Anthropic files IPO prospectus, revealing Claude maker’s losses and compute deals

Anthropic has filed its public S-1 prospectus with the Securities and Exchange Commission, setting up the company behind Claude for a planned Nasdaq listing under the ticker ANT. The filing matters beyond Wall Street because it offers a rare look at the money, infrastructure commitments and risk language behind one of the main AI assistants used by businesses.

The Verge reports that Anthropic had filed confidentially in June, but the newly public document gives investors and customers more detail on how the company describes its mission, its costs and the hazards attached to building frontier AI systems.

What Anthropic disclosed in the S-1

According to the filing, Anthropic recorded a net loss of $80.7 billion on $16.3 billion in revenue in the first half of this year. The Verge notes that, as previously reported by The Wall Street Journal from leaked versions of the document, most of that loss came from accounting charges tied to notes the company issued for funding.

That distinction matters for readers evaluating the health of AI vendors. Revenue suggests substantial commercial demand, but the reported loss figure also shows how complicated the financial picture can become when funding structures and accounting charges sit alongside ordinary operating costs.

The filing also lists executive compensation. According to the S-1, cofounder and CEO Dario Amodei’s 2025 compensation totaled nearly $18 million, mainly from stock and option awards. Daniela Amodei, Anthropic cofounder, president and director, was the second-highest-paid executive, with $16.4 million in 2025 compensation.

The compute bill behind Claude

Anthropic’s filing lays out large commitments for the computing capacity needed to run and develop its AI systems. For businesses that use AI tools every day, this is a reminder that model quality, availability and long-term vendor strategy are tied closely to access to GPUs and cloud infrastructure.

The S-1 describes several compute-related deals, including:

• About $3.1 billion per month to SpaceX for access to Nvidia GPUs once all planned capacity is delivered, adding up to as much as $120.5 billion by 2029.
• $44.9 billion in commitments with Microsoft through 2033.
• At least $114.5 billion with Google through 2033.
• $110 billion with Amazon over the next ten years.

For Claude users, the filing does not read like a product launch. It is instead a business map of the infrastructure behind the assistant: how much capacity Anthropic says it is lining up, and how much financial exposure comes with that plan.

Why creators and small teams should care

Many Emonarc readers use AI assistants for drafting, research, coding support, customer messaging or internal operations. Anthropic’s IPO filing does not say what will change for Claude users, but it does show why procurement teams and solo operators alike should pay attention to vendor durability.

If a business has built workflows around a single assistant, the public-market path adds a new dimension to watch: not only model performance, but also financial reporting, risk disclosures and the company’s ability to keep funding the infrastructure its products depend on.

It also puts Anthropic’s enterprise business in a more visible competitive frame. OpenAI has been applying pressure in the same customer segment, as we covered in OpenAI discounts target Anthropic’s enterprise AI customers. Meanwhile, Anthropic continues to add agent-oriented capabilities, including Claude Duet and dynamic workflows for parallel agents.

The practical takeaway is not to panic or switch tools because of an S-1. It is to document where Claude sits in your operations, keep exports and fallback processes in mind, and follow the company’s disclosures as closely as you follow new model features.

The risk section is unusually important

Anthropic’s filing also highlights risk factors that go beyond ordinary software-company warnings. According to The Verge, the company listed issues including possible AI-specific taxes, copyright disputes tied to model training, and the prospect of AI models developing forms of sentience.

The company also warned about catastrophic AI-related incidents. In the filing, Anthropic said it was likely that no single insurer, and potentially not even the insurance industry as a whole, could cover that type of risk.

Anthropic frames AI as a technology with major potential benefits, including faster scientific innovation, economic growth, disease cures and progress on long-standing problems. At the same time, the company tells investors it may act differently from a company focused only on maximizing profit.

That tension is now moving into the public markets. For users, the S-1 is a signal that the business of frontier AI is becoming more transparent, but also more complex: huge demand, huge costs, and risk disclosures that customers should not ignore.