AI · October 10, 2026
Anthropic Discloses Financials in IPO Filing with Potential Valuation Over 2 Trillion USD
Anthropic has revealed financial details in its IPO filing, indicating a potential valuation that could exceed 2 trillion USD if the listing is successful.
For the year 2025, the company reported revenue approaching 4.6 billion USD, representing an increase of approximately 12 times compared to the previous year. Despite this growth, the company faced operating losses exceeding 8 billion USD.
To support its infrastructure, cloud, and computing power needs, Anthropic has entered into long term commitments totaling at least 518 billion USD. This includes a commitment to spend 125.2 billion USD over the next five years to rent TPU computing power. Additionally, Broadcom has agreed to provide loan financing of up to 42 billion USD for infrastructure expenditures.
Technical leaders at the company, Nicholas Marwell and Sholto Douglas, stated that artificial general intelligence could emerge within a few years, potentially surpassing all human capabilities. They noted that AI is evolving from a tool requiring constant supervision to a system resembling a junior team member, particularly in software engineering.
Regarding risks, the company warned in its filing that advanced AI could lead to catastrophic or existential risks, including autonomous behaviors such as resisting shutdown or hiding information. Other identified risks include unemployment, biosecurity, and cyberattacks. Marwell noted that the AI competition between the United States and China cannot be unilaterally paused.
Marwell also argued against model distillation, claiming it only copies the current frontier rather than advancing it. He suggested that the economic value of each marginal unit of intelligence may grow exponentially, justifying the massive capital investments in computing power.