WSJ: Anthropic's updated draft IPO prospectus shows $80.7B half-year loss, $600B+ compute commitments, 'catastrophic incident' risk warnings
What happened
The Wall Street Journal reviewed a new draft of Anthropic's confidential IPO prospectus covering H1 2026 and reported Tuesday: revenue of ~$16.3B, up nearly 15x from $1.1B a year earlier; but a staggering $80.7B net loss for the half, up from $13B in H1 2025 and nearly double the full-year 2025 deficit of $42B. About $72B of the loss is non-cash accounting charges on convertible notes and other financing instruments; operating losses remain roughly $8B+. Compute/infrastructure commitments exceed $600B (up from $518B in the September draft), with deals under negotiation that could add ~$150B more, pushing total potential obligations to $750B. Two customers account for ~20% of H1 revenue. The draft also lists sprawling risk factors including uncertainty over preventing AI models from causing harm and protecting against cyberattacks and even 'blackmail by uncontrolled models.' Anthropic is targeting a blockbuster IPO that could value it at up to $2T. Anthropic declined to comment.
Why it matters & what’s next are part of Pro, along with the audio summary.
Start your free 14-day trialWho’s affected
- AI infrastructure complex: Nvidia, Broadcom, Microsoft, Amazon (AWS), Google (cloud) — all named as infrastructure counterparties; debt-funded AI trades like CoreWeave; semis broadly. Losses center on a private company, so impact flows through AI-capex sentiment and supplier counterparty risk.