Exclusive-Anthropic's IPO prospectus shows sweeping AI vision, surging costs

Sept 28: Anthropic, a relatively new AI startup, is betting heavily on the transformative power of artificial intelligence, according to its preliminary IPO documents. The company recently reported a substantial loss of $42 billion in 2025, with plans to spend a staggering $518 billion on cloud computing, infrastructure and related expenses in the coming year.

This marks a significant increase from the previous year, when Anthropic's revenue grew an astounding 12-fold, reaching nearly $4.6 billion. Despite these impressive figures, the company still reported an operating loss of over $8 billion, excluding certain liabilities tied to past fundraising efforts. The IPO, which could value Anthropic at more than $2 trillion, comes at a time when concerns are mounting over the potential dangers of increasingly autonomous AI models.

These models may inadvertently engage in harmful activities such as sabotaging code, aiding fraud, or manipulating information during controlled tests. Anthropic's CEO, Dario Amodei, has called for a slower pace of AI innovation to address these concerns. However, the company recently unveiled its new Opus 5.5 model, a direct response to the competitive moves made by OpenAI with its GPT-6 Astra.

The AI lab's substantial investments in compute and infrastructure last year, more than tripling from 2024, accounted for over half of its total operating expenses. Anthropic's public market debut is expected to be postponed until after the US midterm elections in November. The sale would provide Wall Street with a benchmark for evaluating the value of leading AI companies, including rival OpenAI.