Anthropic Filed Its S-1. Claude Now Has Quarterly Earnings.

Anthropic just filed the paperwork to go public, and the headline number — an $80.7 billion net loss — isn't the story. The real story is what changes for you once a company you depend on answers to analysts instead of just shipping blog posts.

On October 26, Anthropic, PBC filed a public Form S-1 with the SEC, confirming a Nasdaq listing under the ticker "ANT" [1][2]. This is the real filing, not the confidential draft Anthropic quietly submitted back in June. It's the document the public actually gets to read, and it tells you more about Anthropic's incentives than any product launch has this year.

I build client work on Claude. If you do too, this filing is worth five minutes — not because the IPO itself matters to your business, but because the disclosures inside it tell you exactly who Anthropic optimizes for, and it might not be you.

The $80.7 billion loss isn't what it looks like

Here's the number every non-technical outlet is going to run with: Anthropic posted a net loss of $80.7 billion for the first half of 2026. That reads like a company on fire. It isn't.

That figure is driven almost entirely by a non-cash charge from convertible-note accounting — a paper revaluation, not cash leaving the building [2]. The number that actually tells you how the business is running is the operating loss, and it went the other direction: down to $693 million for H1 2026, from $2.1 billion in H1 2025 [2]. Revenue over the same stretch jumped from $1.2 billion to $16.3 billion — a 13x increase [2].

So the real picture: Anthropic's core business is losing less while growing revenue thirteenfold. The scary number on the cover page is an accounting artifact from how convertible notes get marked when a company's valuation moves. Don't let a headline writer's $80 billion sell you an instability story the operating numbers don't support.

Who Anthropic is actually built to serve

The part of the S-1 that matters more than the loss number is the customer breakdown, because it tells you where Anthropic's attention goes.

As of the end of September, Anthropic had more than 11,900 customers spending $100,000 or more a year, over 1,900 spending $1 million or more, and roughly 250 spending $10 million or more annually [3]. Net revenue retention sits around 900% [3] — meaning Anthropic's existing big accounts are expanding their spend dramatically faster than new accounts are replacing churn. More than half of the $100k+ customers now use both the Claude platform and a first-party app like Claude Code or Cowork, up from about 30% at the end of last year [3].

The S-1's own risk language is blunt about what that concentration means: "Any adverse change in demand from our largest customers … could adversely affect our business," and the filing specifically flags the risk of "a significant reduction in spending by, or the loss of, one or more of our largest customers" [3].

A newly public company's incentives point at the 250 accounts writing eight-figure checks. If you're not one of them, you're reading the same risk factors they are — just from the other side.

That's not a knock on Anthropic. It's just what a public filing forces into daylight that a blog post never would. A startup chasing logo diversity has different incentives than a public company managing net revenue retention for analysts every quarter.

OpenAI is making the opposite bet

The contrast with OpenAI is the part that actually tells you something about the industry, not just one company.

Anthropic blinked first. OpenAI filed confidential IPO paperwork back in June 2026 and, instead of following through, is reportedly working on a new private raise of around $30 billion at a valuation CNBC pegs near $1.25 trillion and Bloomberg pegs closer to $1.4 trillion — neither number confirmed by OpenAI, which declined to comment [4]. Anthropic's own target valuation is unconfirmed too; the S-1 itself discloses no share price or raise amount [1][2].

Two frontier labs, two different bets on whether quarterly disclosure helps or hurts a business running on compute costs this large. Anthropic is about to find out what it's like to explain a one-time accounting charge to retail investors on an earnings call. OpenAI is choosing to put that conversation off.

What actually changes for you

Here's the operator version, stripped of the IPO theater:

Earnings calls become a new signal source. Once Anthropic lists, its 10-Qs and quarterly calls will surface real information about pricing pressure, compute constraints, and customer concentration that used to only leak through press coverage or your own invoice. Put the earnings date on your calendar the way you'd track a vendor's status page.

Customer-concentration risk is now public, and it's not flattering to the long tail. If your business isn't writing an eight-figure annual check, the S-1 just told you, in the company's own risk-factor language, that your renewal terms and roadmap influence sit structurally behind the accounts that are [3]. That's not new information about Anthropic's priorities — it's just the first time it's been in writing.

Build for portability, not loyalty. A newly public company optimizing for net revenue retention among its biggest accounts doesn't have the same incentive to protect a small account's workflow that a venture-stage startup chasing logo diversity does. If your Claude integration is hard-wired into prompts and tooling that only work with one vendor, that's a cost you're choosing to carry, not one you're stuck with.

Don't panic-read the loss headline, and don't let your clients panic-read it either. The $80.7 billion is a convertible-note artifact. The number worth watching is the operating loss, and it's shrinking.

None of this means pull out of Claude. It means read the filing the way you'd read a vendor's terms of service: not because you expect the worst, but because the worst case is cheaper to plan for now than to discover during a renewal call.

flowchart TD
  A["Jun 1, 2026<br/>Confidential S-1 draft filed"] --> C["Sep 28, 2026<br/>Prospectus leaks:<br/>losses, growth, risk language"]
  C --> D["Oct 26, 2026<br/>Public S-1 filed<br/>Ticker: ANT, Nasdaq"]
  D --> E{"Reported next steps<br/>(unconfirmed by Anthropic)"}
  E -->|"reported"| F["Roadshow begins<br/>week of Nov 9"]
  E -->|"reported"| G["Target listing<br/>week of Nov 16"]
  D --> H["Your move:<br/>audit Claude portability<br/>before the lock-up expires"]

Sources

[1] Anthropic — Anthropic Files S-1 — anthropic.com

[2] TechCrunch — Anthropic officially files for its public debut — techcrunch.com

[3] TechCrunch — Anthropic's IPO filing reveals its strategy for keeping enterprise customers hooked — techcrunch.com

[4] TechCrunch — OpenAI working on new $30B raise at $1.25T valuation ahead of IPO — techcrunch.com

Story: Anthropic — Anthropic Files S-1. Image via TechCrunch.

If you're running a business on top of Claude, Anthropic's IPO prep isn't a spectator event. It's the first time the company's actual incentive structure has been put in writing for anyone to read, and it's worth five minutes before the roadshow turns it into a stock story instead of an operator one.

The short version

• Anthropic filed a public S-1 on Oct 26, 2026 — Nasdaq ticker "ANT"
• The $80.7B net loss headline is mostly a non-cash convertible-note accounting charge; operating loss actually shrank to $693M from $2.1B year-over-year
• Revenue hit $16.3B in H1 2026, up from $1.2B — and the S-1's own risk language says the business depends disproportionately on roughly 250 customers spending $10M+ a year
• OpenAI is making the opposite bet — staying private with a new ~$30B raise instead of following Anthropic to the public markets
• Anthropic's own valuation target isn't disclosed in the filing itself
• If you build on Claude, the move now is portability — not panic, not loyalty