Chip stocks get crushed: Philadelphia Semiconductor Index plunges 12.25% in a single day—the AI bill has come due
On Thursday, November 5, the Philadelphia Semiconductor Index closed at 9,011.14, down 12.25% for the day—its largest single-day drop since 2020. All 30 of its constituent stocks fell.
The loss leaderboard was horrifying: Sandisk -22.26%, Micron Technology -17.46%, Advanced Micro Devices -17.3%, Qualcomm down more than 13%, Intel -10.33%, Nvidia -9.3%, and Broadcom -9.3%. The broader market was dragged down too: the Nasdaq Composite closed at 24,315.00, down 5.52%, and had fallen 11.9% from its October 6 record closing high, officially entering correction territory—a correction is a decline of more than 10% from a record high, which technically confirms a market pullback. The S&P 500 fell 4.36%, the Dow Jones Industrial Average dropped 3.44%, and the entire market lost about $2.3 trillion in market value in a single day. In October alone, the Nasdaq Composite had already shed more than $2.5 trillion.
No company in this wave of selling suffered a blowup, and there was no news of demand collapsing. The sell-off was triggered by three forces converging:
The first was politics. In the November 3 midterm elections, Democrats won both the House and the Senate, prompting investors to worry that the new Congress might tighten oversight of data center construction, energy use, and environmental standards. UBS estimates that about 30% of the data center capacity the U.S. plans to add by 2028 could be delayed or blocked. The newly elected governor of Texas campaigned on a platform that included pausing new data center construction. One key detail: Congress does not take over until January next year, and no laws have been enacted yet. What fell today was expectations, not reality.
The second was returns: the AI bill has come due. Tech giants are borrowing money to build data centers—Meta Platforms issued more than $20 billion in bonds over two days, and Alphabet issued €6 billion. Hundreds of billions of dollars are pouring into chips and data centers, and the money is being spent faster than it is being earned. The market is starting to ask: when will the returns arrive? Even Elon Musk has acknowledged that Grok's service costs are high relative to its revenue. Once the “spend first, earn later” narrative is put under the microscope, valuations become hard to justify.
The third was valuation. Michael Burry, who predicted the housing bubble in 2008, is once again in the spotlight. His criticism focuses on AI companies' depreciation assumptions. Depreciation means spreading the cost of equipment over its useful life. AI chips and servers may lose value much faster than companies assume in their books, which could mean profits are overstated. According to a September report, he has already bought put options on Nvidia, Micron Technology, and Palantir. A put option is a bet on a stock falling—you make money if it drops.
One telling detail: the hardest-hit stocks were in memory and among newer cloud providers. Sandisk, Micron Technology, Advanced Micro Devices, and Nebius all fell by around 17%—not because demand had collapsed, but because these stocks had the most froth and the loftiest expectations after their recent rallies. Nvidia fell 9.3%, which was not the worst on the leaderboard. That suggests the market has not rejected AI itself; it has rejected the narrative of “spend without limit, and returns will come on their own.” The anchor for this correction has shifted from “How much can AI earn?” to “How much will AI cost, and how long will it take to pay back?”
Next, watch two things. Tonight at 20:30, the U.S. October nonfarm payrolls report will be released. The forecast is for 65,000 new jobs; a stronger-than-expected figure could increase bets on rate hikes and put further pressure on tech stocks. On November 17, Nvidia reports earnings—the company's guidance is the next key test.
There are three risks. First, policy has not yet been enacted, so sentiment may overshoot; a drop that goes too far can also be followed by an excessive rebound. Second, if tonight's jobs report is strong and the odds of a December Federal Reserve rate hike rise again, chip stocks could face a second wave of selling. Third, if Nvidia's November 17 earnings guidance falls short of expectations, the entire sector could take another leg down.
The AI story isn't over, but today the market took its first serious look at the cost of telling it.
#美股 #ArtificialIntelligence
On Thursday, November 5, the Philadelphia Semiconductor Index closed at 9,011.14, down 12.25% for the day—its largest single-day drop since 2020. All 30 of its constituent stocks fell.
The loss leaderboard was horrifying: Sandisk -22.26%, Micron Technology -17.46%, Advanced Micro Devices -17.3%, Qualcomm down more than 13%, Intel -10.33%, Nvidia -9.3%, and Broadcom -9.3%. The broader market was dragged down too: the Nasdaq Composite closed at 24,315.00, down 5.52%, and had fallen 11.9% from its October 6 record closing high, officially entering correction territory—a correction is a decline of more than 10% from a record high, which technically confirms a market pullback. The S&P 500 fell 4.36%, the Dow Jones Industrial Average dropped 3.44%, and the entire market lost about $2.3 trillion in market value in a single day. In October alone, the Nasdaq Composite had already shed more than $2.5 trillion.
No company in this wave of selling suffered a blowup, and there was no news of demand collapsing. The sell-off was triggered by three forces converging:
The first was politics. In the November 3 midterm elections, Democrats won both the House and the Senate, prompting investors to worry that the new Congress might tighten oversight of data center construction, energy use, and environmental standards. UBS estimates that about 30% of the data center capacity the U.S. plans to add by 2028 could be delayed or blocked. The newly elected governor of Texas campaigned on a platform that included pausing new data center construction. One key detail: Congress does not take over until January next year, and no laws have been enacted yet. What fell today was expectations, not reality.
The second was returns: the AI bill has come due. Tech giants are borrowing money to build data centers—Meta Platforms issued more than $20 billion in bonds over two days, and Alphabet issued €6 billion. Hundreds of billions of dollars are pouring into chips and data centers, and the money is being spent faster than it is being earned. The market is starting to ask: when will the returns arrive? Even Elon Musk has acknowledged that Grok's service costs are high relative to its revenue. Once the “spend first, earn later” narrative is put under the microscope, valuations become hard to justify.
The third was valuation. Michael Burry, who predicted the housing bubble in 2008, is once again in the spotlight. His criticism focuses on AI companies' depreciation assumptions. Depreciation means spreading the cost of equipment over its useful life. AI chips and servers may lose value much faster than companies assume in their books, which could mean profits are overstated. According to a September report, he has already bought put options on Nvidia, Micron Technology, and Palantir. A put option is a bet on a stock falling—you make money if it drops.
One telling detail: the hardest-hit stocks were in memory and among newer cloud providers. Sandisk, Micron Technology, Advanced Micro Devices, and Nebius all fell by around 17%—not because demand had collapsed, but because these stocks had the most froth and the loftiest expectations after their recent rallies. Nvidia fell 9.3%, which was not the worst on the leaderboard. That suggests the market has not rejected AI itself; it has rejected the narrative of “spend without limit, and returns will come on their own.” The anchor for this correction has shifted from “How much can AI earn?” to “How much will AI cost, and how long will it take to pay back?”
Next, watch two things. Tonight at 20:30, the U.S. October nonfarm payrolls report will be released. The forecast is for 65,000 new jobs; a stronger-than-expected figure could increase bets on rate hikes and put further pressure on tech stocks. On November 17, Nvidia reports earnings—the company's guidance is the next key test.
There are three risks. First, policy has not yet been enacted, so sentiment may overshoot; a drop that goes too far can also be followed by an excessive rebound. Second, if tonight's jobs report is strong and the odds of a December Federal Reserve rate hike rise again, chip stocks could face a second wave of selling. Third, if Nvidia's November 17 earnings guidance falls short of expectations, the entire sector could take another leg down.
The AI story isn't over, but today the market took its first serious look at the cost of telling it.
#美股 #ArtificialIntelligence