The number moving around Wall Street is $2 trillion, in October, on a listing that would be the largest initial public offering in history. The number is probably worth taking seriously. It is worth being precise about where it came from, because the answer is not the company.
Anthropic’s backers, speaking to the Financial Times, are the ones putting $2 trillion or more on an October debut. Reporting on the same story has been consistent on an awkward detail: senior Anthropic executives had not fixed a valuation target for the IPO, even in private conversations. The company filed a confidential draft registration statement with the Securities and Exchange Commission on June 1. It has not named an exchange, a ticker, a roadshow date, or a price.
So the largest IPO in history currently has a valuation, a month, and a venue attributed to it by the people who stand to sell into it.
The Gap That Has to Be Explained
Anthropic’s last priced round was real and recent. On May 28 the company raised $65 billion in Series H at a $965 billion post-money valuation, led by Altimeter Capital, Dragoneer, Greenoaks and Sequoia Capital. That is the last arm’s-length mark anyone actually paid.
Getting from $965 billion to $2 trillion is a doubling in roughly five months. Some of that is defensible on operating performance, and the performance is genuinely extraordinary. Anthropic told investors its annualized revenue run rate reached $65 billion at the end of July, up from $47 billion in May and around $9 billion at the close of 2025. CNBC’s account of the same investor update has backers expecting the year to finish somewhere between $100 billion and $120 billion.
Growth like that does not need embellishment. Which is what makes the framing choice interesting.
Read the Profitability Line Carefully
Anthropic posted its first profitable quarter in Q2, with revenue of roughly $11.5 billion and positive adjusted operating income. Two qualifiers in that sentence are doing real work.
The first is “adjusted.” The second is more important, and it came from Anthropic itself: the company cautioned investors that planned compute infrastructure spending across late 2026 and 2027 is likely to push operating results back into negative territory. That is not a skeptic’s reading. That is guidance.
A single profitable quarter, on an adjusted basis, that the company expects to reverse, is a milestone worth reporting and a thin foundation for a permanent rerating. The honest version is that Anthropic demonstrated frontier AI can generate operating profit, then told its shareholders it intends to spend past that point on purpose, because compute is how you defend the position. That is a defensible strategy. It is not a profitability story.
What $2 Trillion Actually Prices
At $2 trillion against a $65 billion run rate, buyers are paying roughly 31 times annualized revenue. Against the optimistic $100 billion to $120 billion exit rate for the year, call it 17 to 20 times. Those are software multiples applied at a scale where no software company has ever operated, on a business whose largest cost line is a physical input priced by a supplier that is itself the most valuable company in the world.
Fortune’s analysis put the tension plainly, arguing the underlying business is nowhere near the valuation it wants. Whether that holds depends almost entirely on whether enterprise revenue proves sticky when the current buildout cycle stops subsidizing everyone’s inference costs.
There is also the risk-factor question. CNBC reported last week that Anthropic’s filing is expected to flag AI backlash as a risk factor, which is the sort of disclosure that reads as boilerplate until a jury or a legislature makes it concrete.
The SpaceX Comparison Cuts Both Ways
Every version of this story anchors on SpaceX, and it is the right comparison for one reason and the wrong one for another.
SpaceX priced at $135 and closed its first session at $161, a 19% pop, on a $1.77 trillion valuation on June 12, selling more than 555 million shares for a raise near $75 billion. It worked. It is the benchmark Anthropic’s investors are explicitly trying to clear.
Here is the part that cuts the other way. A 19% first-day pop means the deal was priced 19% below where the market cleared, which is money that went to allocated buyers rather than to the issuer. Anthropic’s existing holders watched that happen. Anchoring $2 trillion in the press months before a roadshow is one of the few tools available to shift where the book opens. This is not a conspiracy; it is standard pre-IPO positioning, and recognizing it is just literacy.
The appetite is not in doubt, incidentally. Unitree opened 629% above its IPO price in Shanghai last week. Retail and institutional demand for anything adjacent to AI and robotics is extraordinarily strong right now, which is precisely the environment in which valuations get set by narrative rather than by cash flow.
What Would Make the Number Real
Three things, none of which are visible yet.
The public S-1, with audited financials rather than investor-deck figures, will show gross margin after compute, customer concentration, and the actual shape of that “adjusted” operating income. Anthropic’s move to build a custom silicon team is the clearest signal of where management thinks the margin problem sits, and progress there matters more to a 2028 valuation than any 2026 run rate. And the competitive picture keeps moving: the pricing squeeze between OpenAI, Anthropic and DeepSeek is the mechanism that would turn 31 times revenue into a mistake.
None of this makes $2 trillion wrong. Anthropic has grown revenue roughly sevenfold in eight months, which almost nothing in corporate history has done at this absolute scale. It does mean the number is a bid from sellers, not a mark from the market, and those are different objects that happen to be denominated in the same currency.
The S-1 will settle it. Until it is public, everything anyone knows about Anthropic’s finances comes from Anthropic, filtered through people who own a piece of it.