The key facts

  • Anthropic has scheduled its IPO for November 2026: 825 million shares at $70–78, with the valuation cut from $2 trillion to $1.2–1.3 trillion
  • After the US midterm elections, the 'Magnificent Seven' lost more than $1 trillion in a single trading day
  • Nvidia shares fell nearly 10% amid an investigation into CEO Jensen Huang's ties, after which he resigned

Anthropic's IPO coincides with market turbulence

Anthropic, the developer of the Claude chatbot, plans to hold its IPO in November 2026, despite the market's sharp cooling toward AI stocks. This is reported by Deutsche Welle. On Tuesday, the company announced that it will offer 825 million shares priced between $70 and $78 per share. The initial valuation target exceeded $2 trillion, but Anthropic now expects $1.2–1.3 trillion. OpenAI, the second-largest player, by contrast, has postponed its stock market debut to next year.

What sent the stocks tumbling

The second half of 2026 brought the market a serious shock. After the Democrats gained control of the US Congress in the midterm elections, technology stocks fell sharply: the 'Magnificent Seven', which includes Meta, Apple, Amazon and Tesla, lost more than $1 trillion in market capitalization in a single trading day. Then Nvidia shares plunged nearly 10% over two days after reports that the company's board of directors, together with the FBI, was examining CEO Jensen Huang's ties with a woman whom American intelligence suspects of working for Chinese special services. On Thursday, after reports of a second extramarital affair, Huang resigned, and the shares fell even further.

Doubts about return on spending

Wall Street is growing increasingly wary of the scale of AI spending. According to a Goldman Sachs estimate cited in the publication, hyperscalers need about $300 billion in annual AI revenue to recoup their costs, and the gap is roughly $230 billion. This year, about $800 billion has been invested in capital expenditures, with another $1.2 trillion and $1.4 trillion planned for 2027 and 2028 respectively. Meanwhile, borrowing costs for infrastructure companies like CoreWeave and Oracle have risen sharply: according to CNBC, the cost of insuring Nvidia's debt against default rose 76% in a month. Oracle shares have fallen 65% from their highs. Nevertheless, Oracle, Meta, Microsoft and Alphabet have announced new AI bond issues totaling $67 billion since the end of last month.

Bubble or healthy correction

Analysts' opinions diverge. Keith Richards of Quilter called last week's decline a 'healthy correction' and noted that 'we are still at the very earliest stages of building AI infrastructure'. Economist Gary Smith, in his forthcoming book 'AI Mania', compares the boom to previous bubbles — around railroads, automobiles, television and the internet — and warns of a 'terrible crash'; for comparison, when the dot-com bubble burst in 2000, the Nasdaq index lost almost 80% of its value. Joachim Klement of Panmure Liberum believes the reckoning may come in 2027 or 2028. Dennis Dick, a strategist at Stock Trader Network, commenting on Anthropic's decision to go public in the middle of a sell-off, wondered: 'Is this the best time?' According to him, the company's valuation depends on whether it can fulfill Goldman Sachs' forecast of revenue exceeding $600 billion by 2030 — at the current $52 billion, that looks overvalued. More optimistic is Vivek Seth of Shaires Holdings, who called Anthropic a 'once-in-a-generation company' and noted that the IPO will establish a market valuation as conditions stabilize.