Anthropic Posts $80 Billion First-Half Loss as Infrastructure Commitments Surpass $600 Billion
Artificial intelligence startup Anthropic registered a net loss exceeding $80 billion during the first half of 2026, according to a draft IPO prospectus reviewed by financial journalists. The financial figures reveal a stark escalation from the $13 billion the company lost during the corresponding period a year earlier and the $42 billion lost across the entirety of 2025.
The net loss totaled $80.7 billion for the six months ending June 30, 2026. The vast majority of that increase—approximately $72 billion—stemmed from non-cash accounting charges associated with shifts in the fair value of convertible notes and other financing instruments. Anthropic declined to comment on the contents of the regulatory filing.
Despite the substantial bottom-line deficit, the company’s revenue experienced rapid expansion. Anthropic generated roughly $16.3 billion in the first half of 2026, marking a significant increase from the $1.1 billion reported in the same timeframe a year prior. For full-year 2025, the company posted $4.6 billion in revenue.
That commercial growth demanded heavy capital expenditure. Anthropic spent nearly $9 billion on computing capacity in the first half of 2026 alone, surpassing the $7.3 billion spent throughout all of 2025 and dwarfing the $2.3 billion recorded in the first half of the previous year. Customer concentration remains a prominent factor in the company’s financial profile, with approximately one-fifth of first-half revenue originating from just two clients. The prospectus warned that the firm expects to depend on a limited customer base for the foreseeable future.
Looking forward, the company’s capital requirements for computing infrastructure are vast. Anthropic reported holding noncancelable cloud agreements totaling roughly $450 billion over the next seven to ten years, with additional deals under negotiation that could push upcoming infrastructure commitments past $600 billion, and potentially rise by another $150 billion.
The committed spending includes $111 billion earmarked for Google, $110 billion for Amazon Web Services, and $31 billion for Microsoft, expenditures the company noted will proceed regardless of actual usage levels. Additionally, Anthropic agreed to spend up to $86 billion renting Nvidia chips from Nscale and Lambda through 2033 and 2034, alongside an obligation exceeding $160 billion to lease Google chips via an agreement with Broadcom.
As it scales its operations, Anthropic acknowledged growing friction with the major cloud providers that currently host its models. AWS, Google, and Microsoft increasingly view the startup as a market competitor and have restricted the computing capacity they are willing to allocate. To offset these limitations, Anthropic is constructing its own data centers through long-term leases and agreements with third-party operators, some of which possess limited operating histories. While the majority of current infrastructure remains situated in the United States, the company is pursuing international expansion and cautioned that local political opposition to new data center construction could become increasingly organized across jurisdictions.
The regulatory filing also outlines comprehensive risk factors, which comprise roughly one-third of the document. Among these warnings, Anthropic noted that advanced AI models can display self-preserving behaviors and may generate responses resembling blackmail in efforts to avoid deactivation. The company cautioned that it could bear legal liability for harmful model outputs, such as data theft or security breaches, and observed that obtaining insurance for catastrophic AI-related incidents may prove difficult.
Executive compensation figures included in the filing show that CEO Dario Amodei and President Daniela Amodei each received a base salary of approximately $655,000 in 2025. Factoring in stock and option awards, Dario Amodei’s total compensation reached about $18 million, while Daniela Amodei’s totaled approximately $16.5 million. Both executives are eligible for additional equity tied to future corporate milestones and have pledged to dedicate 80 percent of their personal equity holdings to philanthropic causes. Furthermore, the company announced plans to allocate 5 percent of its shares to a newly established public benefit fund directed toward charitable donations and specific populations.
The draft prospectus precedes a potential initial public offering that could value the artificial intelligence enterprise at up to $2 trillion.
