Overview
- Anthropic filed an S‑1 in late October 2026 that discloses minimum multi‑year compute obligations of about $417 billion and optional capacity raising total commitments up to roughly $567 billion.
- The prospectus shows $16.3 billion in revenue for H1 2026 and an $80.7 billion headline net loss that the company attributes largely to non‑cash accounting remeasurements tied to convertible instruments, while reporting an operating profit in Q2 2026.
- SpaceX’s Super Intelligence unit reported an operating profit for Q3 2026 with Anthropic accounting for about 34.5 percent of SpaceX’s revenue, roughly $5 billion, making Anthropic the unit’s largest single customer.
- The SpaceX–Anthropic compute agreement will reach full service capability targeted for December 2026 and includes 90‑day termination rights once non‑cancellable windows end in January and March 2027, creating near‑term concentration and timing risk for providers and investors.
- Anthropic’s planned multi‑class share structure gives founders and a long‑term trust outsized voting control, and the S‑1 devotes extensive disclosure to AI safety and regulatory risks that could shape investor appetite and future commercial deals.