Anthropic is “seeking to go public as soon as the middle of November,” Bloomberg reported on October 1. Formal marketing to investors could begin as soon as the week of November 9, and prospective investors put fair value at $1.8 trillion to $2 trillion. Bloomberg also cautioned that “deliberations are ongoing and the timeline could continue to shift.” As of October 17 no public registration statement appears on SEC EDGAR, so the filing remains confidential.
On October 14 the company gave institutional investors its third-quarter figures, CNBC reported, citing people familiar with the matter. Revenue reached $16.8 billion, up from $11.6 billion in the second quarter and $4.7 billion in the first. CNBC said the company was profitable on an adjusted basis. None of these figures has been audited in public yet.
The debt behind the compute
The growth runs on borrowed capacity. The Financial Times reported a $60 billion debt package, known as Project North Star, to procure about 1.5 gigawatts of Google-designed chips. Blackstone funds are taking $9 billion of an $18 billion junior tranche at a 10% coupon, secured by Anthropic’s five-year chip lease payments. A smaller deal four months earlier priced at 8.5%. OpenAI has taken the other road. It postponed its own listing beyond 2026 and is seeking at least $30 billion privately, Bloomberg reported.
The model under your vendor’s product is about to become a public company with a balance sheet you can read. Read it.
Why a bank’s risk team should care
Many of the AI features in core, compliance and customer-service platforms run on a handful of foundation models, either directly or through Amazon Bedrock, Google Cloud or Microsoft Foundry. A listing would give third-party risk teams audited financials, risk factors and customer-concentration disclosures for one of those models for the first time. It would also tie the vendor’s pricing and capacity decisions to public-market pressure and to the cost of the debt behind its chips.
- Map which of your vendors’ AI features depend on Anthropic, OpenAI or Google models, including those reached through a cloud platform.
- Read the registration statement when it becomes public. Add its risk factors on compute commitments and customer concentration to your vendor files.
- Ask critical vendors how they would switch models if a provider’s pricing, capacity or terms changed after a listing.
Zovos records which foundation model sits behind each AI feature in your vendor stack. A change in that provider’s ownership or terms then opens a review on every affected vendor.
This is for information only and is not legal or investment advice. The IPO figures are reported and unaudited, so verify them against the linked sources and any registration statement before acting.