Something happened on June 1 that did not get nearly enough attention.

Anthropic, a company that did not exist before 2021, quietly submitted a confidential S-1 to the SEC. That filing came less than two weeks after the company closed a $65 billion Series H at a $965 billion post-money valuation. No share price set. No listing date confirmed. No public share count. Just a company worth nearly a trillion dollars walking up to the window and saying: we are ready when you are.

Not billion. Trillion.

The revenue numbers behind that valuation are the part worth slowing down on. Anthropic exited 2025 at roughly $9 billion in annualized revenue. By May 2026, that run-rate had crossed $47 billion. Q1 2026 came in at $4.8 billion. Q2 2026 is projected at $10.9 billion, which would be a 130% sequential increase and would, in one quarter, exceed everything Anthropic earned across all of 2025. Dario Amodei said publicly that the pace outstripped internal forecasts by a factor of eight. That is not a polished investor day quote. That is a CEO visibly surprised by his own company’s numbers.

For context, OpenAI posted roughly $5.7 billion in Q1 2026 revenue. OpenAI, which has a four-year head start and the most recognized consumer AI brand on the planet. Anthropic was faster last quarter. That gap is not something analysts were modeling twelve months ago.

Slight tangent, but it matters. SpaceX went public on June 12. Raised $75 billion at $135 per share on the Nasdaq under SPCX. Largest IPO in stock market history. Closed its first day at roughly $161, up about 19%. OpenAI is targeting a September 2026 debut at a $1 trillion-plus valuation. Goldman projects 2026 total U.S. IPO proceeds could reach $160 billion. Some analysts are now putting the full-year number closer to $225 billion when you factor in lockup expirations and secondary volume.

Three companies. North of $3.7 trillion in combined implied valuation. All going public within roughly six months of each other. That capital has to come from somewhere. It almost certainly comes from existing large-cap tech positions. Even investors who never touch a single IPO share could feel it as a headwind in things they already own. Hyperscalers. Semis. The names that have led portfolios for three years. That second-order effect is not getting the attention it deserves right now.

Back to Anthropic specifically.

The Q2 2026 projection includes something that would have seemed far-fetched even six months ago: Anthropic’s first-ever quarterly operating profit. Roughly $559 million on $10.9 billion in revenue. The company spent 71 cents on compute for every dollar of revenue in Q1. That ratio is projected to drop to 56 cents in Q2. Whether that holds into the second half is a real question. Data center spending is expected to ramp sharply in H2 2026, and the company has committed to approximately $80 billion in cloud infrastructure spend through 2029. Longer-term projections shared with investors point to $70 billion in revenue and $17 billion in cash flow by 2028. None of those figures are audited. None are public. They are what Anthropic showed its Series H investors to justify the valuation, and they should be read accordingly.

At $965 billion against a $47 billion revenue run-rate, Anthropic is priced at roughly 20x forward revenue on private market terms. OpenAI’s implied multiple on the same math sits closer to 34x. That discount is either a buying opportunity or a signal that the market knows something about which model wins the enterprise. Probably both, depending on who you ask.

There are two risks embedded in the filing that are not getting enough airtime.

First: the gross-to-net revenue question. Anthropic reports revenue from cloud resellers like AWS and Google on a gross basis, counting full end-customer spend as top-line and booking partner payouts as cost. If the SEC pushes for net revenue accounting during review, headline annualized revenue could drop 20% to 40% in a single disclosure. The multiple math that institutions are carrying right now does not survive that scenario intact. It is a known risk. It is not being priced like one.

Second: the Department of Defense situation. In February 2026, the DOD designated Anthropic a supply chain risk and ordered federal contractors off Claude, after Anthropic refused to allow its models to be used for autonomous weaponry and mass domestic surveillance. Two simultaneous lawsuits followed. A California federal judge granted Anthropic a preliminary injunction in March, temporarily blocking enforcement. The D.C. Circuit denied a parallel motion. As of now, Anthropic remains locked out of DOD contracts while litigation runs. President Trump told CNBC in April that a deal is possible. That situation is live, not resolved. For commercial enterprise clients with ethical procurement requirements, the refusal actually plays well. For government contract revenue, it is a ceiling. Investors going into the IPO need to decide which version of that story they are buying.

Worth noting: Anthropic is a Public Benefit Corporation. Its corporate charter legally incorporates obligations beyond shareholder returns. That structure will almost certainly be the most uncomfortable question on the roadshow. What does fiduciary duty look like when the charter says something more complicated than maximize returns?

For most retail investors, none of this is actionable before the listing anyway. Pre-IPO access is limited to accredited investors through secondary markets, tender offers, or late-stage venture funds with high minimums and thin liquidity. The practical move right now is to track the beneficiaries. Amazon and Alphabet are both strategic investors and primary compute providers. Any revenue growth at Anthropic flows directly through AWS and Google Cloud. That is a more liquid way to express a view on what Anthropic is building, without waiting for a prospectus.

The public S-1, when it eventually drops, will be the most consequential financial disclosure document in the AI industry’s history. Audited revenue. Real gross margins. Actual compute cost structures. Retention data nobody has seen. When that lands, every AI valuation conversation currently happening across the sector gets reset. The direction is genuinely unclear from here.

That part, at least, is worth paying attention to.

For informational purposes only.