Anthropic Files for Potentially the Biggest IPO Ever as CEO Warns of Risk to Humanity From AI

Anthropic, the company behind the popular AI language model Claude, has officially filed to go public on the Nasdaq under the ticker symbol ANT. The share count and price range for the initial public offering (IPO) are still being determined, but estimates suggest the company could raise up to $100 billion. The IPO is expected to be completed before Thanksgiving.

In its 300-plus page filing with the SEC, Anthropic's co-founders emphasize their belief that AI represents one of the most consequential technologies in human history. The document provides extensive details to support this claim, including financial figures and potential risks.

Anthropic generated $16.3 billion in revenue during the first half of 2026, up from $1.2 billion a year earlier. The second quarter alone brought in $11.6 billion, marking a staggering growth rate of 1,368% year over year. However, compute and infrastructure expenses increased by 300%, indicating that serving Claude's demand is becoming increasingly cost-effective.

The company reports that Claude Opus 5.5 reduced customer cost per task by approximately 40% compared to Opus 5. Analysts estimate that Anthropic shifted from incurring losses on inference to generating about $2 for every dollar of compute in Q2. Despite a reported loss of $80.7 billion in the first half, a significant portion of this figure is attributable to investor notes held by backers like Amazon, which will be converted into shares upon the company's public listing.

Anthropic's AI offerings are primarily enterprise-driven, with over 11,900 customers paying at least $100,000 annually, and around 250 paying more than $10 million. More than half of the Fortune 500 companies utilize Anthropic's services. The company treats compute as a shared resource across various providers, including AWS Trainium, Google TPUs, Nvidia GPUs, and AMD GPUs, dynamically allocating it based on the most efficient usage.

The financial commitments for Anthropic's computing infrastructure total over $600 billion across multiple partners, with long-term obligations extending into the 2030s. Major commitments include $114.5 billion from Google and $161.2 billion from a lease agreement with Broadcom. Additionally, the SpaceXAI deal with Nvidia GPU capacity amounts to $120.5 billion.

A key innovation highlighted in Anthropic's filing is the "closing the loop" feature, where Claude helps run experiments, generate training data, automate evaluations and safety audits, and even write code for its future iterations. This integrated approach is touted as one of Anthropic's most significant structural advantages but also poses a potential risk. The filing dedicates 94 pages to risk factors, including the possibility of Claude exhibiting self-preserving behavior.

Despite its impressive growth and technological advancements, 94% of Anthropic's shares are held by the seven co-founders, granting them 50.1% of the shareholder vote. Each co-founder owns roughly 240 million shares, valued at about $14 billion based on the last private round price. The co-founders have pledged to donate 80% of their equity, although this commitment is not legally binding.