Anthropic IPO filing puts explosive AI growth beside a vast compute bill

Anthropic's public S-1 describes a sharply expanding AI business, but also discloses heavy infrastructure commitments, continuing losses and a governance structure that would keep substantial voting power with its founders and long-term-benefit trust.

jajoy39

Oct 27, 2026 - 06:46

0 5

Rooftop deck at Foundry Square III in San Francisco, part of a complex where Anthropic has leased offices (file photo). Source: STUDIOS Architecture.

Anthropic IPO filing puts explosive AI growth beside a vast compute bill

Anthropic's public filing for a proposed initial public offering gives investors a detailed look at a business growing at extraordinary speed while depending on equally extraordinary amounts of computing capacity. Reuters reported that the Claude maker filed its registration statement on October 26. The filing itself remains preliminary: it says the information can change, the offering price is not yet set and the shares cannot be sold until the registration statement becomes effective.

Revenue is scaling faster than the cost problem is disappearing

According to figures reported from the filing by Reuters, Anthropic's revenue rose to $4.59 billion in 2025 from $386 million in 2024 and $41 million in 2023. Revenue for the first six months of 2026 reached $16.29 billion, compared with $1.16 billion in the same period of 2025. The company still recorded a $693 million operating loss in the first half of 2026, although that was narrower than the $2.14 billion loss in the first half of 2025.

Anthropic makes most of its revenue from business customers paying according to model usage, alongside subscription products. That consumption model can expand rapidly as customers use Claude for coding, enterprise work and agentic tasks, but it also ties revenue to continued access to expensive computing resources.

Long-term commitments create exposure to infrastructure demand

Reuters said Anthropic has committed more than $600 billion in spending with about eight partners over a decade, with some agreements described as non-cancelable or payable regardless of usage. The reported commitments include at least $110 billion with Amazon, about $114.5 billion with Google and $44.9 billion with Microsoft for cloud computing over seven to 10 years. Amazon and Google are also major Anthropic investors, creating the circular financing pattern the Reuters report highlighted: suppliers of compute are simultaneously financing a major customer.

These commitments are not the same as immediate cash outlays, and the filing's preliminary status matters. They nevertheless show why model competition is also a balance-sheet and energy-infrastructure contest. If demand, pricing or model economics change, a company can be left paying for capacity that no longer matches usage.

Control will not be distributed like a conventional public company

The SEC filing says Anthropic will remain a Delaware public benefit corporation. It describes a Long-Term Benefit Trust intended to help keep the company focused on long-term AI safety and responsibility, with authority over a majority of certain board seats. The filing also says a Founder LLC would hold a single Class F share carrying 50.1% of total voting power on general shareholder matters. That structure is disclosed, not an allegation of wrongdoing, but it means public shareholders would have limited influence over key governance decisions.

Anthropic's filing therefore frames the IPO as a test of more than AI enthusiasm. Investors will be assessing whether fast enterprise adoption can support a business with unusually large compute commitments, while accepting a governance model designed to preserve founder and mission influence.