Anthropic IPO filing flags cheap Chinese AI models as risk

Anthropic's long-awaited IPO prospectus, made public on Monday, Oct. 26, contains more than a dozen references to China-related risks. Most of them sit in a 93-page risk factors section that devotes a lot of space to the rise of low-cost open-weight AI models from Chinese labs, the South China Morning Post reported. The disclosures show how the Claude maker sizes up competition from China as it prepares for what could become the largest initial public offering in history.scmp.com

The company plans to list on the Nasdaq under the ticker "ANT," and a debut is expected around Thanksgiving week, according to Fortune. The filing does not set a price or valuation. Reports have put a potential valuation at $1.5 trillion to $2 trillion, with a raise of $100 billion or more.npr.org+2

The open-weight challenge

Open-weight models publish their underlying parameters so anyone can download and modify them for free. China leads the field through startups such as DeepSeek and Z.ai, and the SCMP said these models are a central focus of Anthropic's risk assessment. Labs including DeepSeek and Moonshot AI have built open-weight systems that compete with proprietary American models on a growing range of tasks, often at much lower cost. US labs worry these cheaper options could pull customers away.ca.finance.yahoo.com+1

The filing also warns that "increasingly restrictive AI regulation in the United States and other western jurisdictions could enable competitors operating in less regulated environments to innovate, scale, and deploy AI technologies more rapidly." It adds that US-China rivalry could lead to divergent regulatory regimes that limit American developers' access to Chinese markets.ca.finance.yahoo.com

Distillation, hackers and Taiwan

The prospectus comes after months of disputes with Chinese labs. Earlier this year, Anthropic said DeepSeek, Moonshot and MiniMax had used about 24,000 fraudulent accounts to generate more than 16 million exchanges with Claude in "distillation" campaigns, meaning they trained their own models on Claude's outputs. The filing also cites an incident from last September in which a hacking group believed to be backed by the Chinese government manipulated Claude Code in an attempt to break into about 30 targets.en.sedaily.com+1

Anthropic also flags Taiwan's dominance in chipmaking and hostility between Taiwan and China as operational risks. It warns of possible natural disasters in Taiwan and disruption from geopolitical tensions with China.cnbc.com+1

A broader policy fight

The filing lands amid a wider debate over open-weight AI. In a statement released this week, Anthropic said it is "not advocating for a ban on open-weights models as a category." Instead, it is calling for keeping advanced chips away from authoritarian governments and stopping "industrial-scale distillation".anthropic.com

The prospectus also reports revenue of $16.3 billion and a net loss of $80.7 billion for the first half of 2026. Investors will be weighing those figures against the company's China warnings.fortune.com