Anthropic S-1 Explained: The $80.7B Loss, the $600B Compute Bet, and the $1.5T IPO

The biggest AI IPO in history is about to go public — and the first look at its finances is already shaking up Wall Street. Yahoo Finance reported on October 21 that Anthropic could publish its S-1 prospectus as early as Friday, October 23, moving the Claude maker one step closer to a November listing that could be the largest stock market debut ever. But the draft figures — reviewed by Reuters and The Wall Street Journal — need careful reading: the headline $80.7 billion loss is mostly not what it looks like.

The headline numbers

Here is what the draft prospectus reportedly contains, per Reuters' review of the document:

| Metric | Reported figure | |---|---| | H1 2026 net loss | $80.7 billion (vs. $13B a year earlier) | | Non-cash accounting charge | ~$72 billion | | H1 2026 revenue | $16.3 billion (vs. $1.1B a year earlier) | | Q3 2026 revenue | $16.8 billion, up 45% sequentially | | H1 2026 compute/infrastructure spend | Nearly $9 billion (vs. $2.3B a year earlier) | | Decade-long infrastructure commitments | At least $518 billion | | Expected listing | November 2026; investors weighing ~$1.5T valuation |

A few caveats before we go further: this is a draft prospectus, reported by journalists — not a public SEC filing and not confirmed by Anthropic. The public S-1 itself has not yet appeared on EDGAR. Everything below should be read as "reported," and Anthropic did not respond to Reuters' request for comment.

Why the $80.7B loss is mostly paper

The single most misunderstood number in this filing is the loss itself. About $72 billion of the $80.7 billion came from an accounting charge, not from running the business.

Here's what happened: Anthropic raised capital using bonds that can convert into company shares. When Anthropic's valuation rises, the estimated fair value of those bonds rises too — and accounting rules require the company to record that increase as a loss, even though no money left the company. In other words, the worse the headline loss looks, the better Anthropic is doing at being valued more highly. It's a paradox of paper accounting: success at fundraising makes the income statement look catastrophic.

This isn't new for Anthropic either. Reuters reported that a similar charge drove most of its $42 billion net loss in 2025 — the estimated value of those notes tripled year over year as the valuation climbed. Strip the charge out and the 2025 picture was an operating loss of about $8.1 billion on $4.6 billion of revenue. That operating loss is the number that actually describes the economics of the business.

Why this matters to you: when the headlines scream "biggest loss ever," the takeaway isn't that Anthropic burned $80 billion in six months. The real story is underneath: revenue growing roughly 15x year over year, compute spending scaling with it, and a business that is still loss-making on operations while pouring money into infrastructure.

The $600B compute bet

If the accounting loss is paper, the infrastructure commitments are very real. Reuters confirms Anthropic has committed to spending at least $518 billion over a decade on cloud services, computing and infrastructure — and reports of the draft put total future compute commitments above $600 billion, with even more under negotiation.

Some context on the scale: the company spent nearly $9 billion on computing capacity and infrastructure in just the first half of 2026, up from $2.3 billion a year earlier. Revenue is scaling faster than the operating loss, but compute spending is growing nearly in lockstep with revenue — a pattern that is unusual for software companies, where margins normally expand as the business grows.

This is the structural story of the frontier-AI business: the money goes to chips and data centers, and it has to keep flowing for the revenue to keep flowing. It also explains why the draft reportedly shows about one-fifth of first-half revenue coming from just two customers — when your costs are this fixed and concentrated, customer concentration is a real risk factor, not a footnote.

The $1.5T IPO question

Reuters reported last week that some investors are now weighing a valuation of around $1.5 trillion — down from earlier talk of roughly $2 trillion — after the financial disclosures. The offering is expected to top SpaceX's June IPO, and Yahoo Finance called it "widely viewed as a crucial test of investor appetite for the artificial intelligence boom."

Two things to watch:

1. Timing. Under SEC rules, companies that file confidentially must make their documents public 15 calendar days before they start pitching the deal to investors. A public filing as early as Friday clears the path for a November listing. But Anthropic has announced nothing — no date, no ticker, no price, no valuation. Treat every date as reported, not confirmed.

2. The Q3 signal. At an investor meeting on October 14, Anthropic told prospective IPO investors that third-quarter revenue jumped 45% to $16.8 billion from the previous quarter. That eased some concerns about costs — but some investors were disappointed by the slowing sequential growth rate, since revenue had more than doubled from Q1 to $11.6 billion in Q2. Reuters notes the company is said to be profitable on an adjusted basis, but the official audited picture only arrives with the public filing.

The stakes extend beyond Anthropic. Reuters framed the IPO as a referendum on the AI boom itself: a cool reception could weigh on the wider market, and investors see the listing as a precursor to a potential OpenAI listing next year.

What remains uncertain

• The public S-1 has not been filed yet — the filing date is a report, not a fact.
• The valuation ($1.5T vs. $2T), the raise size (up to $100B per the WSJ), and the November date are all press-reported; none is company-confirmed.
• How much of the compute commitments are cancellable versus locked in will be in the public filing's risk disclosures — that's the document worth reading closely when it lands.

Bottom line

The S-1 story so far is really two stories. One is an eye-catching accounting loss — $80.7 billion, mostly paper, driven by a valuation that kept rising. The other is a company generating real revenue at extraordinary speed while committing hundreds of billions to the infrastructure needed to stay in the AI race. Which story the public markets believe will decide whether this becomes the biggest IPO in history — and what that means for every other AI company behind it.

This post follows Anthropic's Public Benefit Fund restructuring — a separate October development in how the company is organized. For more on Anthropic's product roadmap, see the latest Claude Opus 6 rumors, and for the broader IPO landscape including its chief rival, read about OpenAI's Dots AI agents.

Sources: Reuters (Echo Wang, Oct. 20, 2026) via live.euronext.com; Yahoo Finance (Oct. 21, 2026) via PYMNTS; The Wall Street Journal's reporting relayed by Yahoo Finance and PYMNTS. All figures come from a draft prospectus reviewed by these outlets, not a public filing. Featured image is an AI-generated illustration, not a photograph.