01What the Anthropic S-1 Filing Is and What It Says
The Anthropic S-1 filing is the document Anthropic, the company behind Claude, filed with the SEC on October 26, 2026, so it can sell shares on the stock market. Inside it, the company warns that its own AI could pose catastrophic or existential risks to humanity.
An S-1 is the document every company must file before it goes public. It opens the company's books to the world: revenue, losses, costs, risks and who controls the company.
Here are the key deal details from this episode:
- Anthropic plans to list on the NASDAQ under the ticker ANT.
- Goldman Sachs and Morgan Stanley lead the deal, with around 22 banks behind them.
- The Wall Street Journal reports trading could start the week of November 16.
- Some reports say the raise could be as big as $100 billion.
The filing still leaves the share price, the number of shares and the size of the raise blank. Those details get filed closer to the listing. Even so, the market is treating the listing as close to certain, with some people putting the chance of an IPO announcement this year at 95%.
02The Existential Risk Warning in the Anthropic S-1 Filing
The most unusual part of the Anthropic S-1 filing is the risk section. The filing runs about 335 pages, and close to 100 of them are about risks.
Some risks are normal for any company going public, such as keeping staff, competition and regulation. But this filing goes much further. It says Anthropic's models could pose catastrophic or existential risks to humanity.
Existential means a risk to the survival of humanity itself.
That phrase sits in a legal document where lawyers check every word, because a company can be sued for leaving risks out. Here is what else the risk section covers:
- The word "humanity" shows up 21 times.
- Bad actors could use AI to help develop chemicals.
- Authoritarian governments could use the technology against their own citizens.
- Allegations about its technology could bring in regulators and damage its reputation.
- The whole insurance industry might not be big enough to cover one catastrophic AI incident.
- Increasingly strict AI regulation is flagged as a business risk.
That last insurance point is striking. If something went badly wrong, there might be no insurer on earth able to pay for it.
03The Pentagon Fight and Why Anthropic Warns So Loudly
The filing also covers a fight with the US government. The Pentagon labeled Anthropic a supply chain risk after Anthropic refused to allow unrestricted military use of its technology.
- There are two court cases about it.
- The courts have split on them.
- Both cases are now on appeal.
None of the risk talk is new for Anthropic. It matches what its CEO, Dario, has said in public for years. Back in September, he called on the whole industry to slow down development so safety could keep up.
So the company building some of the most powerful AI in the world is also the one warning about it the loudest.
Not everyone agrees with the warnings. Some experts say these fears are overblown, and some say they are just PR. That tension sits at the center of the whole IPO.
Here is the simple version of that tension:
- The more powerful Claude gets, the more valuable the company becomes.
- The more powerful Claude gets, the bigger the risks the company warns about.
Both of those things are true at the same time, which is why the risk section is so long and why the company is set up in such an unusual way.
04Who Controls Anthropic After the IPO
Anthropic is a public benefit corporation. That means it is legally allowed to put its mission ahead of pure profit. The S-1 shows how the founders plan to keep that mission in charge after the company goes public.
- Anthropic has seven co-founders, including Dario and his sister Daniela, who is the company's president.
- Together the founders will hold 50.1% of the votes.
- They do this through one special share called a Class F share.
- So even after the IPO, the founders stay in control.
There is a second layer of control. A group called the Long Term Benefit Trust elects a majority of the board. The trust exists to stop money from outweighing public safety.
As for what the founders get, reports say each one could hold about 1.4% of the company. At the top valuation being talked about, that could be worth up to $28 billion each.
For anyone watching the IPO, this setup matters. Public shareholders will own part of the company, but they will not control it. The founders and the trust keep the final say on how fast and how safely Claude is built.
05Anthropic Revenue Growth in the S-1 Filing
The revenue numbers in the Anthropic S-1 filing are the ones everyone is talking about. They show growth that is rare for any company, let alone one this size.
- First half of 2025: about $1.2 billion in revenue.
- First half of 2026: $16.3 billion, about 14 times bigger in one year.
- First quarter of 2026: $4.7 billion.
- Second quarter of 2026: $11.6 billion, more than double in a single quarter.
- All of 2025: about $4.6 billion.
Put that together and Anthropic made more than double its whole 2025 revenue in just the second quarter of 2026. Most companies would be thrilled to grow 20 or 30% in a year. Anthropic grew more than 100% in three months.
The bulk of this money comes from businesses paying for AI that does real work, such as writing code and sorting files. More than 1,900 customers now spend over a million dollars a year each with Claude.
The growth also sped up through the year, which is the opposite of what usually happens as a company gets bigger. That is a big part of why the market expects the listing to draw so much attention.
06The $80.7 Billion Loss and the First Operating Profit
The headline that scared many people is the net loss of $80.7 billion for the first half of 2026. At first glance it sounds like the company is on fire. The filing tells a different story.
About $72.4 billion of that loss comes from one accounting charge: the change in value of convertible notes, derivatives and SAFEs. In plain English, those are early investment deals.
- Investors gave Anthropic money in the past.
- In return, they get shares later.
- As the company's value shoots up, those deals become worth much more.
- Accounting rules say that increase must show up as a loss on paper.
- So the faster Anthropic grows, the bigger the paper loss gets.
No cash leaves the bank for that $72.4 billion.
A better number to watch is operating income: what the business makes from its products minus what it costs to run. In the first quarter of 2026, Anthropic had an operating loss of about $1.7 billion. In the second quarter, it made an operating profit of $974 million, the first in the six quarters the filing shows.
It also generated $3.9 billion in cash from operations in the first half, and CNBC reports an implied gross margin of around 85%.
07Claude Code, Claude Co-Work and How Customers Grow
Two products keep coming up as the drivers of growth: Claude Code and Claude Co-Work.
- Claude Code is Claude working inside your computer to build things. You describe what you want, and it writes the code and runs it.
- Claude Co-Work does the same kind of thing for everyday people. You hand it a task, and it gets on with it.
The big change is that people stopped asking AI questions and started giving AI jobs.
The filing shows how that spreads inside a business. A number called net revenue retention is about 900%. In normal words, businesses that paid Anthropic a year ago now pay about nine times more on average.
That happens when a company tries Claude on one team, it works, and they roll it out to everyone. One user becomes ten, ten becomes a department, and a department becomes the whole company.
Efficiency is high too. Anthropic has around 5,300 employees, nearly triple a year ago, while revenue grew about 13 times. Revenue per employee is now over $6.6 million a year, up from about $1.4 million.
For comparison, Nvidia is around $7.2 million, Meta about $3.6 million, Apple around $3.1 million and Amazon around half a million.
08What the Anthropic S-1 Filing Means for Your Business
One more line in the Anthropic S-1 filing matters for anyone using AI at work. Anthropic says the number of tasks AI can reliably complete on its own is doubling about every four months.
Doubling every four months works out to about eight times more in one year. So a year from now, the jobs AI can handle alone could be around eight times what it handles today. That is Anthropic's own claim in a legal document, and it is the same reason the risk section is so long.
Here is what this means if you run a business:
- Businesses are already paying for Claude to do real work, not just answer questions.
- Companies that start with one team tend to spread Claude across the whole business.
- The range of tasks AI can do alone is growing fast, so the jobs you can hand off will keep growing.
A simple way to get started:
- Set up Claude Code or Claude Co-Work.
- Pick one real task in your business to hand over first.
- Build one agent workflow around it, then expand.
Julian Goldie's AI Profit Boardroom includes a 30-day Claude roadmap for exactly these steps, plus weekly coaching calls and daily tutorials on each new Claude feature.