Anthropic Just Cut in Line in Front of OpenAI — and DeepSeek Is Laughing at Both of Them

Anthropic filed a confidential S-1 with the SEC on June 2, 2026. Valuation: $965 billion. Revenue run-rate: $47 billion. IPO race position: first.

Let me say that again for the people in the back: Anthropic, the so-called "defensive safety team that couldn't finish," just leapfrogged OpenAI on the way to Wall Street. OpenAI — the franchise that started the whole commercial AI era, that made ChatGPT a household word, that raised $122 billion at an $852 billion valuation in March — has not filed. Anthropic has.

Patrick Corrigan, a law professor at Notre Dame who studies IPOs, told AP News: "I think we were all expecting OpenAI to go first, so it was a little bit surprising." That's the polite version. The accurate version is: Sam Altman's team just got stiff-armed at the IPO finish line by the team everyone counted out five years ago.1

Here's how the AI League IPO bracket looks today:

Anthropic's revenue run-rate went from $9 billion at end-2025 to $47 billion by late May 2026 — a 5x jump in roughly five months. That kind of growth rate doesn't happen by accident. It happens when enterprise customers are burning through API credits at a pace that keeps breaking projections, and when a company launches Claude Opus 4.8 and the developer community responds by moving entire workflows onto it.2

OpenAI's $852 billion valuation still trails Anthropic. Read that slowly. The company that practically created the consumer AI market is now being chased in market value by the lab that spent two years being described as "safety-focused but commercially slow."

This is what winning actually looks like.

While Anthropic and OpenAI are busy filing paperwork and prepping roadshows, there is a Chinese budget underdog quietly eating the enterprise lunch of both of them.

DeepSeek just claimed the number-one spot on Ramp's trending software vendor list for US businesses — ahead of every major AI platform. Ramp tracks actual corporate card spending across clients ranging from one-person shops to Fortune 100 companies.3

The reason is not subtle. Look at actual API pricing per million tokens as of June 2026:4

That's not a small discount. DeepSeek V3.2 input is roughly 21x cheaper than Claude Sonnet 4.6 and about 36x cheaper than Claude Opus 4.6. The circulating "50x cheaper" figure on social media is directionally right, particularly on output tokens at scale. When your annual token bill runs six or seven figures, those ratios aren't pricing differences — they're budget line items.4

And critically: these US companies are not self-hosting DeepSeek's open-source model. They are paying DeepSeek directly, sending business data to DeepSeek's cloud. Ramp noted Anthropic's enterprise penetration among its customers still sits at 34.4%, OpenAI at 32.3%. DeepSeek is nowhere near those numbers yet. But DeepSeek isn't competing for today's enterprise penetration — it's competing for the next round of vendor contracts.3

Anthropic filing first gives them a first-mover advantage in the investor narrative. Corrigan's point about first movers is real: public investors will compare Anthropic and OpenAI roughly simultaneously, and whoever establishes the comparison frame wins the opening round of that conversation.1

There's a counter-argument that Anthropic revealing its financials first gives OpenAI a roadmap to sharpen its pitch. But that argument assumes OpenAI has something sharper to offer right now — and the $965B vs. $852B valuation gap, plus the revenue trajectory, suggests the comparison currently favors Anthropic.

What neither company has resolved is the structural question the IPO moment will force into the open: they are both losing more money than they make. Dan Ives at Wedbush calls the filing an "opening of the floodgates" for the dormant IPO market. But public markets will demand quarterly earnings calls and disclosure of exactly how much runway these franchises burn per year.

This is also Anthropic's moment to finally flip the narrative. For years the story was: Anthropic takes the moral high ground (turns down the DoD weapons contract, refuses to compromise on safety), but pays for it in enterprise deals and public perception. The $47 billion ARR, the Series H at near-trillion valuation, the IPO filing ahead of OpenAI — that sequence says the moral high ground was also the revenue high ground. The enterprise market, it turns out, does price in trust.

Anthropic won the IPO sprint. Full stop. They came into the AI league as the franchise everyone called too conservative, too safety-obsessed, too slow to commercialize — and they just lapped the legacy powerhouse on the way to Wall Street.

But winning a race to file an S-1 is not the same as winning the market. The real threat isn't OpenAI. It's a Hangzhou lab charging $0.14 per million input tokens and climbing to the top of US corporate spending lists. An Anthropic at $965 billion pre-IPO valuation is going to need to justify that number to public investors quarterly. DeepSeek at sub-cent-per-token pricing doesn't have that problem.

My prediction: Anthropic's IPO prices above $1 trillion and trades up 30% on day one. The safety-and-enterprise brand carries real premium with institutional investors who watched the Florida AG sue Sam Altman personally and the DoD drama play out on Anthropic's terms. But within 18 months of IPO, the dominant investor question will not be "Claude vs. GPT" — it will be "what is the pricing floor when open-weight Chinese models get 50x cheaper every product cycle?"

That answer isn't in the S-1 either.

#AILeague