Anthropic’s operating loss for the nine months to 30 September 2026 was $1,260 million on revenue of $33,242 million, according to the amended Form S-1 the company filed with the SEC on Monday, 2 November. That is 3.8% of revenue, by our arithmetic.
A year earlier, the same table shows an operating loss of $5,224 million on revenue of $2,545 million, or 205% of revenue, by our arithmetic.
The key insight: Anthropic’s amended S-1 shows a nine-month operating loss of $1,260 million on $33,242 million of revenue, 3.8% of revenue by our arithmetic, against 205% a year earlier.
Revenue grew about 13 times between the two nine-month periods while total operating expenses grew about 4.4 times, from $7,769 million to $34,502 million, by our arithmetic from the filing’s table.
Compute and infrastructure, the largest line, rose from $4,381 million to $18,808 million. As a share of revenue that is 172% falling to 57%, by our arithmetic. Personnel costs rose from $1,027 million to $4,863 million, or 40% of revenue falling to 15%.
Sales, marketing and partnerships rose from $322 million to $5,529 million, the one major line that grew faster than revenue: 13% of revenue rising to 17%, as the filing’s own comparison table shows. Other operating and administrative expense rose from $539 million to $5,302 million.
The 2025 period also carried the $1,500 million Bartz settlement charge, which the table shows as nil in 2026. Without it, the 2025 nine-month operating loss would have been $3,724 million, or 146% of revenue, by our arithmetic.
Anthropic operating expense lines and operating loss as a percentage of revenue, nine months to 30 September 2025 vs 2026; percentages are FourWeekMBA arithmetic from the filing. Source: Anthropic PBC, Form S-1/A filed with the SEC on 2 November 2026.
Business Pill · A Cost Test
Business Pill: A Cost Test, checking whether revenue now covers the costs behind it.
From Operating Loss to Net Loss
Below the operating line, the nine-month 2026 table shows interest income of $2,359 million, a $94,324 million change in the fair value of convertible notes, derivative liabilities and SAFEs, and other expense of $10,397 million, ending in a net loss of $104,039 million.
The filing explains the other-expense line, $309 million, $8.4 billion and $1.6 billion in the first three quarters of 2026: it “was primarily driven by the initial fair value recognition and subsequent change in fair value of the assets and liabilities recognized under the partner financing arrangements”. A note to the accounts puts that item at $11,008 million for the nine months, partly offset by a $783 million gain on warrant rights.
The note names the arrangements: note purchase agreements with Google in March 2026 and with Amazon Web Services in April 2026, each including a commitment to sell Series G-2 preferred shares at $259.1364 per share, and a July 2026 hardware supply agreement under which AMD committed to buy, at Anthropic’s election, up to $5.0 billion of its Class N common stock if certain conditions are achieved.
As we read the table, the narrowing came from revenue outgrowing costs rather than costs falling: total operating expenses rose about 4.4 times while revenue rose about 13 times, by our arithmetic.
As we read the bottom half of the table, the net loss is set mostly by accounting marks: $94.3 billion of fair-value changes on notes and an $11.0 billion charge on partner financing arrangements, against a $1.26 billion operating loss.
Anthropic’s amended Form S-1
“was primarily driven by the initial fair value recognition and subsequent change in fair value of the assets and liabilities recognized under the partner financing arrangements”
Three Implications
FOR INVESTORS The nine-month operating loss was $1,260 million, against a net loss of $104,039 million, so the two lines read very differently.
FOR GOOGLE, AMAZON AND AMD The note lists their 2026 agreements with Anthropic as the partner financing arrangements behind an $11,008 million charge.
FOR COST WATCHERS Sales, marketing and partnerships rose from 13% to 17% of revenue, the one major line that grew faster than revenue.
The passages we quote do not explain the operating-margin change line by line beyond the expense comparisons. The note says the fair value of the partner financing arrangements was determined using a probability-weighted expected return method, so the charge depends on the company’s own valuation assumptions. The 3.8%, 205%, 146%, 13 times, 4.4 times and per-line shares of revenue are our arithmetic.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Anthropic’s amended S-1 shows its nine-month operating loss falling to $1.26 billion, about 4% of revenue, from $5.22 billion, about 205% of revenue, a year earlier, as revenue grew faster than every major cost line except sales, marketing and partnerships. Below the operating line, $94.3 billion of fair-value changes on its notes and an $11.0 billion charge on its partner financing arrangements with Google, Amazon and AMD took the nine-month net loss to $104.0 billion.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. All figures and quotations are from Anthropic’s amended Form S-1 filed with the SEC on 2 November 2026. The percentages of revenue, growth multiples and the 146% figure excluding the Bartz settlement are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
