Anthropic's confidential draft S-1, submitted to the SEC on June 1, sits alongside a $965 billion private valuation from its $65 billion Series H and a disclosed revenue run rate reportedly reaching $47 billion -- up from roughly $10 billion a year earlier. That's about as strong a growth story as any company preparing to go public this year can present. But Anthropic's eventual listing will land in a public-market environment that's shown a clear pattern in 2026: the biggest, best-received IPOs have gone to physical AI infrastructure -- SK Hynix's record Nasdaq debut, Csquare's data-center listing, Standard Nuclear's SMR-focused raise -- not to software-first AI labs, because none of the pure frontier-model companies have actually listed yet.
That makes Anthropic's IPO, whenever it lands, a genuinely novel test case rather than a continuation of an established pattern. Nothing this year has told investors how the public market will price a company whose core asset is a set of frontier AI models and the revenue relationships built on top of them, as opposed to a company whose core asset is a fab, a data center or a defense-AI platform with government contracts. Anthropic's $47 billion revenue run rate is a real, demonstrated commercial base -- distinct from earlier-stage story companies -- but it's also concentrated in a category, foundation models, that faces the fastest commoditization pressure of anything in the current AI stack, given the model-release pileup already reshaping buyer behavior.
“That makes Anthropic's IPO, whenever it lands, a genuinely novel test case rather than a continuation of an established pattern.”
The backdrop adds real risk to the comparison: Nvidia's roughly $1 trillion market-cap decline since May is a live reminder that public markets aren't extending unconditional enthusiasm to everything AI-labeled -- they're pricing specific structural risks, in Nvidia's case a hyperscaler ASIC shift threatening its position in the value chain. Anthropic will face its own version of that scrutiny: whether its revenue growth and enterprise penetration (8 of the Fortune 10, roughly 70% of the Fortune 100 as customers) is durable enough to justify a valuation multiple that infrastructure companies, with their years-long replication moats, don't have to defend in the same way.
For LPs and pre-IPO investors holding Anthropic exposure, the company's actual public listing -- whenever it happens -- will be the real price-discovery event that either validates or challenges its private and secondary-market marks, in a public environment that has so far rewarded infrastructure durability over software-layer growth rates. For infrastructure-focused investors, watching how Anthropic prices relative to this year's infrastructure winners will be a clean read on whether the public market's current preference is AI-infrastructure-specific or simply reflects which companies happened to list first.
The bear case: Anthropic's revenue growth from $10 billion to $47 billion in a year is genuinely exceptional by any public-company standard, and a strong IPO could just as easily reset the market's framework toward rewarding software-layer growth again, rather than confirming the infrastructure-favoring pattern seen so far. What to watch next: whether Anthropic's IPO timeline holds, and if OpenAI's own competing listing process lands first or second relative to Anthropic's, since whichever lab lists first will set the pricing benchmark the other has to clear.