Anthropic IPO Timing Threatened as AI Selloff Hits Public-Market Appetite

Anthropic's planned Nasdaq listing — the first for a frontier AI lab — is running into a wall of seller resistance. Potential investors are pushing for a lower valuation and questioning whether the company can proceed with its roadshow next week for a November debut, four people familiar with the discussions told Reuters [1]. The roadshow was previously expected to start the week of November 9, with pricing the following week [1]. Prediction-market odds on Kalshi for a filing by November 30 fell from 84% to 60% on Thursday [5].

The pressure reflects a broader re-pricing of AI assets. The Nasdaq 100 dropped as much as 6% and the S&P 500 3.8% Thursday as a liquidation in AI and data-center stocks intensified [2][4]. More than 50 S&P 1500 technology stocks fell 10% or more; retail selling was the strongest since the COVID-19 pandemic, according to Vanda Research [1]. Tech has underperformed the S&P 500 by 10.7% over the past month — the most extreme one-month gap since May 2002 [1].

Valuation Expectations Cut

Earlier this year, investors anticipated an October IPO at a valuation as high as $2 trillion [1]. The Wall Street Journal reported Thursday that IPO investors are skeptical Anthropic can achieve $1.5 trillion in a public listing [1]. Institutional investors have reportedly advised a range of $1.5–1.6 trillion [9]. NYU professor Aswath Damodaran, who tracks the company closely, trimmed his estimate to $1.45 trillion from $1.475 trillion [9]. Anthropic's most recent private round in May valued it at $965 billion [12].

The company filed its S-1 on October 26 and amended it November 2, targeting a listing before Thanksgiving and a raise of roughly $100 billion [2][4]. It has not yet set formal price range or share count [5].

Financials Show Growth — and Enormous Commitments

The amended filing shows third-quarter revenue of approximately $16.96 billion, up more than twelvefold year-over-year [2][4]. But net loss widened to $23.29 billion, and the company posted a $567 million operating loss after a $974 million operating profit in Q2 [2][4][9]. Revenue growth is decelerating: 47% quarter-over-quarter versus 144% in Q2 and 132% in Q1 [9].

Most strikingly, Anthropic disclosed approximately $895.4 billion in aggregate purchase commitments and lease agreements, substantially all for compute and infrastructure [2][4][10]. About $447.5 billion is due by end of 2030; nearly $70 billion comes due in the 12 months after September 30 [2][4][10]. The company held $146.6 billion in cash and liquid assets at quarter-end [9][10]. It warned that if third-party financing for its infrastructure partners becomes unavailable or more expensive, contracted capacity could be delayed, repriced, reduced, or not delivered [2][4].

A Circular Feedback Loop

The market dynamic is increasingly circular: falling AI stock prices threaten Anthropic's mega-IPO, while doubts about the IPO raise fresh questions about the financing available for future data-center investment [2][4]. Vital Knowledge Media wrote on X that the "extremely ugly price action" in pick-and-shovel stocks raises "serious doubts about whether Anthropic can raise ~$100B at a $1.5T+ valuation by Thanksgiving" [2][4].

Neocloud operators Nebius, IREN, and CoreWeave have each lost roughly half their market value over the past month [1]. Microsoft tightened oversight of some data-center projects from neocloud providers Nebius, IREN, and Nscale, Bloomberg reported [1]. Credit-default swaps on CoreWeave and SpaceX hit records implying rising default expectations [1].

Technical Progress Accelerating — With Caveats

Amid the financial turbulence, an independent assessment by METR, an AI safety nonprofit, found Anthropic's models are improving nearly twice as fast as a year ago [6][11]. Benchmark scores (measured via Anthropic's AECI index) have grown at an estimated 1.9× the pre-October 2025 rate (95% CI: 1.5–2.8×) since October 2025 — roughly two years of historical progress in one year [11]. METR is confident the acceleration was not driven by increased training compute [6][11].

Interviews and internal surveys are consistent with AI usage driving much of the speedup. Staff self-reported productivity gains averaging roughly 12× for the most capable internal model as of October, though METR cautions Slack polls are weak evidence [7][11]. Researchers told METR they are "now only bottlenecked on ideas & compute, no longer bottlenecked on code" [7]. METR saw no observed bottlenecks that would prevent further acceleration [7][11].

However, Anthropic redacted a majority of the evidence METR found most informative, citing commercial sensitivity and the risk of handing out a roadmap for others to go faster [8][11]. METR notes the public has to rely on its judgment more than ideal [11]. The assessment also found AI is not yet making highly significant, fully autonomous contributions to the most important stages of model training — though researchers described "signs of life" that this could change soon [6][11].

Political Overhang

Tuesday's midterm elections, which gave Democrats control of Congress, have crystallized opposition to data-center development [5]. Meta president Dina Powell McCormick said Wednesday that building data centers will be harder after the results [5]. OpenAI on Thursday outlined policy proposals including ending NDAs with towns and paying full electricity and infrastructure costs for its facilities [5].

What's at Stake

Anthropic's IPO is a stress test for the entire AI financing chain. Its $895 billion in commitments make it one of the largest customers for Broadcom ($161 billion in equipment leases), Google ($121 billion in cloud commitments), SpaceX (up to $120.5 billion), and Amazon ($110 billion) [2][4][9]. A delayed or downsized offering could reinforce concerns about capital availability across the ecosystem [2][4].

For now, the company and its underwriters — Goldman Sachs and Morgan Stanley lead a 22-bank syndicate [10] — face a narrow window between Veterans Day (November 11, when the SEC is closed) and Thanksgiving (November 26). Whether they proceed, reprice, or wait will signal how much evidence public markets now require before funding the next phase of the AI buildout.