Anthropic’s original IPO paperwork, filed last week, offered investors financials only through the first half of 2026. Monday’s amended S-1 filled in the missing third quarter—and turned the coming listing into a sharper test of whether AI revenue can keep pace with AI infrastructure costs.¹

The topline is dramatic. Anthropic reported roughly $17 billion in third-quarter revenue, up from $11.6 billion in the prior quarter and more than 12 times the level of a year earlier. Revenue for the first nine months reached $33.2 billion, while the company ended September with $131.8 billion in cash and short-term investments.²

But the filing also showed the price of that expansion. Anthropic disclosed $895.4 billion in purchase commitments and leases, “substantially all” tied to compute and infrastructure. About $447.5 billion is expected by the end of 2030, including nearly $70 billion due in the 12 months after September 30, 2026; the balance stretches through 2045.²

In the third quarter, the company swung from $974 million in operating income to a $567 million operating loss. Anthropic attributed the reversal primarily to “$5.0 billion of compute under short-term commitments,” which it said were materially costlier than comparable multiyear deals. Compute and infrastructure charges rose to $9.9 billion, four times their level a year earlier.³

The broader loss figures look still harsher: a $23.29 billion net loss for the quarter and $104 billion for the first nine months, though much of the latter reflected accounting changes tied to convertible notes and derivative liabilities.⁴

There is another measure of the strategy: talent. Personnel expenses reached $1.7 billion for 5,300 workers in the third quarter, including $1.1 billion in stock-based compensation. Yet personnel costs were only 15% of nine-month revenue—a sign that, for Anthropic, compute rather than headcount has become the defining cost of growth. The company’s own formulation is blunt: “Talent density matters more than talent mass.”⁵