TECH & SCIENCE
It took a little over five years to go from a handful of researchers to a company Wall Street is whispering could be worth two trillion dollars. This week, Anthropic — maker of the Claude family of AI models — finally let the public look under the bonnet.
On Monday 26 October 2026, the company made its S-1 registration statement public, the document every American company must file before it can sell shares on a stock exchange. It plans to list on the Nasdaq under the ticker “ANT,” with Goldman Sachs and Morgan Stanley leading a syndicate of more than twenty underwriters. Reports put the target valuation somewhere between $1.5 trillion and $2 trillion. Trading could begin in mid-to-late November, according to press reports, though no date has been set.
The prospectus runs to more than 300 pages. We read the coverage so you do not have to. Here is what is inside — the jaw-dropping growth, the jaw-dropping losses, and one warning that stopped even seasoned market watchers in their tracks.
A Rocket Ship in the Revenue Column
Start with the number that made investors sit up: $16.3 billion. That is Anthropic’s revenue for the first six months of 2026 alone — more than three times what it booked in the whole of 2025 ($4.6 billion), and roughly fourteen times the $1.2 billion it made in the first half of last year.
Rewind further and the curve looks almost made-up. According to analysis of the filing, Anthropic’s revenue was about $41 million in 2023 and $386 million in 2024. Within the first half of 2026 the quarterly pace jumped from around $4.7 billion in the first quarter to $11.6 billion in the second, based on figures reported from the filing.
The customer list tells the same story. As of 30 September, more than 11,900 customers were each paying the company at least $100,000 a year. More than 1,900 pay at least $1 million. About 250 are writing cheques of $10 million or more. Revenue per employee, according to one live-blog summary of the filing, topped $6.6 million, with a headcount of roughly 5,300 — a productivity figure that would make any chief executive in the Fortune 500 weep into their spreadsheet.
The company also claims, in the filing, that the number of tasks AI systems can reliably complete on their own is doubling roughly every four months — a bold statement that sits at the heart of its growth pitch.
That $80.7 Billion ‘Loss’ — Don’t Panic (Yet)
Then comes the number that looks like a typo. Anthropic reported a net loss of $80.7 billion for the first half of 2026. That is more than the annual GDP of many countries.
Before you rush to sell your tech stocks, the fine print matters. According to one detailed analysis of the filing, about $72.4 billion of that loss is a non-cash accounting charge. It comes from revaluing convertible notes and similar instruments: as the company’s valuation soared, the value of what it owes those early investors soared with it, and accountants forced that onto the loss line.
Strip that out and the operating picture is dramatically better. Reports put the first-half GAAP operating loss at about $693 million, compared with $8.1 billion for the whole of 2025. One account says the company swung to a roughly $974 million operating profit in the second quarter after a $1.7 billion loss in the first. Different outlets summarising the same 300 pages have reported some figures differently, so readers should check the final numbers on the SEC’s EDGAR database. But the direction of travel is clear: the losses are shrinking fast.
Cash is plentiful. Depending on the summary you read, Anthropic holds well over $100 billion in cash and short-term investments, funded by roughly $141 billion raised since its founding in 2021.
The Compute Bill That Could Power a Small Nation
Where does all that money go? Mostly into computing power. The filing says Anthropic’s business depends on specialised chips and data-centre capacity bought years in advance from a small number of suppliers — some of which are also investors, customers or even competitors.
One live summary reports compute costs of $5.6 billion in the second quarter alone, up from $3.4 billion in the first. Another reports that compute commitments stretch as far as 2045, though the total dollar figure is left blank in this draft. A separate weekly roundup puts the mostly non-cancellable commitments at more than $475 billion over seven to ten years.
There is also a deal with SpaceXAI. Under agreements described in the coverage, Anthropic has committed up to $120.5 billion through 2029 — about $3.1 billion a month once the capacity is fully delivered, which is expected in December 2026. And through an initiative called “Anthropic Direct,” the company is leasing and operating dedicated data centres, reducing its reliance on cloud giants such as Amazon and Google.
In other words: this is a company betting hundreds of billions of dollars that demand for AI will keep exploding. If it is right, the payoff is historic. If it is wrong, those fixed commitments become a very expensive anchor.
Who Really Controls ANT?
Here is where the filing gets truly unusual. Anthropic is a Delaware public benefit corporation. Its stated legal purpose, in the company’s own words, is “the responsible development and maintenance of advanced AI for the long-term benefit of humanity.” The word “humanity” appears 21 times in the document, according to one count.
That is not just branding. Ordinary investors buying Class A shares will get one vote each — but together those shares will hold less than half of the voting power. Two special single shares do the real steering. The Long-Term Benefit Trust, a body of five financially disinterested trustees, holds one Class T share that generally elects a majority of the board. And a Founder LLC representing the seven co-founders holds a Class F share carrying 50.1 per cent of the vote on most matters.
The result: Anthropic will be a “controlled company” under Nasdaq rules. If you buy ANT, you are buying a slice of the profits — not a seat at the table.
There are sweeteners. Five per cent of fully diluted shares will go into a public benefit fund, and the board has approved a framework to issue up to an extra two per cent each time the share price climbs 50 per cent above its prior milestone. The co-founders have pledged 80 per cent of their personal equity to philanthropy. Early lock-up releases are also unusual: some insiders could sell small amounts almost immediately, with releases staggered across more than twenty dates over a year.
The Warning Buried in the Risk Factors
Every prospectus has a risk-factors section that reads like a horror anthology. Anthropic’s stands out. The company warns that its AI models might pose a “catastrophic or existential risk to humanity” if misused or given too much autonomy.
AI researcher Dave Banerjee flagged on X that the filing suggests the entire insurance industry may not be able to cover Anthropic’s liability in a worst-case AI incident. The company also tells investors it may deliberately delay model releases for safety reasons — even if rivals race ahead — and warns of an “evolving and uncertain regulatory and political environment” that could raise its costs or restrict its operations.
That candour is either refreshing or terrifying, depending on your mood. For a company that has built its brand on safety, writing the risk down in a legal filing is consistent. For a first-time investor, it is a reminder that you are not buying a software firm. You are buying a stake in a technology its own makers say needs careful handling.
The Rivalry: OpenAI Is Watching
No Anthropic story is complete without its great rival. OpenAI, according to reports, is chasing a valuation of around $1.25 trillion in a pre-IPO funding round and may not list until early 2027. That means Anthropic could grab the first-mover prize of the AI public-market era.
The competition is playing out on product shelves too. This week OpenAI launched GPT-6.1 Terra, a cheaper workplace model priced at $1 per million input tokens and $5 per million output tokens, claiming big cost advantages over rivals. Chinese lab Moonshot is preparing to release the full weights of its open Kimi K3.1 Pro model. The race is on every front — price, performance and trust.
Markets seemed to approve of the filing. One live blog noted the Nasdaq Composite up about 0.8 per cent at midday after the news landed. Anthropic had been valued at $965 billion as of May, so a $1.5–2 trillion target would mean investors expect it to nearly double again in months.
Expert Perspective
Market analysts quoted in live coverage of the filing were cautiously enthusiastic. Nancy Tengler of Laffer Tengler Investments and Avery Marquez of Renaissance Capital were among those weighing demand and competition, with the broad view being that appetite for the stock will be intense — but that the valuation leaves very little room for disappointment.
The sceptics’ case is easy to make. A company with $80.7 billion in headline losses, hundreds of billions in long-term compute obligations and a founder-controlled voting structure is not a conventional bet. The bulls’ case is just as easy: revenue is growing at a pace rarely seen in business history, margins are turning, and AI is being adopted by corporations at breakneck speed.
Note also that the filing is preliminary. Share count, price range and third-quarter results are still blank, and will arrive in an amended filing before the roadshow.
What This Means for You
If you are an everyday investor, you may get a chance to own a piece. Retail brokerage allocations have not been announced, but IPOs of this size often include some. Remember, though, that a hot IPO can swing wildly in its first days, and nothing here is financial advice — speak to a qualified adviser before putting a penny in.
If you use AI at work, the IPO means one more thing: pressure. Public companies must hit quarterly numbers. Expect sharper pricing battles, more aggressive product launches and more AI features embedded in the software you already use.
If you are simply a bystander, the filing is a rare window into how fast this industry is moving — and how seriously its own builders take the risks. When a firm tells its shareholders that its product could be catastrophic if mishandled, the rest of us are entitled to pay attention.
The Bottom Line
Anthropic’s S-1 is one of the most anticipated corporate documents of the decade, and it does not disappoint. It shows a business growing faster than almost any in history, spending money at a scale that boggles the mind, and governed by a structure designed to keep mission ahead of markets.
Whether ANT becomes the defining stock of the AI age or a cautionary tale will depend on a handful of things: whether those compute bets pay off, whether competitors like OpenAI close the gap, and whether regulators leave the field open. For now, the countdown to the ticker has begun.
Disclosure: this article was written by an AI model developed by Anthropic, the company it covers. All figures come from published press and analyst summaries of the filing, which differ in places; consult the SEC’s EDGAR database for the official numbers.
