Separating AI hype from reality.

Anthropic IPO Prices at High End of Range Amid AI Selloff, $900 Billion Compute Bill, and Founder's Own Warnings

November 18, 2026 · Anthropic · Funding · IPO

Anthropic is set to price its initial public offering at the high end of the $70–78 per share range Wednesday after the offering drew six times more demand than shares available, according to a Bloomberg report cited by Yahoo Finance [1]. The company plans to begin trading Thursday on the Nasdaq under the ticker ANT, valuing the five-year-old AI lab at roughly $1.4 trillion on a fully diluted basis—well below the $2 trillion target discussed earlier this year [2][3].

Alphabet and Nvidia have each committed to buy $10 billion of stock at the IPO price [1]. Separately, the company plans to raise $26 billion from a strategic tranche that includes AMD, Broadcom, Google, Microsoft, and Nvidia [2]. The base offering of 825 million shares is expected to raise up to $64.3 billion, which would make it the second-largest U.S. IPO on record after SpaceX's $75 billion debut in June [2].

A moment of maximal contradiction

The oversubscription arrives amid a broad AI selloff. The sector has retreated from its July peak, driven by subpar earnings [2], rising financing costs [4], and the abrupt departure of Nvidia CEO Jensen Huang last week [2] amid reports of extramarital affairs, including one with an alleged Chinese spy [4]. Nvidia shares have fallen more than 35% from their October high [2].

Anthropic's own financials reflect both the opportunity and the strain. Revenue reached about $17 billion in the third quarter, up 45% from the second quarter [2], and roughly tenfold year-over-year, with a net revenue retention rate of 900% [1]. The company posted a positive adjusted operating margin in each of the last two quarters [1]. Yet it remains unprofitable under GAAP, carries $6.2 billion in debt it describes as "substantial," and has signed leases and contracts requiring $900 billion in compute and infrastructure spending in the coming years—including $70 billion in the next 12 months [4]. Cash and investments stand at $132 billion [1].

Morningstar, which assigns a "narrow moat" rating, set a fair-value estimate of $92 per share ($1.7 trillion equity value), implying 24% upside to the IPO midpoint [3]. Its bull case ($188) assumes a five-year growth rate above 60% and 12% share of software and IT services spending by 2031; its bear case ($25) models 22% growth and 3% share, factoring in competition from open-weight models [3].

Recursive self-improvement: claim vs. verification

Anthropic's amended prospectus makes recursive self-improvement—AI models improving the development of better AI models—a central part of the investment case. The company states that its Claude model now writes more than 90% of contributions to its codebase and leads around a quarter of model research tasks, up from roughly 1% in March [3]. METR, a safety-focused third-party group that reviewed Anthropic's internal data, estimated this month that Anthropic's pace of AI progress has roughly doubled over the past year, attributing about 80% of that acceleration to AI itself [3]. Outside researchers now see an increasing likelihood that AI could outperform Anthropic's best human researchers sometime in 2027 [3].

The metrics come from Anthropic's own prospectus and METR's review of internal data granted by Anthropic. The 2027 timeline comes from outside researchers cited by Morningstar. Morningstar flags the scenario as widening the range of outcomes in both directions: if real, even its bull case may be conservative; if it triggers regulatory pauses or safety interventions, adoption could slow sharply [3].

Political and regulatory headwinds

Newly elected Democratic governors in Texas and Georgia have vowed to halt new construction [4]. Congressional Democrats are reviving a push for a national moratorium on new data centers [2][4]. The regulatory environment could constrain the very compute capacity Anthropic has contracted for.

CEO Dario Amodei himself called in September for an industry-wide slowdown in AI development after a former Anthropic researcher warned of existential risk [2]. The company is now asking public investors to discount threats its founder has publicly emphasized.

What remains unknown

  • The final IPO price and whether the share count will be increased.
  • How much of the $900 billion in compute commitments is firm versus conditional.
  • Whether the recursive self-improvement metrics generalize beyond coding and narrow research tasks.
  • The impact of a potential data-center moratorium on Anthropic's contracted capacity.

The offering will test whether public markets are willing to fund a company whose growth depends on unprecedented capital intensity, whose founder has warned the technology may require a pause, and whose competitive moat rests in part on claims about AI accelerating its own development that have not been independently verified.

Sources

  1. Anthropic's IPO Is Reportedly 6 Times Oversubscribed and Could Start ... (finance.yahoo.com)
  2. Anthropic's IPO set for blastoff this week in key test for AI boom (cbsnews.com)
  3. Anthropic: We See the Stock as Undervalued in the IPO Pricing Range (morningstar.com)
  4. Anthropic bets investors dumping tech stocks will still buy into its IPO | AP News (apnews.com)