Anthropic faces a difficult capacity problem: it needs substantial computing resources, but the cloud providers it relies on increasingly regard the Claude-maker as a competitor. A leaked S-1 filing describes the challenge, along with the company’s spending plans, financial results and concerns about potential liability.
Cloud capacity is becoming harder to secure
According to the filing, Anthropic is having trouble securing enough capacity from AWS, Google and Microsoft. Those companies are not only infrastructure providers; they increasingly see Anthropic as a rival, complicating the company’s efforts to obtain the computing resources it needs.
The filing says Anthropic believes it could spend more $750 billion on cloud contracts. Some of those arrangements are still being negotiated, so the figure reflects planned and potential commitments rather than only finalized deals.
Even that level of cloud spending may not meet the company’s needs. Anthropic is also leasing its own data centers and working with third-party data center operators, widening the set of ways it can access capacity.
Alternative infrastructure brings its own risks
Using outside data center operators may help Anthropic expand beyond the major cloud providers. But the company’s filing raises a concern about those operators: they may have a “limited operating history.” Anthropic says that could make them less reliable than the hyperscalers.
The trade-off is consequential. Depending on a small group of large providers can leave the company exposed when those providers have competing interests. Turning to less-established operators could broaden its options, but Anthropic itself flags uncertainty about their reliability.
The filing therefore presents capacity as more than a question of how much Anthropic is willing to spend. It also concerns where the company can obtain computing resources and how dependable those sources may be.
Large losses sit alongside rising revenue
Anthropic reported an $80.7 billion net loss in the first half of the year, while revenue for that period was $16.3 billion. According to the WSJ report, only two customers accounted for about a fifth of sales.
The company has also told prospective investors that third-quarter revenue alone has reached $16.8 billion, up 45% from the previous quarter. That increase offers a different view of the business from the first-half loss, but the filing’s figures and reported growth do not remove the infrastructure challenge.
Anthropic’s potential cloud commitments are immense in the context of these figures. The filing describes a company pursuing more computing capacity while managing steep losses and a notable share of sales from two customers.
Investor disclosures include AI liability concerns
The filing also warns that Anthropic could be held liable for damages caused by harmful AI behavior. The potential damages, it says, could exceed the company’s available insurance coverage.
This warning adds a separate kind of risk to the picture. Securing computing resources and funding growth are operational and financial challenges; potential liability from AI behavior could expose the company to costs beyond its insurance protection.
Anthropic is reportedly targeting a November IPO. Its disclosures give prospective investors a view of the competing pressures: demand for cloud capacity, reliance on providers and operators with different reliability profiles, large reported losses, and possible liability related to harmful AI behavior.