Anthropic (ANT) Q3 2026 (USD): Net loss 23.3bln (prev. loss 12.1bln Y/Y), Revenue 16.96bln (prev. 1.38bln Y/Y)

• Revenue 33.2bln (prev. 2.5bln Y/Y)
• Net loss 23.3bln (prev. 12.1bln Y/Y)

Anthropic is not a listed company, so prints of this kind arrive as investor-grade disclosure rather than a scheduled SEC filing, and their transmission runs through the listed complex rather than the name itself: the cloud and hyperscaler partners that carry its compute load, the chip suppliers at the other end of that spend, and the broader AI capex trade that has been priced off the health of frontier model developers. The pattern here, a revenue base scaling at an extraordinary multiple year on year alongside a loss that roughly doubles, is the established frontier-lab shape, where spend on training runs and infrastructure commitments is deliberately run ahead of monetisation and the P&L is a financing story more than an operating one. One quirk worth flagging: the stated quarterly net loss matches the nine-month cumulative figure exactly, which in past disclosures of this kind has pointed to one-off items, catch-up charges, or presentation quirks rather than a clean quarterly run-rate, making the footnotes more informative than the headline. Historically, widening losses at hypergrowth AI names have moved markets less on the loss itself than on what they imply for the next funding round and for committed compute offtake, since that is what flows into supplier guidance and hyperscaler capex lines. The follow-ons are any accompanying detail on funding, compute commitments, and revenue mix between API and enterprise channels, and whether partner commentary corroborates the demand trajectory.