AiStaffo

Anthropic Reports $80.7 Billion First-Half Loss in IPO Draft

Stock photo: laptop desk documents
Photo: Pavel Danilyuk / Pexels

Anthropic's draft IPO prospectus shows a net loss of $80.7 billion for the first half of 2026, on revenue of $16.3 billion. According to reports on the draft, about $72 billion of that loss came from non-cash accounting charges tied to convertible instruments the company issued to raise money. The filing is a draft, the figures can still change, and the company is reportedly targeting a November listing. For businesses that automate work with AI tools, the numbers matter less for the headline loss than for two points: how much the AI vendor spends on computing capacity, and how concentrated its customer base is. Both are disclosed in the draft and both affect the vendor you depend on.

In short

  • Anthropic's draft IPO prospectus shows an $80.7 billion first-half 2026 net loss on $16.3 billion revenue.
  • About $72 billion of the loss is a non-cash charge tied to convertible bonds.
  • The filing is a draft, and reports say a November 2026 listing is being targeted.
  • Compute spending reached nearly $9 billion in the first half, a signal for AI vendor cost pressure.
  • Businesses should review AI vendor contracts for price-change clauses and data export rights.

What happened

According to a draft prospectus reviewed by the Wall Street Journal and reported on 20 October 2026, Anthropic, the San Francisco company behind the Claude models, recorded a net loss of $80.7 billion for the six months to 30 June 2026. Revenue for the same period was $16.3 billion. Reuters also reported on the draft, and the figures match across both accounts. The loss was up from about $13 billion in the first half of 2025.

The bulk of the loss is an accounting item. The draft shows about $72 billion came from a charge reflecting a rise in the estimated fair value of convertible bonds, which holders can exchange for Anthropic shares. The charge does not represent cash spent running the business. Once that item is set aside, the remaining net loss is about $8.7 billion, a figure that reporters derived from the filing rather than one the company stated as an operating loss.

The draft is not final. Anthropic has not announced a listing date. Reports say the company is targeting a November 2026 listing, and Reuters reported that the roadshow could start the week of 9 November, though that timing is not final.

The figures at a glance

MeasureFirst half 2026First half 2025
Revenue$16.3 billion$1.1 billion
Net loss$80.7 billion$13 billion
Non-cash convertible charge within the lossAbout $72 billionNot reported in the sources reviewed
Compute and infrastructure spendingNearly $9 billion$2.3 billion

The table uses figures from the same draft as reported by the Wall Street Journal, Reuters and Benzinga. The first-half 2025 charge is not given in the reports reviewed, so that cell is left open.

Why it matters for businesses that automate work

Most companies that automate routine work with AI do not buy a model directly from its maker. They buy a tool built on top of one, and their cost and service depend on the maker's own economics. A draft prospectus is one of the few places where those economics are written down in detail.

Two parts of the draft stand out. First, compute and infrastructure spending reached nearly $9 billion in the first half, more than the $7.3 billion spent in all of 2025, according to the same draft. Second, reports say commitments to computing capacity exceed $600 billion, with more deals under negotiation. A vendor with commitments on that scale has a strong reason to keep usage growing, and it also has large fixed obligations to meet. The reports reviewed do not say how these costs will reach the prices that business customers pay.

The draft also shows customer concentration. Reports on the filing say around a fifth of first-half revenue came from two customers. For a business, the lesson is general. Any single AI supplier can change its pricing, limits or terms, and a supplier's dependence on a few large buyers can shape its priorities.

What changes in practice

For a business that automates work, the filing does not require a new tool or a new process. It is a reason to check what already exists. The steps below apply whether the company has five staff or several thousand.

  1. List every AI tool that touches invoices, customer follow-ups, reports or staff records, and name the model behind each one.
  2. Read the price-change and usage-limit clauses in each contract, and note how much notice the vendor must give.
  3. Confirm you can export your prompts, templates, logs and outputs in a usable format.
  4. Keep one tested fallback for each critical workflow, even if it is a slower manual process for a few weeks.
  5. When the final prospectus is public, read its risk factors section before you renew a large contract.

Automation also has limits here. An AI worker can draft a reconciliation or a follow-up email on schedule, but it cannot judge whether a supplier's accounting or a vendor's financing will hold up. Those judgments stay with the finance lead. Results from any automation depend on the business's own volumes and error rates, which a review of its own figures will show.

What to watch next

  • The public S-1 filing, once Anthropic makes it available, which will replace the draft figures.
  • Whether the final figures change, especially the $72 billion charge and the revenue total.
  • The listing date and the valuation range. Reuters reported some investors discussing about $1.5 trillion, while earlier expectations were between $1.8 trillion and $2 trillion.
  • Any change in the compute commitments or in the share of revenue from its largest customers.

How AiStaffo would automate this

AiStaffo would start with a list of the AI tools and vendors that touch your invoices, follow-ups, reconciliations and reports, so each dependency is visible in one place. AI workers would then be connected to the systems where that routine work already happens, and they would run the scheduled tasks, such as drafting payment reminders or matching records, without a person starting each one. A person in finance or operations still approves exceptions, decides on vendor changes and reviews any figure that goes to a lender or regulator. Book a free automation audit to see which of your workflows depend on a single AI supplier and what a fallback would involve.

Questions people ask

What is the Anthropic IPO filing loss?
According to a draft prospectus reviewed by the Wall Street Journal, Anthropic recorded a net loss of $80.7 billion in the first half of 2026. Revenue for the same period was $16.3 billion. About $72 billion of the loss came from non-cash accounting charges tied to convertible instruments.
Did Anthropic lose $80.7 billion in cash?
No. The reports reviewed say about $72 billion of the loss came from accounting charges linked to the rising value of convertible bonds. Those charges are not cash spent on operations. Compute and infrastructure spending was nearly $9 billion in the same half.
Is the Anthropic IPO confirmed for November?
Not officially. Reports say the company is targeting a November 2026 listing, and Reuters reported that the roadshow could start the week of 9 November, though the timing is not final. The prospectus is still a draft, so figures can change.
Why does an AI company's IPO filing matter to my business?
Most businesses use AI through tools built on a model maker's platform. The draft shows large computing commitments and revenue concentrated in a few customers. Both can influence an AI vendor's pricing, limits and priorities, so it is worth checking your own contracts.
What should a business do after reading this news?
List the AI tools that handle routine work, check their price-change and data export terms, and keep a manual fallback for critical workflows. Review the final prospectus's risk factors section when it is public. Decisions about which vendor to use should rest on your own cost and volume figures.
Can automation fix the risks shown in the filing?
No. Automation can run scheduled tasks such as reminders and reconciliations, but it cannot judge a vendor's financing or accounting. People still review exceptions and vendor decisions. The value of any automation depends on your own volumes and error rates.

Book a free automation audit

Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.

We use these details only to reply to this request and prepare your audit, and keep them for up to two years. We send no marketing unless you tick the box above. Privacy policy

anthropicai ipoai spendingbusiness automationvendor risk