The Anthropic IPO is officially in motion. Anthropic, the company behind Claude, has confidentially filed to go public. The filing landed on June 1, four days after the company closed a 65-billion-dollar Series H round that valued it at 965 billion dollars. If it lists anywhere near that number, and bankers seem to think it will list above a trillion, it walks straight into the top tier of the entire US stock market on day one. I have been writing this Data and AI history for weeks now, and this is the rare case where a current event slots perfectly into the through-line of everything that came before it.
The revenue curve does not look real
Start with the number that matters, because it explains the valuation better than anything else. Anthropic’s run-rate revenue was about 87 million dollars in January 2024. By the end of 2025, it was around 9 billion. By February 2026, it was 14 billion; it reached 19 billion in March, 30 billion in April, and crossed 47 billion by late May. That is roughly a 540-fold jump in about 28 months. Dario Amodei reportedly called the growth “crazy”. He said it beat the company’s internal forecasts by a factor of eight, which is the kind of thing you say when even the people running the company did not see it coming.
I want to be honest about my own reaction here, because pretending to be neutral would be silly. A curve like that triggers two thoughts at once, and they do not resolve cleanly. The first is that this is real demand: companies are paying for Claude inside actual workflows, coding, support, legal review, research, finance, and they keep paying more. The second is that a 540-fold increase in just over two years is exactly the shape of thing that makes you nervous, because nothing in business usually moves like that without something snapping back. Both thoughts are probably a little bit right.
Where the money is actually coming from
The engine is enterprise, not consumer. That is the cleanest way to separate Anthropic from OpenAI in your head. OpenAI owns consumer reach with ChatGPT; Anthropic built its business selling Claude to companies and developers. The standout is Claude Code, the coding tool, which hit around $ 2.5 billion in annualized revenue by February 2026, with enterprise use accounting for more than half and a customer list that reportedly includes Netflix, Spotify, KPMG, L’Oreal, and Salesforce. When developers adopt a coding agent, and it becomes part of how their team ships software, that spending does not casually churn out. It compounds, the same way Snowflake’s consumption model compounded, the same way every sticky enterprise platform in this whole series has compounded.
There is also a profitability wrinkle worth flagging, because it is unusual for a frontier AI lab. Anthropic has told investors it expects its first profitable quarter around June 2026, the idea being that revenue growth is finally outpacing the brutal cost of compute. If that holds up in the audited filing, it changes the story from burn-the-cash-and-pray to something a public-market investor can actually model. The word if is doing a lot of work in that sentence.
The accounting question nobody should skip
Here is the part that gets glossed over in the breathless coverage. Anthropic reportedly books revenue on a gross basis, which can inflate the headline number relative to peers that report net revenue. The confidential S-1 is exactly where that gets resolved, because public markets do not accept a run-rate figure on faith. They want audited financials, a real revenue mix, cost of revenue, and the risk factors spelled out. A confidential filing is not an IPO; it opens the SEC’s review window, and the actual decision to list depends on whether the numbers survive that scrutiny and whether the market still feels like buying when the window opens, reportedly targeted for around October 2026 on NASDAQ.
This is the same lesson that runs through Snowflake’s record-breaking IPO and the Databricks rounds I covered earlier in this series. A private valuation is a number that a handful of investors agreed on in a room. A public valuation is a number that has to defend itself every single quarter in front of people who can sell. They are not the same thing, and the gap between them is where a lot of pain has historically lived.
Three trillion-dollar debuts in one year
Anthropic is not filing into a quiet market. It is one of three expected trillion-dollar listings of 2026, alongside Elon Musk’s SpaceX, which has already filed and is reportedly pursuing a 75 billion raise at a 1.75 trillion-dollar valuation, and OpenAI, which is preparing its own filing. A PitchBook analyst put it about as sharply as anyone could, calling it the largest concentration of pre-IPO capital ever brought to market at the same time, and warning that the 2026 window could either rival the dot-com era in importance or become the most expensive lesson in narrative-versus-fundamentals that public markets have ever taught. I keep rereading that line. It manages to be bullish and ominous in the same breath, which feels exactly right for this moment.
The timing against OpenAI is not an accident either. Filing first matters, because both companies will be raising tens of billions in quick succession, and there is a real argument that whoever reaches the market first does better, simply because the second one is competing for the same pool of capital after the novelty has worn off. Sam Altman, for his part, said OpenAI will go public when it makes sense and denied there is a race. Make of that what you will.
The governance detail that makes Anthropic different
One thing I do not want to lose in the valuation noise: Anthropic is structured as a Public Benefit Corporation, with a Long-Term Benefit Trust above it, designed to protect the company’s safety mission as it scales. That is an unusual structure for a company about to ask public shareholders for a trillion dollars, because public shareholders generally expect to be the priority, and here there is an explicit structure stating that the mission can override pure shareholder returns. How that holds up under the pressure of quarterly earnings is one of the most interesting open questions of the whole listing, and it is the kind of thing the S-1 will have to address head-on.
What it means for the market
Step back, and this filing is the data-and-AI story I have been tracing, reaching its loudest moment yet. The whole arc, which I tried to pull together in how the data revolution and the AI revolution became one story, has been about data infrastructure becoming the foundation of AI, and the platforms that sit on top becoming some of the most valuable companies on earth. Anthropic filing at 965 billion dollars, profitable or nearly so, built on enterprise demand and a coding agent that companies cannot easily quit, is that arc reaching its public market test. It also drags the entire AI boom into the daylight, because an IPO forces the first real look at the financials behind the hype, and after a year of bubble talk, that look is overdue.
For the rest of us watching, the benefit is clarity. We are about to find out, in audited numbers rather than press releases, whether the revenue is as real and durable as the run rate suggests. The risk is the obvious one: three near-simultaneous trillion-dollar listings represent an enormous bet that demand keeps outpacing the staggering cost of building this stuff, and if that bet wobbles, it wobbles for everyone at once. I am not going to pretend I know which way it breaks. What I do know is that the company writing the model I use to research and draft these very posts is about to open its books to the world, and for once, the most important story in tech is one I get to watch from unusually close up.
Quick disclosure, since it would be weird not to say it: I write a lot of this blog with Claude’s help, so I am hardly a neutral observer of Anthropic. Take the enthusiasm with the appropriate grain of salt, and watch the S-1 when it goes public. That is where the real answers are.