A leaked draft of the Anthropic S-1 filing paints a picture of a company growing at a wild pace while struggling to find enough computing power to keep up. The document, which the company would use to go public, says the big cloud providers that host Claude now increasingly treat Anthropic as a rival. It also lays out huge losses, heavy reliance on a few customers, and new legal risks tied to AI behavior.
The draft was obtained and reported by The Wall Street Journal on October 20. Anthropic has not published the filing, and it did not immediately respond to requests for comment.
What the Anthropic S-1 Filing Reveals
An S-1 is the paperwork a US company files before an initial public offering. It has to spell out the business, its finances, and the risks investors should know about. That makes it one of the most honest looks you will ever get at a private company.
Here are the headline numbers reported from the draft:
- Revenue of about $16.3 billion in the first half of 2026.
- A net loss of $80.7 billion over the same six months, driven largely by accounting charges.
- Roughly one fifth of first-half sales came from just two customers.
- Potential spending of more than $750 billion on cloud contracts, including some deals still being negotiated.
The company also warns that a significant share of its revenue will keep coming from a limited number of customers for the foreseeable future. That kind of concentration is a classic risk flag in any IPO document.
Why Cloud Giants Now See Anthropic as a Rival
The most striking detail is about computing power. According to the reported filing, Anthropic is having trouble securing enough capacity from Amazon Web Services, Google, and Microsoft. Those same companies sell Claude to their customers, but they also build and sell their own AI models.
In other words, the landlords are also competitors. As Claude grows, the hyperscalers have less reason to hand over their scarcest chips to a company competing for the same business customers.
Takeaway: the companies that power Claude are also racing to replace it, and that tension now shows up in Anthropic’s own risk disclosures.
Leasing Its Own Data Centers
To fill the gap, Anthropic is leasing its own data centers and working with third-party data center operators. The filing reportedly flags a worry here too: some of those operators have a limited operating history. That could make them less dependable than the established cloud giants.
This fits a pattern we have been tracking. Anthropic was recently reported to be in talks for a second Nscale deal worth about $45 billion, one of several large bets on newer infrastructure partners.
The Numbers at a Glance
| Item | Reported figure | Period |
|---|---|---|
| Revenue | About $16.3 billion | First half of 2026 |
| Net loss | $80.7 billion | First half of 2026 |
| Third-quarter revenue | $16.8 billion, up 45% from Q2 | Q3 2026 (shared with investors last week) |
| Possible cloud spending | More than $750 billion | Coming years, some deals pending |
| Top two customers | About 20% of sales | First half of 2026 |
The third-quarter figure came separately, from what Anthropic told prospective investors last week. We covered that jump in our report on Anthropic’s Q3 revenue.
Liability Risks From Harmful AI Behavior
The draft also expands on earlier warnings about catastrophic risks. It reportedly says Anthropic could be held liable for damage caused by harmful AI behavior, and that such costs could exceed its available insurance coverage.
That is a notable admission for a company that has built its brand around safety. It matches the wider pressure on AI labs this month, including the FTC safety probe into Anthropic and OpenAI.
How the Anthropic IPO Timeline Is Shaping Up
Anthropic is reportedly aiming for a November IPO. The leaked draft adds detail to figures that other outlets first reported in late September, and a formal public filing would likely follow before any listing.
Not everyone is eager. New York City’s top money manager recently asked the company to pause its IPO.
What This Means for You
If you use Claude every day, none of this changes the product today. But it does hint at what could shape your experience in the months ahead.
- Capacity limits matter: tight computing supply can mean usage caps, slower rollouts, or waitlists for new Claude features.
- Cloud access may shift: if you use Claude through AWS, Google Cloud, or Microsoft, watch for changes in pricing or availability.
- Have a backup: teams that depend on one AI model should test alternatives, as our model comparison guide explains.
- Read the real filing: once Anthropic files publicly, the official S-1 will be the best source for exact numbers.
Sources
Original report on the leaked draft from The Wall Street Journal. Additional reporting from TechCrunch and Yahoo Finance.
Conclusion
The leaked Anthropic S-1 filing shows a company with staggering revenue growth, even bigger losses, and a growing dependence on rivals for the computing power it needs. Its cloud partners are now competitors, its newer data center partners carry their own risks, and a handful of customers drive a large share of sales. For investors, those are serious questions ahead of a possible November listing. For Claude users, the key thing to watch is whether Anthropic can secure enough capacity to keep up with demand.
