1. AI chip stocks rebound after a two-day rout
Shares of AI chip companies recovered some ground on Friday, November 7, after a sharp selloff that began on Wall Street and spread to Asia. Nvidia rose 2.78% to $186.64 and AMD gained 11.04% to $449.04, though both remain well under their October highs and the broader question of a bubble is unresolved.
One session of gains is a thin basis for conclusions. The advance followed steep declines on Thursday, when Nvidia dropped 9.3% and AMD fell 17.3%, so the rebound only returned a portion of what was lost. Investors are now trying to judge whether the slide was an overdue reset in valuations or the first stage of something more damaging.
The distinction matters well beyond semiconductor shareholders. Chipmakers sit at the center of the artificial intelligence build-out, and their share prices have become a shorthand for confidence in the entire spending cycle. When those prices fall quickly, lenders, utilities, construction firms and cloud providers all begin to reassess their own plans.
2. How far did Nvidia and AMD fall, and how much came back?
Both stocks are still deeply below where they traded last month. Nvidia's Friday close of $186.64 followed a 9.3% decline in a single session, and the 2.78% gain recovered only a small slice of that drop. AMD's 11.04% jump looks larger in percentage terms, but it came after a 17.3% fall, which means the stock needed a much bigger rise just to return to its starting point.
That arithmetic is worth keeping in mind whenever a large rebound is reported. A stock that falls 17.3% must rise by roughly 21% to break even, and a stock that falls 9.3% must rise by about 10%. Neither company has come close to that, so Friday's moves are better described as a pause than a reversal.
The US Tech 100 index, tracked here through a cash-market proxy, added 1.85% to reach 27,007.3. It is still down more than 5.5% for the week. The bounce therefore sits on top of a much larger slide from October levels, rather than signaling a return to the earlier trend.
3. Why did the selloff spread to Asia?
The turmoil moved quickly across the Pacific because many of the companies that manufacture and assemble AI hardware are based there. South Korea's KOSPI benchmark fell 6% on Friday, and Taiwan's main index lost more than 8% over the week, its worst weekly performance in months.
TSMC, the Taiwanese contract chipmaker that produces processors for many of the leading AI designers, reached its daily limit-down threshold, the maximum amount a stock is permitted to fall in one session. Taiwanese officials reportedly asked state-owned banks to help support prices, a step that signals how seriously authorities viewed the pressure on their market.
Taiwan and South Korea together account for a large share of the world's advanced chip and memory production, which explains why their benchmarks are so sensitive to sentiment toward American technology stocks. A sharp fall in one place tends to invite selling in the other, particularly when overseas investors hold sizable positions.
The episode shows how tightly the AI supply chain links markets that otherwise have little in common. A change in expectations for American data-center spending travels directly into order books for fabrication plants, memory suppliers and equipment makers in East Asia, and share prices there react almost immediately.
4. What is driving the AI trade to unravel?
Two forces are in play: a political shock and a financial one. On the political side, the U.S. midterm elections raised concern that Democratic gains would bring tougher scrutiny of data centers. Ballotpedia found that on November 3, voters favored limiting development in 40 of 41 local ballot contests, including some that rejected measures supporting construction. Some of those votes were advisory rather than binding.
That matters because data centers are the largest source of growth for chipmakers. Each facility requires heavy spending on processors, servers, electrical power and cooling equipment. If local rules restrict new projects, orders could be delayed, or construction could move to other locations, which would push revenue further into the future.
The financial concern predates the election. AI companies and their customers have been spending heavily on capacity, frequently before the returns on that investment are clear, and higher borrowing costs make such spending harder to sustain. Market strategist Steve Sosnick warned that heavily owned stocks were "priced to perfection," meaning the prices left little room for disappointment.
Data centers also depend on local approval for land, grid connections and water use, so even non-binding votes can influence how developers and their financiers judge the risk of a project. A signal of public resistance does not stop construction by itself, but it raises the odds of delays, and delays are costly when equipment and capital are already committed.
Taken together, the two pressures reinforce each other. A political obstacle to building raises the risk that spending will not generate returns on schedule, and a higher cost of capital makes that delay more expensive. Investors who had been willing to look past both are now reconsidering.
5. Is this a correction or a bubble bursting?
A correction is conventionally defined as a decline of at least 10% from a recent peak, and it does not necessarily mean a business is getting smaller. A stock trading at 60 times earnings that falls to 40 times earnings loses a third of its value even if profits do not change. In that case, the market is simply paying a lower price for the same earnings.
A bubble bursting is a more serious event. It occurs when expectations prove unrealistic and the money that supported the boom disappears, leaving companies unable to fund their plans or repay what they borrowed.
History offers a reference point in the dot-com era. According to Goldman Sachs, the Nasdaq lost roughly 77% between March 2000 and October 2002. The internet was not a passing fad, but investors had overestimated how quickly companies could turn it into profit.
The telecom industry supplies a closer parallel. Operators laid fiber-optic cable for customers who had not yet arrived, and Global Crossing filed for bankruptcy in 2002. The technology survived and eventually became essential, yet many of the companies that built it, and their shareholders, did not.
6. Where do AI valuations resemble earlier booms?
AI could face a similar gap between its long-term promise and its near-term returns. Signs of enthusiasm outrunning fundamentals are visible in several places: the shortage of memory chips has been treated as a business opportunity, former cryptocurrency miners now rent out computing hardware, and a growing number of cloud and inference startups are competing for the same customers.
A simple hypothetical shows how the risk could travel. A company borrows billions of dollars to construct a data center filled with Nvidia chips, expecting to collect premium rental income. If customers cut their spending, if borrowing costs rise, or if construction is delayed, the debt still has to be repaid. The operator may cancel chip orders, contractors lose work, and utilities lose customers they had planned for. Lenders and bondholders may take losses, and a selloff in equities can then spill into the wider economy.
There is an important difference from 2000, however. Microsoft, Alphabet, Amazon and Meta are profitable, have established customers and generate substantial cash flow. They can absorb setbacks far more easily than smaller AI-cloud providers that depend on repeated borrowing to keep expanding.
The broader lesson is that a technology can transform industries even while many of the companies built around it fail, and investors in the eventual winners can still pay too much. A real long-term opportunity does not guarantee a good entry price.
7. Which warning signs are investors watching?
One indicator of stress appeared in the credit market on Friday. The cost of insuring against a default by CoreWeave, an AI-cloud provider, rose above 1,000 basis points. Such a price signals growing concern about whether the company can meet its repayment obligations, because a higher cost of protection means buyers see a greater chance that something goes wrong.
Broader bond markets, by contrast, were still working in a relatively normal way. That pattern suggests the strain is concentrated in the riskiest corners of AI financing, where companies rely heavily on debt, rather than reflecting a systemic crisis across the financial system.
Credit-default protection works like insurance on a loan. The buyer pays a recurring premium, quoted in basis points, and receives compensation if the borrower fails to pay. A level above 1,000 basis points means the annual premium exceeds 10% of the insured amount, a price typically associated with companies the market regards as financially fragile.
Whether that distinction holds will depend on how lenders treat the next round of refinancing. Borrowers that cannot roll over their debt on acceptable terms would find their expansion plans cut short, while those with strong cash flow would be able to wait out the turbulence.
8. What happens next for AI stocks?
The coming week brings several tests. Earnings from CoreWeave, Applied Materials and Cisco should show whether customers are still ordering computing capacity and the equipment that supports it. Applied Materials is a useful gauge of chip manufacturing demand, and Cisco reflects spending on networking inside data centers.
Inflation data will also matter, because it shapes expectations for borrowing costs. A reading that points to persistently high prices would suggest rates could stay elevated, which would add pressure on companies that are financing their growth with debt.
Stable orders and manageable debt would support the view that the recent decline was a correction. Canceled projects, failed refinancing or shrinking sales would point instead to a crack in the financial foundation of the boom. The question facing the sector is less whether AI has a future than whether every company betting billions of dollars on it can afford to reach that future.