Anthropic's $2trn Number Is Investor Arithmetic on a May Revenue Figure

Six of Anthropic's backers told the Financial Times that rising revenue would let the five-year-old lab more than double its valuation in an autumn listing at $2 trillion ^[1], a figure the company has not endorsed and which several investors said executives had not settled on even privately ^[3]. For anyone buying Claude capacity, the number that matters is not the valuation but the revenue underneath it: the last figure Anthropic disclosed itself was annualised revenue past $47bn, in May ^[5], while investors now expect $100bn to $120bn by the end of 2026 ^[6].

That gap is between $53bn and $73bn of new revenue ^[1], and it comes from somewhere. It comes from enterprise invoices.

The listing is not speculative machinery. Morgan Stanley, Goldman Sachs and JPMorgan are leading the offering, Quartz reported ^[4], and Anthropic filed confidentially with the SEC in June, which puts it in a quiet period and limits what it can say about its finances ^[17]. So the numbers in circulation come from people with a position in the outcome.

On the disclosed figure, $2 trillion against $47bn is roughly 43 times revenue ^[8]. The FT notes that companies treated as AI beneficiaries, naming Palantir and the cloud group Nebius, have traded this year at about 55 times ^[9], which puts the ask about 22 percent below the comparison set ^[3]. One investor went further, telling the FT that at 800 percent growth "at the incredibly low end they would trade at 30 times," making Anthropic a $3 trillion company ^[10]. If revenue reaches the expected range, the multiple at $2 trillion falls to between 20 and 17 times ^[2]. The valuation is underwritten by growth that has not happened yet, and $2 trillion already sits about 67 percent above the $1.2 trillion secondary marks ^[4] set after Anthropic's May round at $965bn post-money ^[19].

Read the measure before the number. Annualised revenue infers a full year from recent performance rather than counting twelve months of receipts, which means a strong month flatters it, and it is not what a public company reports ^[7].

The buy-side signal is already visible. Anthropic's market-leading model costs more than two and a half times as much to use as OpenAI's flagship, according to Artificial Analysis, while Chinese open-weight models cost a fraction of either and improved sharply this year ^[14]. Ramp payments data shows Anthropic gaining share among US businesses last month, with the same analysts finding those businesses were "hitting their limit on AI spend" and many moving to cheaper alternatives ^[15]. Some customers have reversed internal instructions telling staff to maximise AI use ^[16]. A company that needs to double or triple revenue in a year to justify its multiple is not the counterparty that discounts into a budget freeze.

The other exposure is regulatory. Anthropic is in active litigation against the Department of Defense, which labelled it a supply-chain risk ^[11]; the Commerce Department imposed export controls in June and the company briefly pulled its two leading models ^[12]. Two investors told the FT the ban slowed revenue growth that month before a rebound, and it spooked customers who had built on those models ^[13].

Watch for the public filing, which will show audited revenue against the annualised framing ^[7]. Watch renewal terms offered between now and the listing. And watch whether Anthropic holds list price while cheaper models close the capability gap ^[14].