Nvidia fell as much as 4.4% in the sharpest reports of an AI stock sell-off that rolled through markets during the week of Oct. 21 to 23, while debt-funded cloud provider CoreWeave dropped roughly 10%. The selling spread from Wall Street to Asia and, on Friday, to European markets, arriving alongside a jump in oil that pushed Brent crude toward $113 a barrel. Reports disagree on which day was the worst, but they describe the drop as among the harshest for AI stocks since the technology went mainstream.
What the AI stock sell-off did from Oct. 21 to 23
Chip and AI stocks fell sharply on Wall Street and in Asia, and the drop spread to Europe on Friday, according to Euronews. CNBC led its live coverage with the Nasdaq's drop of more than 2%, its worst day since June, which it tied to AI spending worries and oil prices.
The S&P 500 closed at 7,663.43 in one report, down 1.2% and its worst day since July. Another report put the decline at 1.4%, according to Newsy Today. That gap matters, because many index funds track the S&P 500, so readers should check the official closing data before quoting a figure.
Meta also fell after its earnings report, but coverage conflicts on why. One account says the company beat on revenue but missed on profit. Another says it beat on revenue and posted upbeat guidance. Meta shares fell roughly 5.3%, to $660.51, in one report and about 7% in another, per Archyde.
Why Nvidia and CoreWeave led the decline
Nvidia is the largest US company by market value, which made it the biggest single drag on the S&P 500. AP put its early decline at about 1.8%, while the sharpest reports showed declines of 4.3% to 4.4%. The gap depends on the time of day and the outlet, so any single figure should be dated when you cite it.
CoreWeave's slide reflected worries about its debt. The company borrows money to buy Nvidia chips and then rents them out as cloud capacity, which makes its economics sensitive to borrowing costs. USA 24 has already covered the wider neocloud debt question, including SpaceX's $40 billion GPU debt, so this story focuses on the market-wide reaction. Whether CoreWeave's debt concerns turn into a credit event is still unknown.
Nvidia's own leadership complicates the picture. Jensen Huang told CNBC on Thursday that demand for neoclouds is "off the charts," while adding that capital is scarce and holding back growth. Earlier in October, Nvidia's chief executive had warned about the funding AI requires, and investors cited that warning as one trigger, according to Benzinga.
Oil above $110 added a second shock
Oil rose to its highest level since May on renewed fears of a return to war with Iran. AP reported Brent crude at $112.38 a barrel, while Newsy Today put it at $112.80. Both readings were up about 3.3% to 3.7% on the day, after the month began below $97.
Trump blamed Iran for the surge in a Truth Social post, alleging that Tehran was fabricating attacks to push prices higher before the midterms. Treat that as a political claim, not an established finding.
The oil move reaches well beyond gasoline prices. Archynewsy linked oil above $110 to fuel and inflation concerns that feed into Treasury yields and the Federal Reserve's decision this week. Investors moved into US Treasury bonds, and in one report the 10-year yield fell from 5.13% to 4.96%. Euronews also tied that bond buying to growing French debt fears.
Which parts of the AI trade fell hardest
The damage landed hardest on companies that sell the chips, power and buildings, not on the Big Tech firms writing the checks. Yahoo Finance's calculations show the median decline since the S&P 500's Oct. 6 record close:
| Group | Median change since Oct. 6 record close |
|---|---|
| Data-center builders and equipment makers | -23% |
| Power producers | -14% |
| Nvidia | -11% |
| Big Tech spenders | -5% |
That ordering is telling. The companies building data centers have absorbed far more pain than the firms paying for them, which suggests investors are questioning whether the build-out's returns will arrive on schedule.
One report says this week's losses topped $1 trillion in market value. That figure rests on a single source so far, as Minute Mirror reported, so verify it with a market-data provider before repeating it.
What the sell-off means for 401(k)s and index funds
Many US index funds hold large weights in Nvidia, Meta and other AI names, so moves in those stocks flow directly into 401(k)s and index funds. Nvidia's heavy weight in the S&P 500 means its moves count for more than any other single stock. That is why a sell-off concentrated in a few names can reach retirement accounts that hold no individual AI stocks at all.
Practical steps, offered as general information rather than investment advice:
- Check how much of your portfolio sits in a handful of AI-linked companies, including through index funds.
- Read your fund's top holdings to see how heavily it weights Nvidia, Meta and other AI names.
- Avoid rapid trades based on single-day moves, since reports disagree on the size and timing of this drop.
- Talk with a licensed financial adviser if a large share of your savings is tied to one sector.
Businesses that rely on borrowed cloud capacity face a related risk. If neocloud financing tightens, their compute costs could rise.
Whether the AI bubble is popping and what to watch
Analysts have pointed to a broader worry that AI spending has outrun returns and that financing it has become expensive. One week of selling does not settle that argument. It does show how quickly the market can reprice companies that depend on borrowed money and heavy capital spending.
USA 24 has also reported that Anthropic's IPO paperwork is nearing public view and shows a reported $80.7 billion loss. That filing could give investors a closer look at how a leading AI company's economics add up.
The next test arrives this week. Microsoft, Amazon and Alphabet report earnings, and their AI spending plans will show whether the build-out holds up. Watch the spending guidance first, then oil. If Brent stays above $110 while the Fed weighs inflation risk, the pressure on AI shares will be hard to separate from the energy shock.




