As AI developer Anthropic prepares for what is expected to be one of the most closely watched public offerings of the year, newly leaked financial documents reveal deep operational hurdles. According to an S-1 registration filing reviewed by The Wall Street Journal, the maker of the Claude AI assistant is struggling to secure adequate cloud computing infrastructure. Major tech giants that once welcomed the partnership—specifically Amazon Web Services (AWS), Google, and Microsoft—increasingly view Anthropic as a formidable direct competitor rather than merely a high-paying enterprise customer, complicating the company’s efforts to scale its computing power.

The newly leaked filing expands significantly upon financial figures and strategic outlines initially reported by Reuters and the Financial Times in late September. Among the most striking revelations in the prospectus is Anthropic’s projection that its total spending on cloud contracts could ultimately surpass $750 billion. This staggering figure accounts for both active agreements and massive deals that remain under active negotiation as the company races to maintain the computational infrastructure required to train and run its frontier artificial intelligence models.

Even with these massive financial commitments, the supply of available enterprise-grade infrastructure falls short of Anthropic’s internal demands. To bridge the gap, the company has resorted to leasing its own data center spaces and collaborating directly with third-party data center operators. However, these alternative arrangements have introduced new vulnerabilities. According to the filing, Anthropic executives harbor concerns that these third-party operators possess a "limited operating history," potentially rendering them far less reliable than the established cloud hyperscalers during moments of high-demand stress or systemic technical outages.

Astronomical Costs and Revenue Concentration

The aggressive pursuit of computing capacity comes on the heels of staggering financial losses. The S-1 filing reveals that Anthropic recorded an $80.7 billion net loss during the first half of the year alone, highlighting the brutal economic realities of the generative AI race, where training next-generation models requires multi-billion-dollar investments before yielding commercial returns.

Despite these immense losses, revenue generation has surged. Anthropic reported $16.3 billion in revenue during the first half of the year. However, the financial disclosures highlight a notable concentration risk: only two primary customers accounted for approximately one-fifth of the company’s total sales during that period, pointing to a heavy reliance on a narrow base of major enterprise clients.

The financial pressure is compounded by escalating liability warnings. Building on previously disclosed catastrophic risk assessments included in earlier regulatory discussions, the filing explicitly warns investors that Anthropic could face legal liabilities for damages caused by harmful or unintended AI model behavior. The company notes that potential legal judgments and damages could scale to a degree that vastly exceeds its available insurance coverage, posing a unique systemic risk for prospective shareholders.

Growth Momentum and Upcoming Public Debut

Despite the immense capital burn, cloud bottlenecks, and liability warnings, commercial momentum for Claude and Anthropic’s enterprise offerings remains remarkably strong. Just last week, the company informed prospective investors during roadshow presentations that its third-quarter revenue alone had reached $16.8 billion, representing a significant 45% increase compared to the previous quarter.

Capitalizing on this rapid top-line growth, Anthropic is moving swiftly toward the public markets. Following initial market positioning, the company is reportedly targeting a November initial public offering to secure the massive pools of capital necessary to fund its ongoing infrastructure expansion and research initiatives.

Representatives for Anthropic did not immediately respond to requests for comment from TechCrunch regarding the leaked S-1 filing, the status of its cloud contract negotiations with major hyperscalers, or its upcoming public market debut.