The Nasdaq fell 5.28% to 22,617.73, down 1,261 points—the worst day in what was itself the worst week since March 2020. All three major U.S. indexes entered technical correction the same day. The Nasdaq erased all of its 2026 year-to-date gains, while the S&P 500 and Dow were both more than 11% below their October record highs. This was a broad repricing of the AI trade, not the result of any single piece of bad news.
Four reasons, ordered by impact
1. Expected good news failed to work, and sentiment turned to disappointment
Nvidia had just reported revenue that doubled, and Anthropic—which listed on Thursday, Nov. 19, in the largest IPO on record—rose 17% on its first day. The market had hoped these catalysts would turn around two weeks of anxiety in the AI sector. Instead, Nvidia fell 14.01% the next day to a 52-week low, while Anthropic traded as low as $83.05 intraday—just five cents above its $83 IPO price—and closed down 8.5%. Good news was exhausted and buyers did not follow through. That hurts sentiment more than any bad news. “Stronger earnings can’t produce a gain” means valuation is already fully reflected in the price.
2. The core doubt shifted from “how many chips were sold” to “can AI capex be recouped”
The Philadelphia Semiconductor Index plunged about 10.7% in a single day, with Broadcom, AMD, and Micron all down more than 12%. What really got hit was the narrative: GPU delivery is only the starting point of capex. Only if data centers come online on schedule and paid utilization meets targets can projects cover power, debt service, and recover equipment investment. Once that chain is questioned, selling spreads from model companies to chips, cloud compute leasing, and data-center power equipment—the entire supply chain comes under pressure together.
3. The Fed turned hawkish, raising the discount hurdle for future profits
Wednesday’s meeting minutes showed almost all officials agreed on the need to raise rates “in the coming months” to bring inflation back to 2%. Governor Waller reiterated this week that he wants a “50 basis point hike in December,” and Cook emphasized on Nov. 20 that “not acting now would require larger moves later.” Rate futures priced a 25 bp December hike at nearly 100%. Rising rates directly compress the premium investors are willing to pay for high-valuation growth stocks.
4. Several overlooked amplifiers
Liquidity: Anthropic’s $100 billion-scale IPO drained substantial capital from the secondary market, while the dollar strengthened and financial conditions tightened. Policy expectations: after Democrats retook control of Congress, the market began pricing stricter AI regulation and slower data-center construction. Mechanical deleveraging: crowded leveraged AI longs cut positions together, and CTA-style strategies automatically sold as prices fell, creating a negative feedback loop. The VIX hit an intraday high of 40.03, closed at 35.89, and rose 26.2% on the day.
My view and the limits of this call
This decline is the AI trade switching from “narrative pricing” to “cash-flow pricing.” That is healthy in nature but violent in pace. The key point is that the decline itself does not prove the bubble has cleared. The Nasdaq has fully erased its year-to-date gains but is still about 18% below its October high, while the Nasdaq-100 is about 20% below its high. That suggests most of the prior gain was a retracement of valuation premium, not a fundamental re-rating.
I do not think the short-term move is over: JPMorgan’s technical strategists say indicators show selling is not finished, and Capital Economics also expects further downside in tech stocks. Next week brings the Fed’s preferred inflation gauge; if it confirms the case for a hike, pressure will increase another round.
The key to whether this call is right lies in two things: if inflation data before the December FOMC meeting falls more than expected, this could be a panic overshoot followed by a quick rebound; if the hike is delivered and any cloud provider cuts its data-center capex guidance, that would be the start of a trend decline, not a correction.
The above is not investment advice. Please fully assess your own risk tolerance, make your own investment decisions, and independently bear investment risks.

