Anthropic’s confidentially filed IPO prospectus lays out the enormous capital demands behind its push to become a dominant frontier artificial-intelligence company. The Claude developer expects $518 billion in future obligations for cloud services, computing and infrastructure, while its 2025 revenue reached roughly $4.6 billion, about 12 times the prior year’s figure. Compute and infrastructure spending rose to $7.33 billion, exceeding annual revenue and accounting for more than half of total operating expenses of $12.65 billion. Anthropic reported a net loss of nearly $42 billion, including almost $34 billion from an accounting charge tied to the projected value of financing instruments that could convert into shares. Its operating loss widened to $8.06 billion from $2.98 billion in 2024. The prospectus comes as Anthropic’s annualized revenue run rate climbed to $30 billion in April 2026 from nearly $9 billion at the end of 2025, with more than 1,000 commercial customers spending at least $1 million annually. To support that expansion, Anthropic has secured or pursued capacity from Google, Broadcom, Amazon, Microsoft, SpaceX and other infrastructure partners, including about 5 gigawatts of next-generation Google TPU capacity from 2027, access to more than 500,000 Amazon Trainium2 chips and a SpaceX deal covering more than 300 megawatts and more than 220,000 Nvidia GPUs. A May fundraising round valued Anthropic at $965 billion, making a public valuation above $2 trillion more than double that figure. The filing also reiterates that more capable models require stronger safeguards under Anthropic’s Responsible Scaling Policy. Investors are weighing whether extraordinary revenue growth can justify the company’s operating losses, hardware commitments, customer concentration and other risks.