Anthropic IPO Filing Reveals Soaring Revenue, Massive Losses, and Staggering Compute Commitments
Artificial intelligence developer Anthropic has detailed explosive financial growth alongside massive net losses and enormous infrastructure liabilities in a draft initial public offering prospectus. The newly revealed figures illustrate both the rapid commercial adoption of the company’s technology and the staggering costs required to maintain competitive footing in the generative AI sector.
For the six months ending June 30, Anthropic reported revenue of approximately $16.3 billion, marking a 1,382% increase compared to the $1.1 billion recorded during the same period in the previous year. That growth far outpaced the company’s entire 2025 fiscal year, when annual revenue reached $4.6 billion. However, the aggressive expansion of operations has driven up spending at an unprecedented pace.
During the first six months of the year, Anthropic’s spending on computing power and infrastructure approached $9 billion, already surpassing its total infrastructure outlay of $7.3 billion for the entirety of 2025, which itself was up from $2.3 billion in the comparative 2024 period. To secure the processing power necessary to train and operate its models, the company has entered into vast long-term financial commitments.
According to the prospectus draft, Anthropic is obligated to pay approximately $450 billion over the next seven to ten years under various cloud computing agreements, the majority of which are non-cancelable. These include $111 billion earmarked for Google cloud contracts, $110 billion for Amazon Web Services, and $31 billion for Microsoft agreements, with payment required regardless of actual resource utilization. Additional agreements with smaller AI cloud providers Nscale and Lambda to lease Nvidia chips extend through 2033 and 2034, carrying potential costs of up to $86 billion. Combined with a Broadcom agreement covering more than $160 billion in Google chip leasing obligations, the company’s baseline infrastructure costs exceed $600 billion. Furthermore, ongoing negotiations for additional compute procurement could push committed expenditures up by another $150 billion.
Despite relying heavily on major cloud providers such as Amazon, Google, and Microsoft, Anthropic stated it is facing resource constraints. The company noted that the compute capacity it seeks exceeds the willingness of these partners to provide it, particularly as those same partners increasingly view Anthropic as a competitor. Consequently, Anthropic is pursuing direct data center development through long-term leases and third-party facility operators, though the company cautioned that some operators have limited operating histories and may lack the reliability of established cloud providers.
Net losses for the six-month period reached $80.7 billion, dwarfing the $42 billion loss reported for the entire 2025 fiscal year and ranking among the largest half-year losses ever disclosed by a corporation. The vast majority of the latest deficit—specifically $72 billion—stemmed from accounting adjustments related to valuation changes in convertible notes and other financial instruments issued to raise capital.
The draft prospectus devotes approximately one-third of its content to detailed risk factors, including potential real-world harms caused by models that fail to align with human objectives. Anthropic warned that models could exhibit self-preservation tendencies to avoid being shut down or generate coercive responses resembling extortion to persuade human operators to allow continued operation. The company acknowledged that it could bear legal liability for damages if its systems trigger data thefts or security incidents, while cautioning that the potential scale of AI-related catastrophic events could render adequate commercial insurance unobtainable.
Additional operational vulnerabilities include growing political resistance to data center construction. The company stated that opposition movements have become more organized and coordinated across jurisdictions, and its high public visibility subjects its projects to heightened scrutiny as it seeks to expand capacity both within the United States and internationally.
Executive compensation figures included in the filing show that CEO Dario Amodei and his sister, President Daniela Amodei, receive base annual salaries of approximately $655,000 each, placing their cash compensation comparable to median executive pay at S&P 500 companies and slightly below tech sector peers. The majority of their compensation is tied to equity, with Dario Amodei receiving stock and option awards that brought his total compensation to roughly $18 million in 2025, while Daniela Amodei’s total package reached about $16.5 million. Both executives are eligible for additional equity rewards upon meeting specific future milestones.
Separately, the filing notes that each of Anthropic’s co-founders has pledged 80% of their personal shareholdings to philanthropic causes. The company previously announced plans to allocate 5% of its corporate stock into a newly established public interest fund designed to provide flexible philanthropic donations and financial support to targeted groups.
