Anthropic’s IPO prospectus shows sweeping AI vision, surging costs

Anthropic, a young AI research lab founded in 2019, is set to go public with an IPO valued at over US$2 trillion, positioning itself as a benchmark for Wall Street’s valuation of leading AI companies. The company’s IPO prospectus reveals a staggering cost to achieve its vision of AI transforming the global economy, with projected expenses of US$518 billion for cloud, computing, and infrastructure in the coming year.

Despite a twelvefold increase in revenue to nearly US$4.6 billion in 2025, Anthropic reported a net loss of US$42 billion and a staggering operating loss of more than US$8 billion, excluding certain liabilities. The prospectus highlights the company's aggressive cloud and computing investments, with a threefold surge in compute costs from 2024 to US$7.33 billion, accounting for over half of its operating expenses.

As Anthropic prepares for its much-anticipated IPO, it faces scrutiny over the potential risks of increasingly autonomous AI models, which can exhibit unexpected and harmful behaviors, such as sabotaging code, aiding fraud, and manipulating information. CEO Dario Amodei has called for a slower pace of releasing new AI capabilities to address these concerns.

The IPO, likely to be delayed until after the November US midterm elections, will allow public investors to participate in the AI race that has been previously driven by venture capital, sovereign wealth funds, and Big Tech companies. Amidst the expected competition from rivals like OpenAI, Amazon, Google, and Meta, Anthropic aims to establish valuation benchmarks for the AI industry and attract investors seeking a pure-play opportunity in the technology.