Anthropic's IPO Filing Puts a $2 Trillion Question to Wall Street
Anthropic publicly filed for an initial public offering on October 26, setting up what bankers and financial press are calling a shot at the largest public stock offering in history. The company plans to list on Nasdaq under the ticker ANT, with the Wall Street Journal reporting a targeted valuation between $1.5 trillion and $2 trillion and proceeds that could top $100 billion — enough to eclipse SpaceX's $75 billion IPO from this past June. Goldman Sachs and Morgan Stanley are leading the underwriting.
The S-1's growth figures are the headline case for the high end of that range. Revenue hit $16.3 billion in the first half of 2026, up from $1.2 billion in the same period a year earlier — a roughly thirteen-fold jump. The number of customers paying at least $100,000 annually climbed to more than 11,900 from 950, and net revenue retention, a measure of how much existing customers expand their spending, came in near 900% as of September 30. Just as notable: operating cash flow turned positive at $3.9 billion in the first half, versus negative $1.5 billion a year earlier, even as the headline net loss hit $80.7 billion. The gap is almost entirely an accounting artifact — about $72 billion of that loss came from a non-cash charge tied to revaluing convertible notes, leaving an operating loss of roughly $690 million. In other words, the business that actually runs day to day is far closer to break-even than the GAAP bottom line suggests.
A Retail Push Rare for AI Megacaps
Anthropic is reportedly weighing reserving 15% to 20% of IPO shares for retail investors worldwide, with JPMorgan Chase managing the U.S. retail tranche and Citigroup coordinating the international offering, according to Bloomberg reporting cited by multiple outlets. That would be an unusually large retail carve-out for a company of this size, following — and potentially dwarfing in dollar terms — the retail allocation strategy SpaceX used in its own June debut. Giving ordinary investors direct access to the costliest AI IPO on record raises a pointed question: who absorbs the downside if the valuation the banks are floating doesn't hold once shares start trading?
Skeptics Call the Math Aspirational
Not everyone is buying the growth story at face value. PitchBook analyst Harrison Rolfes has pointed out that while Anthropic's S-1 implies an 85% gross margin, reinvestment in next-generation compute leaves roughly 45 cents of each revenue dollar once the costs of staying competitive are factored in — prompting his warning that retail investors tend to 'buy the story, not the model.' The underlying liabilities are substantial: Anthropic's compute spending alone hit $8.9 billion in the first half of 2026, up from $2.3 billion a year earlier, while the company has locked in more than $430 billion in non-cancelable compute contracts and at least $269 billion owed to Amazon, Google and Microsoft through 2036 — obligations that dwarf its current $16.3 billion half-year revenue base, even if they're paid out over many years.
The skepticism predates the filing itself. In an October 8 research note, independent research firm New Constructs called the prospective listing the 'most ridiculous IPO' of the year and argued Anthropic's standalone economics support a valuation closer to $150 billion — a fraction of even the low end of the range now being floated — writing bluntly that 'we don't think Anthropic has a viable business.' Interactive Brokers chief strategist Steve Sosnick was more measured but no less pointed, telling Axios that buying into the offering requires 'a huge amount of, I don't know, faith.'
Why the Debut Matters Beyond Anthropic
The filing also tests whether public markets will extend the same growth multiple private investors already have. Anthropic was valued at $965 billion just five months earlier in its Series H private funding round in May; asking public shareholders to pay up to double that figure means betting that revenue growth holds its current pace well past where most enterprise software companies plateau. The company's governance structure adds another wrinkle for prospective buyers: its seven co-founders will retain 50.1% of voting power through a Class F share structure, with each founder's stake potentially worth $27 billion to $29 billion at a $2 trillion valuation — a concentration of control that limits how much influence public shareholders will actually have over the business they're funding.
A Nasdaq debut targeted for mid-November, if it holds, will function as a reference price for how public markets value frontier AI labs more broadly — a benchmark that OpenAI and other closely held AI companies will eventually be measured against whenever they pursue listings of their own. Given the scale of the compute and cloud commitments embedded in the filing, the real test isn't whether Claude's growth is genuine; the numbers suggest it is. It's whether that growth can keep outrunning the capital intensity required to sustain it, or whether the market decides, as New Constructs and PitchBook have already argued, that it can't.