Anthropic files for an initial public offering, aiming to raise 100 billion dollars.
Article excerpt
Highlighted: the sentence this signal was extracted from
Anthropic states that its Claude Opus 5.5 model reduces the cost per customer task by approximately 40% compared to the Opus 5 version. The company also has over 11,900 customers spending at least $100,000 per year, including about 250 customers spending more than $10 million. More than half of Fortune 500 companies use its services. The report nevertheless indicates that Anthropic recorded a loss of 80.7 billion dollars in the first half of the year. The majority of this loss, roughly 72 billion dollars, is attributable to the revaluation of shareholder investment instruments likely to be converted into shares upon the company's IPO, rather than actual cash outflows. The operating loss stands at only 693 million dollars. To meet AI needs, Anthropic has invested over 600 billion dollars in IT infrastructure with numerous partners, including Google, Broadcom, and SpaceXAI. These financial commitments extend through 2030, demonstrating the scale of investments in the AI race. However, the IPO filing also devotes about 94 pages to describing risks, notably the potential for the AI model to develop self-preserving behavior.
