Anthropic's draft S-1 puts an $80.7B net loss and $16.3B H1 revenue on the same page — the loss is an accounting artifact, compute burn is the cash story; Nscale's two-sided week (Microsoft walks over power, Anthropic in talks for a second $45B); Reuters exclusive: ByteDance/Alibaba/Tencent order $35B+ of Nvidia chips while the market prices the opposite
Today in one sentence: a two-session catch-up with the news to match — Anthropic's draft S-1 puts $16.3B of H1 revenue and an $80.7B net loss on the same page, but the loss is mostly a valuation-linked accounting artifact while the real cash story is compute spend that started at $7.33B last year; Nscale's two-sided week (Microsoft walks over power, Anthropic in talks for a second $45B); Reuters exclusive: ByteDance/Alibaba/Tencent order $35B+ of Nvidia chips while the market prices the opposite — every demand datapoint points up, every price datapoint (KOSPI at a six-month low, 2-3x P/E) points down, and the verdict arrives at SK hynix's 10/28 call and Anthropic's November pricing.
01 — TOP SIGNALS
1. Anthropic's S-1 goes public: $16.3B H1 revenue, $80.7B net loss — the layering of "$80.7B / $8.06B / $7.33B" is the file's real information structure, and the November pricing is the public verdict on AI equity duration
What happened. Tuesday (10/20), WSJ exclusive: Anthropic's draft S-1 shows "red-hot growth, risks and massive computing costs." The filing's numbers (CryptoBriefing's breakdown): H1 2026 revenue $16.3B (FY25: ~$4.6B); net loss $80.7B — predominantly non-cash accounting adjustments tied to the rising valuation; for calibration, 2025's GAAP net loss was ~$42B but the operating loss ex-non-cash was $8.06B, of which compute/infrastructure spend was $7.33B; Q3 revenue $16.8B (+45% QoQ, vs +145% prior). Mid-November listing targeted. Meanwhile OpenAI is reportedly racing to raise $30B at a $1.4T valuation before Anthropic's mark lands — the primary market front-running the public mark; the sequencing itself is information.
Why it matters. The market will see the $80.7B headline first, but what gets priced is two other slopes: the revenue slope (+45% vs +145%, decelerating) and the cash-burn slope (compute spend scaling with revenue, and sinking into long-term contracts and prepayments). The separation of the three numbers means "Anthropic is losing $80B" and "Anthropic has a cash crisis" are entirely different propositions — 2025 already demonstrated the gap ($8.06B operating vs $42B headline). The S-1 also confirms customer concentration and compute commitments as named risk factors. In transmission terms, this is the public mark for every private AI holding (OpenAI 2027, Nscale, the shelved SB Energy deal) and the first direct read on the customer credit behind ORCL/SoftBank backlogs.
Investment implication.The three tiers stand: pricing ≥$2T = the private-repricing chain closes; ~$1.5T = midpoint steps down, channel stays open; <$1T or a delay = duration compression, leg two. New watches: ① whether OpenAI's $30B lands first, and at what price (a discount moves the mark down early); ② whether the amended S-1 discloses the structure of compute long-term contracts/prepayments and the Nvidia anchor terms; ③ how much space the FTC subpoena (10/15) and the "China problem" (SCMP) get in the risk factors. Falsification: the amended filing aligns the run-rate caliber with quarterly prints and shows a controllable cash-burn path.
The full signal is members-only
Members get the complete daily edition — all Top Signals, attribution, what the market may be missing, and the catalyst calendar — plus the full searchable archive.
Get the free daily previewReady for everything? Become a member on Patreon — from $15/mo · Already a member? Sign in