What’s Happening With AI Stocks?
Follow the whole story: every piece in this series is on our AI storyline page.
The Nasdaq 100 fell about 1.4% on Monday and closed more than 10% below its Oct 6 peak, the line that marks a correction. It’s the index’s third correction this year, after March and July. By Wednesday’s close it had bounced to about 9% below the peak. Friday, Oct 30, was the worst day: the Nasdaq 100 fell about 3.5% and chip stocks (the iShares Semiconductor ETF) lost about 8% after Amazon’s and Alphabet’s results and a weak forecast from Intel. So the useful question isn’t “is the market down?” It’s “which part of the AI story did investors stop believing?”
| Stock | Past month | Below its 52-week high | Past 12 months |
|---|---|---|---|
| Intel (INTC) | −51% | −59% | +57% |
| CoreWeave (CRWV) | −46% | −65% | −58% |
| Marvell (MRVL) | −41% | −49% | +82% |
| AMD (AMD) | −23% | −25% | +95% |
| Alphabet (GOOGL) | −16% | −28% | +4% |
| Nvidia (NVDA) | −14% | −16% | +1% |
| Oracle (ORCL) | −13% | −51% | −50% |
| Apple (AAPL) | +5% | 0% (at its high) | +29% |
| Microsoft (MSFT) | +7% | 0% (at its high) | +8% |
The selloff is aimed at the AI builders and the suppliers whose sales depend on them, not at big tech as a whole.
Look at the bottom of the table. Apple and Microsoft, two of the three biggest companies in the world, are up this month and closed Wednesday at 52-week highs. The pain is concentrated in the companies whose next few years depend on AI spending continuing at today’s pace.
See more lines: what’s in the table, and a few more names
Returns are to the close on Wed Nov 4, with the past month measured from the Oct 2 close (Trefis market data). “Below its 52-week high” uses the highest close of the past year. Intel’s month includes a one-day fall of about 29% after its forecast, its worst day in more than four decades. Others in the selloff: Dell −34%, Super Micro −32%, Micron −26%, Meta −13%, Palantir −9%, Broadcom −8% over the past month. Speculative AI-adjacent names fell harder: Cipher Mining −59%, Oklo −38%, Applied Digital −35%, IonQ −33%. Amazon rose 6% over the month.
What Actually Changed?
Not demand. After Monday’s close, Palantir reported a quarter with revenue up 96%, and the stock fell about 3.7% over the next two days, to $171.70. Amazon and Alphabet said last week that AI demand is still running ahead of the capacity they can install. What changed is the price investors will pay for that growth, because the bill has gotten very large. JPMorgan puts this year’s AI spending worldwide at about $900 billion, and $5.5 trillion in total through 2030.
That spending is income for Nvidia, AMD and Marvell, and a cost for Alphabet, Meta and Oracle. When the spenders raise their budgets, investors now ask whether the payback is getting further away. When suppliers report, investors ask how long the budgets can keep growing. Either way, the stock falls. Add a 10-year Treasury yield near 4.9%, which makes profits far in the future worth less today, and you get a correction in exactly the stocks priced furthest into the future.
How Did Past Nasdaq Corrections End?
Mostly well, but not always. Since 1985 the Nasdaq Composite has closed 10% or more below its record about 25 times, and 8 of those, roughly one in three, turned into bear markets of 20% or more. After the Nasdaq first closes in a correction, its median return over the next 12 months has been about +25%. The bad cases are the ones people remember:
| Episode | Nasdaq 100, peak to bottom | Back to the old high |
|---|---|---|
| 2000–2002, dot-com bust | about −83% | about 15 years (2015) |
| 2007–2009, financial crisis | about −54% | about 3 years |
| Late 2018, Fed and trade war | about −23% | about 4 months after the bottom |
| Early 2020, Covid | about −28% | under 3 months after the bottom |
| 2022, rate hikes | about −36% | about 2 years from the peak |
| April 2025, tariffs | about −23% | under 3 months after the bottom |
The pattern: drops caused by a shock from outside (Covid, tariffs, the Fed) healed in months. Drops that came after prices ran far ahead of profits, like 2000 and 2022, took years. This one is the second kind of question: it’s about whether AI spending pays. That doesn’t mean it ends like 2000. It means earnings, not headlines, will decide it. And even giants fall in that kind of market: from their 2021 peaks to their 2022 lows, Meta lost about 77% and Alphabet about 44%.
What Should You Watch Next?
Already in: AMD, Tuesday night (Nov 3). It beat, guided the fourth quarter above estimates and said it expects revenue to grow more than 80% in 2027. The stock closed up only 1% on Wednesday. Same day: the midterm elections, which matter less than you’d think.
Jobs report, Fri Nov 6, and inflation, Tue Nov 10. With the 10-year near 4.9%, a hot number could push yields higher and hit the most expensive stocks again.
Nvidia, Tue Nov 17. The biggest single test. Nvidia is the AI trade’s scoreboard: if its customers are still ordering at the same pace, the “is it paying off?” fear eases.
Not broken, but repricing. AI demand is still growing; what broke is investors’ willingness to pay any price for it. History says about two in three Nasdaq corrections end before becoming bear markets. Whether this one does depends on Nvidia’s report on Nov 17, not on the headlines in between.
How Does It End?
Three paths, with our rough odds.
A pullback that holds (about 55%). AMD has shown its orders are still there; if Nvidia does too, the index bottoms short of a 20% fall, as about two in three corrections have.
A 2022-style bear market (about 35%). Yields stay near 5% and one or two big spenders trim their 2027 budgets. The index falls 25% to 35%, and the most-hyped names fall much further. Recovery takes a year or two.
A 2000-style bust (about 10%). Spending is cut sharply, suppliers’ sales drop and the profits that today’s prices assume never arrive. Rare, and it needs both of the other paths’ triggers at once.
The AI Selloff Guide
One question per piece. Start with the one closest to what you own; every piece in the series is on our AI storyline page.
- The $1.5 Million Risk In AI-Heavy Portfolios
- Nvidia Stock To $129?
- Is The Nasdaq Selloff 2000 Or 2018?
How Much Of Your Portfolio Rides On The AI Trade?
Many $1 million-plus portfolios own these companies twice: directly, and again through index funds, where the biggest AI names are a large share. Before the next leg of this selloff, it’s worth knowing your real number. See what the AI selloff means for your portfolio in a free 15-minute call with a Trefis partner firm.
For questions about this article or underlying research, email help@trefis.com.