Anthropic has pushed its initial public offering from October to November, the Wall Street Journal reported Friday, as the company’s annualized revenue run rate crossed $100 billion ahead of what would be one of the largest tech listings on record. The delay, attributed to advisors who say the extra month lets the company lock in stronger financials and a firmer valuation, comes while rival OpenAI tells investors to expect massive negative cash flow for years.
The two developments, arriving the same weekend, frame the central question for AI markets heading into the fall: whether the sector’s biggest private companies go public as profitable hypergrowth stories or as cash-burning infrastructure bets that need continuous refinancing. Anthropic and OpenAI are giving very different answers.
Anthropic’s fast climb to a near-trillion valuation
Anthropic closed a $65 billion Series H in April at a valuation near $965 billion, and its total disclosed funding now exceeds $134 billion. Revenue growth has kept pace with the fundraising. The company’s run rate, annualized from recent quarterly figures, crossed $100 billion this month, driven largely by enterprise demand for its Claude models and a partner network it built out with a $100 million investment in March.
The company has spent aggressively to support that growth. Its 2026 computing commitments total at least $135 billion, including more than $100 billion to Amazon Web Services over a decade, $45 billion to Nscale for capacity in West Virginia, $35 billion to Nvidia-backed Lambda for a Texas data center, $30 billion to Microsoft Azure, and $10 billion to Volta Infra in Norway. Whether that spending leaves room for profits at listing is one of the details investors will parse from the S-1 when it lands.
| Anthropic 2026 compute commitments | Amount |
|---|---|
| Amazon Web Services (10 years) | $100B+ |
| Nscale, West Virginia | $45B |
| Lambda, Nueces County, Texas | $35B |
| Microsoft Azure | $30B |
| Volta Infra, Norway | $10B |
The November window also lets Anthropic see how markets treat any earlier AI or tech listings and gives the underwriting banks another month of order book building. An IPO at or near its private valuation would rank among the largest ever for a software company. It would also convert a cap table built through years of complex, structured rounds into public shares, a process that has tripped up other late-stage companies when investor rights and preferences surface in the prospectus.
Internally, the company has been reorganizing for public life. Its headcount has grown past 5,000, it has hired senior executives from OpenAI, Google, Microsoft and xAI through the year, and it completed six acquisitions, including the legacy biotech operation it later wound down in the case of one merger vehicle. Leadership has framed 2026 as the year it built the commercial platform rather than the year it maximized model releases.
OpenAI’s very different math
OpenAI, by contrast, is telling investors to expect enormous cash burn for the next several years, according to Investor’s Business Daily, which reported the forecast alongside the IPO delay news. The company behind ChatGPT has committed to hundreds of billions of dollars in data center buildouts and has structured much of its compute purchasing through complex deals with chipmakers and cloud partners.
That does not necessarily make OpenAI’s path weaker. Revenue is growing quickly and the company retains the largest consumer distribution in AI. But the accounting reality is starkly different from Anthropic’s. One company is approaching public markets showing a credible line of sight to profitability. The other is asking public investors to fund a decade of negative cash flow on the promise of eventual scale economics.
How public markets price those two stories side by side will set the reference points for the rest of the private AI market, including xAI, Mistral and the wave of infrastructure companies building around them. A strong Anthropic debut makes every downstream valuation easier. A weak one does the opposite.
Regulation and safety pressure continue in the background
The IPO news lands in a noisy policy week. The President called AI safety claims a hoax on Friday and promised an AI Force, while California’s governor ordered kill-switch requirements for frontier systems the day before. Anthropic and Accenture also committed more than $2 billion to embedded third-party model evaluators, and multiple outlets report the major labs are discussing a joint safety standards body, though the details remain unverified.
Security adds its own overhang. A zero-click remote code execution exploit disclosed September 18 hit Claude Code, OpenAI’s Codex, GitHub Copilot and Gemini CLI through a plugin-loading flaw. Anthropic patched Claude Code in version 2.1.179 and OpenAI patched Codex in 0.146.0, while Microsoft had not shipped a Copilot fix at the time of reporting. For companies pitching enterprise and government customers ahead of an IPO, developer tooling security is no longer a footnote.
There is also the question of narrative discipline. A New York Post report this weekend, citing unnamed insiders, claimed labs including OpenAI and Anthropic overstated AI breach incidents to pressure federal regulators for market protection. Whether or not that holds up, it lands weeks before a listing and gives short sellers a ready-made line of attack.
What to watch in November
The S-1 filing will settle the questions the run-rate headline leaves open: actual margins after compute costs, how much of the $135 billion in commitments is committed versus contracted, and the cap table after years of structured funding rounds. Anthropic’s channel push, now spanning more than 100 partners including Accenture, Cognizant and Slalom, will also be scrutinized for how much revenue flows through intermediaries at lower margins.
The model pipeline matters too. Anthropic has pushed Claude into new territory this month, including a wet-lab automation product that runs biology experiments, and it claims Claude now handles 26% of its internal AI research work. Investors will want evidence that enterprise adoption, not just headline model quality, drives the revenue line.
For OpenAI, the next milestone is its own fundraising and eventually a listing of its own, on a timeline that remains unclear. Until then, Anthropic’s November window is the market’s first real chance to price a frontier lab on public terms, and both companies’ fortunes will be read through it.