Anthropic sets $1.3 trillion IPO price amid $900 billion compute deals

The artificial intelligence sector is moving toward a new phase of capital concentration. Anthropic, the developer of the Claude language model, submitted a pricing proposal on November 10 for an initial public offering that values the company between $1.3 trillion and $1.4 trillion. This figure falls short of the $2 trillion valuation some analysts had projected earlier in the year.

The company’s securities filings detail the scale of its financial obligations. Anthropic has entered into purchase commitments and lease agreements totaling approximately $900.9 billion. These contracts primarily cover computing infrastructure. About $70.7 billion of this total is due within 12 months of September 30, 2026. A further $451.6 billion is expected to be incurred before the end of 2030. The remaining obligations extend through 2045.

Revenue data included in the disclosures shows an annualized run rate of approximately $81 billion as of October. This figure exceeds the $20 billion annual revenue estimates frequently cited in public discussions. However, the filings distinguish this revenue from profit and free cash flow. The company must cover costs for computing, research, personnel, electricity, and facilities to service its customer base.

Hardware economics remain a central issue in the industry. The source notes that a single high-end GPU can cost approximately $20,000 at retail. Under illustrative scenarios, recovering the cost of that hardware from monthly subscription fees can take 50 months, even before accounting for electricity, cooling, and maintenance. Nvidia dominates the supply of these accelerators and the CUDA software ecosystem required to run them.

In September 2026, Nvidia announced an agreement to acquire Hugging Face for approximately $12.93 billion. The deal includes an equity-based employee retention component and is expected to close in 2027, subject to regulatory approval. Hugging Face serves as a major platform for developers to discover, publish, and distribute open-weight machine learning models.

The acquisition places the primary supplier of AI hardware in a position of control over a key distribution channel for competing models. Nvidia stated that Hugging Face will remain open and that developers will retain the freedom to choose competing hardware. The transaction remains pending final regulatory sign-off.

Competing open-weight models from companies such as DeepSeek, Alibaba, and Moonshot AI continue to challenge proprietary subscription models. These models can be downloaded and operated on private infrastructure, offering an alternative to the high-cost cloud services that drive Anthropic’s revenue.