Anthropic PBC on Monday published a revised version of the S-1 filing for its upcoming public offering, adding third-quarter financial results that were missing from the original document released the previous Monday. The filing shows the company booked $17 billion in revenue for the three months ended Sept. 30, with losses widening as it spends heavily on computing infrastructure.
Revenue was more than 11 times the year-earlier figure, but growth slowed on a sequential basis. The top line rose 47% from the prior quarter, compared with 144% three months earlier. The results land as Anthropic faces multipronged competition from open-source models, OpenAI Group PBC and tech giants including Google LLC, whose newest model, Gemini 4 Carbon, outperforms Claude Opus 5.5 across multiple benchmarks.
Anthropic absorbed a $5 billion charge in the quarter after demand for computing hardware came in higher than expected, forcing it to buy equipment at a premium. Hardware purchased on short notice costs more than the capacity commitments the company typically books 12 to 18 months in advance. A separate $2 billion charge tied to its non-cash employee donation matching program contributed to a $536 million operating loss for the quarter, down from $974 million in operating income three months earlier. Adjusted operating income, which excludes stock-based compensation and charitable contributions, rose $300 million quarter over quarter to $2.5 billion.
As of Sept. 30, Anthropic carried $895.4 billion in outstanding infrastructure commitments, nearly a third of which stemmed from leases signed with Google LLC and Broadcom Inc. In April, the two companies agreed to supply Anthropic with several gigawatts of computing power running on Google's TPU chips, which Broadcom helped design. Anthropic also holds computing deals with Amazon.com Inc., Oracle Corp. and SpaceX Technologies Corp. that are each worth more than $100 billion. Purchase commitments climbed from $55 billion at the end of 2025 to about $314 billion in the third quarter, as the company plans to both rent more capacity and build its own data centers.
The revised filing also updates Anthropic's efforts to mitigate AI risks. The company said it is considering requiring human review and approval of certain research tasks now performed automatically by its models, and that it believes its models will soon be able to dramatically accelerate AI development.
Anthropic has not disclosed how much it intends to raise in the offering. Bloomberg reported that the company hopes to at least match the $86.2 billion SpaceX pulled in when it listed shares earlier this year, while Reuters reported that Anthropic is seeking a valuation of $1.5 trillion to $2 trillion.
More company and startup news from TechManNews.
