Anthropic Pushes IPO to November as the ‘Sure Thing’ Trade Loses Its Glow
TREE NEWS reports: Anthropic has pushed its planned initial public offering back to November, giving the AI developer time to publish third-quarter financials and make its case to investors after rival OpenAI released its newest model, Astra, in September. Advisors expect the listing to value the company at roughly $2 trillion, with proceeds potentially reaching $100 billion — both figures that would eclipse the record set by SpaceX’s June listing.
The delay reflects a subtler problem than scheduling. What the market once treated as an obvious winning trade is now being scrutinized far more carefully, as competition intensifies, AI safety debates heat up, and the company’s capital spending plans grow more aggressive.
Revenue Is Surging — Profitability Is Another Matter
The growth numbers are genuinely striking. Quarterly revenue climbed from about $4.7 billion in the first quarter to more than $11.5 billion in the second, roughly a 2.5x jump, while compute spending rose about 65% over the same period, from $3.4 billion to $5.6 billion. Annualized revenue is projected to top $100 billion this year, with existing investors expecting more than $110 billion by year-end — up from just $9 billion at the end of 2025.
But the profitability picture is far less clean. Anthropic is leaning heavily on an adjusted operating margin metric that strips out stock-based compensation and equity donation matching costs — exclusions that exceeded $1 billion in the first quarter alone. That measure improved from negative 13% in Q1 to slightly positive in Q2, and the company has told some investors it could stay profitable on that basis in Q3.
Under standard accounting rules, however, the first quarter produced a net loss exceeding $15 billion, including roughly $12 billion in accounting charges tied to changes in the fair value of equity. Explaining that gap clearly to public-market investors will be a central challenge on the roadshow.
Aggressive Capex Clouds the Near-Term Outlook
Anthropic held roughly $120 billion to $130 billion in cash as of early August, but management has told investors it intends to sharply increase spending on compute infrastructure to train next-generation models. A May agreement with SpaceX commits the company to paying $1.25 billion per month for compute over several years — a single contract equal to about two-thirds of its total average monthly compute spend of roughly $1.9 billion in the quarter ending in June.
Executives say per-unit power pricing in future data center contracts will not stay at these levels, but overall capital expenditure will keep expanding. As one investor put it, public-market shareholders want to believe Anthropic can self-fund its capex, with debt as a bridge rather than an endless borrowing cycle — a question facing the entire AI sector.
Enterprise Traction Is Real
The customer story is a genuine strength. Companies spending more than $100,000 annually grew from about 1,500 in the twelve months through the end of 2025 to roughly 6,000 by the end of Q2 — a fourfold increase. More than 1,000 customers now spend over $1 million a year, and over 100 spend more than $10 million. Anthropic has highlighted AIG, Thomson Reuters, and Spotify as flagship accounts and argues penetration within large financial and insurance enterprises remains early.
The company has also sketched a broader deployment vision spanning drug discovery, autonomous cybersecurity, supply chain automation, personalized education, and legal and financial advisory work. At a recent investor forum at its headquarters, executives including Jared Kaplan, Benjamin Mann, and Andrej Karpathy showcased new products, among them a ‘Model Hardware Standard’ allowing AI agents to control physical equipment such as microscopes and robotic arms. Notably, IPO talk was barely mentioned — the focus stayed on the technical frontier.
Safety Debate and Competitive Pressure
The timing change is largely unrelated to the recent public argument over AI safety; the November target was set before a former researcher’s warning ignited the debate. CEO Dario Amodei argued in a blog post that frontier AI tools carry risks too great to continue developing at the current pace, becoming the first leader among the three major AI firms to publicly call for slowing down. Sam Altman and Elon Musk later voiced similar positions, though Amodei clarified that slowing does not mean halting training or technical progress.
Several Silicon Valley venture firms heavily invested in Anthropic believe the timing is reasonable and will not dampen IPO interest, noting that public shareholders will scrutinize safety risk anyway and an early stance offers protection.
Competition remains the wildcard. OpenAI plans to list as early as 2027 but is in preliminary talks on a new round that would value it above $1.2 trillion, after raising more than $120 billion previously. Some Anthropic investors warn that a comparable OpenAI raise before Anthropic’s debut could divert demand. Low-cost open-source models are another likely line of questioning, as Anthropic must convince the market that enterprise customers will keep paying a premium for flagship intelligence.
Next steps: deeper presentations to research analysts and investors, typically weeks before a public filing, followed by three to four weeks before trading begins. Meetings with potential investors are expected shortly, with the financial impact of a slower AI development pace and the target valuation likely at the center of the discussion.
Key Takeaways for Investors
- Valuation risk is now two-sided: A $2 trillion valuation with up to $100 billion in proceeds would be the largest listing on record, but the shift from ‘sure thing’ to ‘no longer certain’ means pricing and demand could be more volatile than early expectations suggested.
- Watch the accounting, not the headline: Adjusted operating margin excludes substantial non-cash costs. GAAP net losses and equity fair-value charges will be the real test of how public investors judge the earnings path.
- Capex intensity is the sector-wide question: Anthropic’s SpaceX compute commitment signals how capital-hungry frontier AI remains. Investors should assess whether revenue growth can outrun infrastructure spending without endless debt issuance.
- Competitive and financing overhang: A large OpenAI raise before Anthropic lists could siphon demand, while open-source alternatives pressure premium pricing assumptions.
- Safety positioning as risk management: A clear public stance on AI safety may function as downside protection once shareholder scrutiny intensifies.
