This data reveals a key reality: the prosperity of the AI trade is highly concentrated in a small number of top names, while many tech stocks have already fallen behind.
At the same time, Morgan Stanley’s chief strategist Michael Wilson pointed out that the growth momentum in semiconductor stocks is fading, and capital is rotating toward “hyperscale AI cloud service providers” that have lagged relatively in performance. This means capital rotation in the AI space has already begun.
The AI broad-based rally on July 9 is, in essence, a second confirmation of the AI investment thesis—Micron’s $250 billion investment, SK hynix’s oversubscription, and Meta’s commercialization pivot. The three signals together point to one conclusion: capital expenditures on the AI hardware side have not yet peaked, and the memory shortfall is still real.
But even amid the feast, there are concerns: after SK hynix officially listed on Nasdaq on Friday, has near-term sentiment already been fully priced in? Does Nvidia and Google’s down move against the trend imply that the compute leaders are losing pricing power? The AI trade is moving from a phase of “broad-based gains” into a new stage of “refined pricing”—what to buy “matters more than” whether to buy.
