The tech-heavy Nasdaq dropped 10% this week, its worst performance since the COVID-19 lockdowns of March 2020, as the AI trade unraveled across Wall Street.
The selloff wiped out all of the Nasdaq’s gains for the year in a single session Friday, when the index fell 5.28%. The so-called Magnificent Seven — Nvidia, Microsoft, Google, Apple, Amazon, Meta and Tesla — have shed more than $4 trillion in combined market capitalization over the past six weeks, a decline larger than the annual GDP of France, according to an AFP analysis based on Bloomberg data.
“Markets were euphoric” about AI but “there’s a huge repricing going on right now,” said Steve Sostick, a financial analyst with Interactive Brokers, told AFP. Adam Sarhan, head of the financial platform MarketTerminal.com, said in an interview with AFP.
Investors had hoped the stock market debut of AI giant Anthropic would provide a floor. Instead, the company gained 17% on its first day before slumping more than 8% to close near its IPO price on Friday, NPR reported. The broader market rout dragged the S&P 500 and the Dow Jones Industrial Average into correction territory, each down 10% from recent highs.
The leadership vacuum at the semiconductor bellwether has rippled through the sector. NPR said that Anthropic is a financial supporter of NPR.
Nvidia posted a record quarter on Nov. 17 — $115.8 billion in revenue, more than double a year earlier, with a January forecast of $130 billion, about $5 billion above consensus. Yet the stock fell 14% on Friday to $125.53, roughly 48% below its Oct. 6 record, AInvest reported citing Reuters. The after-hours gain evaporated within days.
Data-center revenue reached $107.6 billion, up 110% year over year. But accounts receivable rose about $18 billion in the quarter to $81 billion, nearly matching the revenue increase, and days sales outstanding climbed to 64 from 60 — the highest since the 2022 chip slump. Inventory rose $8.1 billion to $39.7 billion.
The Fed’s preferred inflation gauge sat at 3.6% in the latest reading, with core inflation at 3.1% — well above the 2% target. The October minutes, released Nov. 18, showed almost all officials expect further increases; a December hike is widely expected. The minutes also drew a direct line between AI and financial stability.
An S&P 500 fund is, in part, a bet on this trade.
Frontier U.S. models cost between $2 and $8 per million input tokens and $10 to $40 per million output tokens, while top Chinese models — including Tencent’s Hunyuan Hy4 lineup — range from $0.07 to $2 per million input tokens and $0.28 to $6 per million output tokens, the Center for Economic and Policy Research noted.
At 50 cents per million output tokens, OpenAI’s GPT6 Luna and Anthropic’s Claude Hailku 5.5 are competitive on price, but the volume required to reach hundreds of billions in revenue at that level is implausible.

“It’s just going to take a little bit of time to heal.” He added that AI could fall “out of favor” through next year. NPR reported that Smith still sees future prospects for AI as “not based on a pie in the sky,” but questions whether investors will see the sizable returns they are hoping for.
Earlier coverage: Investors Pull Money From AI Stocks, Nasdaq Slides 3%