Anthropic's IPO: A close look at AI frontrunner's rapid growth, spiraling costs

Anthropic, the AI startup founded in 2021 by former OpenAI researchers Dario and Daniela Amodei, has filed for a US initial public offering (IPO) to go public and face investors' willingness to fund its expanding expenses linked to its AI ambitions. The company, which has seen its revenue surge to $4.59 billion in 2025 from $386 million in 2023, has committed over $500 billion in spending with six partners over a decade, making it one of the largest AI buildout commitments on record.

Anthropic's AI models, including the Claude family, power products such as its chatbot, coding agent, and the enterprise-focused Claude Cowork AI agent. The company's primary revenue stream comes from businesses paying for model usage, measured in tokens. However, this growth has been accompanied by mounting losses, with an operating loss of $693 million for the first half of 2026.

A significant portion of Anthropic's expenses is due to computing power required to run its models. The company has committed substantial computing fees to cloud service providers like Amazon Web Services, Google Cloud, and Microsoft. These commitments are often non-cancelable or require payments regardless of usage, creating a circular financing structure among AI industry players. Amazon, Google, and Microsoft are among Anthropic's top shareholders, indicating the intertwined nature of financing within the AI sector.

Anthropic operates as a public benefit corporation (PBC), balancing profitability with the development of AI for human benefit. The company's board is composed of independent members who elect a majority. Dario Amodei serves as CEO, alongside his sister Daniela, who is the president and chair of the board. This structure aims to limit investor oversight and influence, allowing the founders and an independent trust to retain control over corporate decisions.