Sam Altman settled the question Saturday morning. OpenAI will not go public in 2026, Altman confirmed in a Fortune interview, saying “given everything happening with safety, right now would be an ill-advised moment to go public.” CFO Sarah Friar had already told staff the company “will be a public company in 2027.” That clears the field. The only frontier AI listing arriving this year now belongs to Anthropic, which means October’s offering will be the first time public markets price what it actually costs to build and run a world-class AI lab.
The figure investors are working with is extraordinary. Reuters has reported that some investors expect Anthropic to go public in October at a valuation of $2 trillion or more, which would make it the largest initial public offering in history. The company closed a $65 billion Series H at a $965 billion post-money valuation on May 28, 2026, and its IPO process has been described in the press as a confidential SEC filing. Going from $965 billion to $2 trillion in roughly five months is not a gradual re-rating. It is a demand that public investors accept a 2.1x step-up before they have seen a single public filing.
The revenue trajectory at least earns a serious look. Reuters reported that Anthropic’s annualized revenue run rate topped $65 billion by the end of July 2026, up from $47 billion in May and about $9 billion at the end of 2025. Some backers have floated expectations that the annualized figure could reach $100 billion to $120 billion by year-end, but those are investor projections rather than company guidance. That pace is real. The question disciplined investors must ask is whether the price already reflects it, and then some.
At $2 trillion, Anthropic would be valued at roughly 16.7 times a $120 billion year-end run rate, using the high end of what investors are modeling. That multiple only works if margin expansion continues at the same speed as revenue. The Information reported that Anthropic expected about $559 million in operating profit on $10.9 billion in revenue for the June quarter, which implies a margin of roughly 5%. A 5% operating margin at a $2 trillion valuation is not a business that can afford to miss a quarter.
There is genuine quality here. Anthropic has leaned into enterprise and developer demand rather than the mass-market consumer race, and it has kept spending tighter than some peers. Reporting tied to the company’s recent financial trajectory has pointed to compute costs falling from 71 cents per revenue dollar in Q1 2026 to 56 cents per revenue dollar in Q2 2026 as a key driver of improving profitability. If that ratio keeps compressing as scale increases, the economics get genuinely interesting. But the $2 trillion figure comes from investors rather than Anthropic itself; senior executives have not publicly established an IPO valuation target, and outside models are doing much of the work.
The SpaceX precedent is instructive, and not entirely in the bull case’s favor. Reuters reported SpaceX priced its June 2026 IPO at $135 a share and raised $75 billion, then noted that the stock’s post-IPO market value was about $2.1 trillion shortly after the debut and that the shares later fell more than 30% from their early peak. The market’s most recent test of a frontier, pre-profit-scale technology company at a record valuation produced a sharp drawdown within weeks. Anthropic’s S-1 has not yet been made public. Investors stepping in at the open will be pricing a business whose full cost structure, compute contract obligations, and channel economics they have not verified.
This is where Altman’s delay actually helps Anthropic’s October buyers, at least intellectually. With OpenAI off the table for 2026, there is no competing listing to arbitrage against, and no comparable public peer to anchor valuation. The real diligence question is the gap between a $47 billion May run rate and the $100 billion to $120 billion December figure investors are underwriting, roughly a 2x to 2.5x jump in seven months, built on investor projections rather than company guidance.
The mogul’s discipline here is specific: great businesses bought at the wrong price still produce poor long-term outcomes. Anthropic may well be building something enduring. But October’s price will be set by the most enthusiastic buyers in a market with no public peer and incomplete financial disclosure. Waiting for OpenAI’s 2027 listing, when there will be two sets of audited numbers to read side by side, may be the more rational act of patience.
