Aggressive expansion of data center and computing capacity is at the heart of a new strategy to scale the latest generation of frontier AI models. Anthropic plans to spend roughly $518 billion on cloud and compute infrastructure obligations in the coming year, targeting hardware such as GPUs, TPUs, and Trainium, according to an IPO prospectus seen by Reuters. The filing also outlines risks associated with its autonomous AI systems, specifically the possibility of unexpected behavior and potential misuse, as the firm competes with rivals including OpenAI, Google, and Meta.

Fiscal 2025 revenue grew 1,088% to $4.59 billion, though the company reported a GAAP net loss of $41.97 billion, including a $34 billion non-cash charge tied to the valuation of financing instruments. Operating losses widened to $8.06 billion, with a concentration of revenue where its two largest direct customers each accounted for 12% of total sales. Anthropic may pursue a valuation exceeding $2 trillion when it lists on a public exchange, a move likely to occur after the November US midterm elections.