According to reports, Anthropic is on the verge of going public – and the IPO prospectus filed for the offering provides a rare glimpse into the AI company's finances. It shows that Anthropic recorded a net loss of roughly $42 billion for 2025 amid sharply rising costs, while at the same time planning to spend $518 billion on AI infrastructure. Notably, the prospectus also includes a warning from the company itself that its AI systems could pose catastrophic or existential risks.
- Net loss of roughly $42 billion for 2025, according to the IPO prospectus
- Planned investments of $518 billion in AI infrastructure
- Prospectus explicitly warns of catastrophic to existential risks from its own AI
- New Claude Sonnet 5.5 model scores 70.6 percent on the Terminal-Bench 4.0 test at an unchanged $2/$10 price
- Prediction markets and crypto bettors are debating whether the IPO will still happen in 2026
How large is Anthropic's loss according to the IPO prospectus?
According to reports on the IPO prospectus, Anthropic is reporting a net loss of about $42 billion for 2025, coupled with sharply rising costs. The prospectus lists billions in losses and rising computing costs as key warning signs for potential investors.
Why does Anthropic warn about its own AI in its own IPO prospectus?
Alongside growth figures, the prospectus includes a warning that Anthropic's AI models could pose catastrophic or even existential risks. According to reports, CEO Dario Amodei himself has expressed concerns, even as the company simultaneously tries to convince investors of its growth opportunities – a balancing act between sales pitch and risk disclosure.
How much does Anthropic plan to invest in AI infrastructure?
According to reports, Anthropic plans to invest $518 billion in AI infrastructure. Reactions in pre-IPO perpetual markets have so far been muted.
What's new about Claude Sonnet 5.5?
Anthropic has released a new model, Claude Sonnet 5.5, which scores 70.6 percent on the Terminal-Bench 4.0 test – at an unchanged $2/$10 price compared to the previous model.
The market is already treating the upcoming IPO as a potentially major event for the entire AI stock sector – comparisons to SpaceX are reportedly making the rounds. At the same time, users on prediction markets are betting on whether the IPO will actually happen in 2026, suggesting there is still uncertainty about the exact timing.
As AI models go rogue, do you still trust OpenAI and Anthropic to stop them? I don't and neither should you.
The Guardian
The Anthropic case highlights a fundamental tension within the AI industry: companies are seeking capital for gigantic infrastructure projects while simultaneously warning about the risks of their own technology. Whether investors will accept this risk given the billions in losses and spending plans should become clear once the actual IPO takes place.
