Anthropic confidentially files for IPO

The most surprising line in this story is the order. Anthropic — the smaller, safety-branded lab spun out of OpenAI in 2021 — got to the SEC first. The confidential draft S-1 was submitted June 1, 2026, less than a week after a $65B Series H that valued the company at roughly $965B ^[1]. OpenAI, by contrast, is reportedly still preparing its own prospectus and eyeing a September listing window ^[2]. In a race where being the AI-IPO benchmark matters more than being the biggest balance sheet, Anthropic just bought itself the right to define what investors expect a frontier-model company to look like on paper.

The confidential route is the strategic part. A confidential S-1 lets Anthropic iterate with SEC reviewers without showing its hand publicly, then flip to public marketing only when the comment cycle is clean and market conditions are favorable. Wedbush's Dan Ives described the moment as the opening of the floodgates for a dormant IPO market, with the AI majors now in 'a race to reach public markets over the coming months' ^[3]. The first mover in that race gets to anchor comparables — every later AI listing will be priced against Anthropic's revenue multiple, customer concentration disclosures, and risk-factor language. That is leverage that does not show up in the valuation.

Anthropic's pitch to public investors rests on a single, almost vertical line. Annualized revenue run-rate jumped from around $10B last year to $47B by May 2026 ^[4], and CEO Dario Amodei publicly said Q1 revenue and usage grew 80-fold on an annualized basis, against an internal plan that called for a 10x ^[5]. That kind of curve is what justifies a $965B price tag — a number that surpasses OpenAI's $852B March valuation ^[6]and reframes Anthropic as a hyper-growth enterprise software story rather than a research lab.

The other side of the ledger is the part the S-1 will have to explain. Roughly $15B of the Series H came from hyperscaler and chip partners — Amazon at $5B, plus Samsung, SK Hynix and Micron ^[7]— capital that maps directly to compute and memory commitments rather than working capital. Community chatter on X has fixated on reported compute outlays running through 2029, and TradingKey's analysis warns all three AI majors still lose more than they make, meaning IPO disclosure will force Anthropic to defend margins, training costs and customer concentration that were opaque in private rounds ^[8]. The bullish read is enterprise demand is real and accelerating. The bearish read is that growth has to keep outrunning a fixed, multi-billion-dollar compute floor — and any quarter where it doesn't will be priced ruthlessly by public markets.

Anthropic was founded in 2021 by Dario and Daniela Amodei and several colleagues who left OpenAI specifically because they wanted a safety-first lab ^[9]. Going public does not delete that mission, but it does change who Anthropic answers to. A public board has fiduciary duties to shareholders; a quarterly cadence rewards growth, margin expansion and unlock cycles; and the long-arc bets that 'safety-first' has historically meant — refusing certain contracts, slowing rollouts, eating margin for evals — become line items that activist investors and sell-side analysts will second-guess every 90 days.

The S-1 will surface this tension whether Anthropic wants it to or not. According to public reporting on the company's posture, in February 2026 Amodei refused DoD requests to drop Claude's ban on mass surveillance and autonomous weapons, which led the DoD to label Anthropic a 'supply-chain risk' — a national-security overhang that will likely show up in the risk factors section ^[9]. At the same time, recent hiring patterns reportedly show enterprise-team hires now outpacing research hires ^[10], hinting that the commercial pivot the IPO will formalize is already underway internally. Notre Dame's Patrick Corrigan put the central question bluntly: whether the prices investors pay 'match up to the substance and fundamentals of what AI is really going to do in the real economy and as a business' ^[3]. For a company whose original brand was caution, the answer will now be litigated by the market.

Anthropic isn't filing alone. SpaceX filed financial information for its own IPO in late May, OpenAI is widely reported to be preparing a confidential prospectus of its own, and the trio together represent something close to $3 trillion in combined market cap and more than $200B in potential raise capacity ^[8]. That is not a normal IPO season — that is a market-structure event. Concentrated listings of this magnitude pull index inclusion, passive flows and active capital all at once, and reshape what 'tech weight' inside the S&P 500 means before any of these companies has reported as a public issuer.

Michael Burry, of Big Short fame, has been the loudest voice drawing the 1999 parallel: adjusted for inflation, he argues, potential IPOs from SpaceX, Anthropic and OpenAI could raise as much capital as roughly 300 internet and TMT IPOs did in 2000 ^[11]. The counter-frame, articulated by Wedbush's Ives, is that the AI rally looks more like 1997 than 1999 — early, fundamental, and not yet euphoric ^[3]. The community read on Reddit reflects the gap: enthusiast subs frame the listing as a quality milestone while broader investing forums describe it as the most irrational market they've seen since the dot-com era, with particular anxiety about index funds being structurally forced to buy unprofitable AI mega-caps. That tension — fundamentals story vs. passive-flow concentration risk — is the prism through which the S-1 will be read.