Anthropic, a leading developer of generative AI models, has filed a registration statement for an initial public offering (IPO) planned for late 2026. While the company's surface-level financial figures show rapid revenue growth and a healthy balance sheet, critics argue that a deeper look at its disclosures reveals a precarious financial position. Specifically, the company has entered into nearly $900 billion in spending commitments stretching into the 2030s. These commitments, primarily with tech giants like Amazon, Google, Microsoft, AMD, and Broadcom, are contractually required payments to secure the massive computational power needed to train and run large language models (LLMs). Because these obligations are non-cancelable, they function as effectively immovable debts that Anthropic must service regardless of its actual profitability.
Several economic forces are currently squeezing the margins of AI model developers, making it difficult for Anthropic to meet these vast obligations. First, there are concerns regarding demand; some studies suggest that LLMs are not yet generating positive returns on investment for most businesses, and one study noted a significant drop in workplace AI usage between 2025 (46%) and 2026 (33%). Second, fierce competition from rivals like OpenAI, Meta, and Google has sparked a price war, forcing companies to lower prices even as costs rise. Third, the physical infrastructure required for AI is extraordinarily expensive. A single 1 gigawatt (GW) data center can cost an estimated $31 billion to construct. Anthropic estimates it will eventually need between 20 and 25 GW of capacity, explaining the half-trillion-plus dollars in spending commitments that appear in its registration statement.
The primary concern for regulators and retail investors is the potential for systemic risk. As a private company, Anthropic's shares are largely held by venture capitalists and corporate insiders. If the company goes public, those shares will flow into 401(k)s, retirement accounts, and index funds, shifting the risk of a potential 'financial precipice' onto the general public. Critics argue that the Securities and Exchange Commission (SEC) has the authority to block the IPO because the registration statement includes only unaudited financial results for its period of highest growth and relegates its most significant future obligations to footnotes. The argument suggests that until Anthropic provides consecutive quarters of audited financial statements and fully discloses its spending commitments, the public market remains exposed to a potentially devastating crash in the AI sector.
However, it remains uncertain exactly how much of the remaining $300 billion in contractual commitments will eventually land on Anthropic's balance sheet, as the company continues to negotiate new infrastructure deals. Additionally, while the report highlights the risks of 'hidden' debt, the actual impact of these commitments depends on Anthropic's ability to innovate and capture market share in a rapidly commoditizing industry where models are becoming increasingly indistinguishable in quality.