Anthropic prospectus puts 2025 loss at $42 billion as revenue rises 12-fold

Anthropic’s revenue rose 12-fold to nearly $4.6 billion in 2025 while it posted a $42 billion net loss, Reuters reported on 28 September after reviewing its IPO prospectus.

Anthropic’s $42 billion net loss

The net loss included a roughly $34 billion accounting charge arising from a higher estimated value for financing that could eventually convert into Anthropic shares. That charge was separate from the money spent operating the business. Excluding writedowns of liabilities mostly associated with earlier fundraising, Anthropic lost more than $8 billion on an operating basis.

The figures give prospective public investors two measures of the cost of Anthropic’s expansion: the net result, affected heavily by the value of potential future shares, and the operating loss excluding those fundraising-related writedowns. Revenue grew sharply in the same year that both measures remained negative.

Anthropic held $20.28 billion in cash, cash equivalents and short-term investments as of 31 December. The company’s plans for cloud and computing capacity put its spending requirements alongside the revenue growth disclosed in the prospectus.

The $518 billion infrastructure plan

Anthropic plans to spend $518 billion on cloud, computing and infrastructure obligations in the coming year, according to the prospectus as reported by Reuters. The figure is a forward-looking plan, distinct from its recorded expenditure for 2025.

Compute and infrastructure cost Anthropic $7.33 billion in 2025, three times the 2024 amount. Those costs accounted for more than half of total operating expenses of $12.65 billion. Compute and infrastructure therefore accounted for most of Anthropic’s operating expenses.

Amazon and Alphabet’s Google were early strategic partners. Both invested billions of dollars in Anthropic and supply cloud infrastructure used to train and deploy Claude. Anthropic also competes with Google, Meta and SpaceX’s xAI in building AI infrastructure, as well as with OpenAI for business customers and talent.

Anthropic released its Opus 5.5 model in September as it sought to counter OpenAI’s momentum following the launch of GPT-6 Astra. The two companies are also pursuing public listings: OpenAI confidentially filed for an IPO in June and is expected to list by early 2027, according to media reports cited by Reuters.

Two customers supplied nearly a quarter of revenue

Nearly a quarter of Anthropic’s 2025 revenue came from two customers, the company said in the prospectus, according to Reuters. Anthropic warned that many of its biggest clients lack long-term contracts and could reduce their spending or stop buying from it.

Those customer relationships matter to a company planning a substantial increase in infrastructure obligations. The prospectus places the concentration of last year’s sales beside a warning that spending by large clients can change, even as Anthropic commits resources to serving demand for its models.

Anthropic targets a valuation above $2 trillion

Anthropic’s public offering could value it at more than $2 trillion, more than double the company’s own estimated valuation of $965 billion in May. AI Affairs previously reported on Anthropic’s pursuit of a $2 trillion IPO and its proposal for founder voting control ahead of a listing.

Anthropic confidentially submitted a draft S-1 to the Securities and Exchange Commission on 1 June, Forkast reported. Its account said the filing had not appeared in the SEC’s public database as of 26 September. The financial figures now available to prospective investors were reported from a prospectus seen by Reuters.

The proposed valuation would also exceed the $1.77 trillion valuation at SpaceX’s recent IPO. Its shares gained 19 per cent to $160 on their 12 June debut, compared with the $135 IPO price. They were trading at around $147 when Reuters reported the Anthropic figures.