Anthropic's Confidential IPO Filing Reveals $80.7 Billion H1 Loss and Heavy Computing Costs

Artificial intelligence company Anthropic reported a net loss of $80.7 billion in the first half of 2026, according to a draft prospectus reviewed by The Wall Street Journal. The disclosures come as the company prepares for an initial public offering that could potentially value the firm at $2 trillion.

During the six months ended June 30, Anthropic generated $16.3 billion in revenue. This compares with $1.1 billion in the corresponding period of the previous year. For the full year 2025, the company had reported $4.6 billion in revenue.

The substantial net loss reported for the first half of 2026 was largely driven by accounting adjustments related to financing agreements. This included $72 billion in charges linked to changes in the value of convertible notes and other financial instruments.

Alongside these financial figures, the draft prospectus detailed massive computing and infrastructure expenditures. Anthropic's spending on computing reached nearly $9 billion in the first half of 2026, surpassing the $7.3 billion spent throughout the entirety of 2025.

Looking ahead, Anthropic indicated that its future infrastructure commitments could exceed $600 billion. The company is considering additional computing deals worth up to $150 billion. The prospectus notes that Anthropic has committed to spending approximately $450 billion on cloud computing agreements over the next seven to 10 years, with the majority of these arrangements designated as non-cancellable.

These long-term commitments include $111 billion allocated to Google Cloud, $110 billion to Amazon Web Services, and $31 billion to Microsoft, which must be paid regardless of actual usage. Additionally, the company agreed to spend up to $86 billion renting Nvidia chips through AI cloud providers Nscale and Lambda, alongside a commitment exceeding $160 billion for Google chips through an arrangement with Broadcom.

Anthropic cautioned that access to adequate computing capacity remains uncertain, noting that its requirements have occasionally exceeded what certain cloud partners are willing to supply. The company also identified growing resistance to data centre construction as an emerging business risk.

The draft filing also highlights operational and market vulnerabilities, including customer concentration. Approximately one-fifth of Anthropic's $16.3 billion in first-half revenue was generated from just two customers.

Furthermore, the prospectus outlines potential risks associated with harmful AI behavior, such as models generating responses that resemble blackmail or displaying self-preserving tendencies to avoid shutdown. The company warned that the financial consequences of a catastrophic AI-related incident could surpass its available insurance coverage.

"Anthropic's confidential IPO filing highlights the immense capital intensity required to scale frontier artificial intelligence models. While top-line revenue growth to $16.3 billion in the first half of 2026 is substantial, the massive non-cancellable cloud commitments and non-cash accounting adjustments underscore the high financial stakes involved. For founders and investors, this case demonstrates that infrastructure planning, customer concentration risks, and long-term liquidity management are critical pillars when scaling deep-tech enterprises toward public markets." — Dr. Shishir Gupta, Founder & CEO, StartupLanes