Anthropic is struggling to secure enough computing capacity from Amazon, Google, and Microsoft. Major cloud providers are increasingly viewing the developer of Claude not only as a customer, but also as a competitor. This was reported by TechCrunch, citing a leaked S-1 document prepared by the company as part of its IPO preparations and obtained by The Wall Street Journal.

The document reveals the scale of Anthropic’s future infrastructure spending. The company expects to spend more than $750 billion on cloud contracts, including agreements that are still under negotiation. Despite these plans, Anthropic fears it still may not be able to secure the computing resources it needs, TechCrunch reports.

Anthropic looks for an alternative to cloud giants

To reduce its dependence on the largest cloud platforms, Anthropic is leasing its own data centers and working with third-party data center operators. However, this approach carries risks: in the documents, the company notes that some operators have limited operating experience. As a result, their reliability may fall short of that offered by major cloud providers.

The situation is further complicated by competition for computing capacity. Amazon, Google, and Microsoft are simultaneously providing infrastructure for AI companies while also developing their own models and services. As a result, Anthropic has to seek resources from companies that compete with it for customers and market position.

Massive spending still does not guarantee profitability

According to the document, in the first half of 2026 Anthropic posted a net loss of $80.7 billion on revenue of $16.3 billion. According to The Wall Street Journal, about 20% of sales came from just two customers, pointing to significant dependence on a limited number of major clients.

At the same time, the company’s business continues to grow rapidly. As Reuters reported on October 14, Anthropic’s third-quarter revenue reached $16.8 billion — 45% higher than the previous quarter. Thus, the figure for three months exceeded revenue for the entire first half of the year, although these numbers cannot be directly compared from a profitability standpoint.

Such large-scale infrastructure commitments reflect the high cost of developing advanced AI models. Companies must not only train models, but also provide the computing resources for their day-to-day operation as the number of users and enterprise customers grows.

The company warns of AI-related risks

The documents also mention Anthropic’s potential financial liability for harm caused by the behavior of its AI systems. According to the company’s estimates, possible losses could exceed the amount of insurance coverage.

This warning adds to previously disclosed information about catastrophic risks associated with AI development. For potential investors, such liabilities are no less important than revenue growth rates: a serious incident could lead to expenses far beyond the available protections.

According to The Wall Street Journal, Anthropic is aiming to hold its IPO in November 2026. The leaked document gives potential investors a more detailed picture of the financial side of the business — including massive infrastructure commitments, dependence on cloud partners, and risks that could affect the company’s future results.

At the time of TechCrunch’s publication, Anthropic had not responded to the outlet’s request for comment.