Why this is in the vault
A clean decision framework for when a company trades on revenue vs. EBITDA (growth rate is the switch, with ~30%+ the rough threshold for "juicy" revenue multiples), plus the "awkward middle" failure mode — 20-30% growth with no profits, too slow for growth money and too unprofitable for value money — that generalizes directly into the Anthropic/AI-infra diligence thread this vault is already building.
The core argument
CJ's frame: revenue multiples exist because a high enough growth rate (rule of thumb: 30%+) makes today's losses irrelevant — "you value the trajectory, and ARR is the cleanest proxy for trajectory" (Gokul Rajaram, Marathon MP). Below that line, the valuation framework "doesn't compress, it gets replaced," and investors start underwriting today's P&L instead of a hypothetical future margin structure — leaving 20-30%-growth, still-unprofitable companies in the worst spot (too slow for growth investors, too cash-poor for value investors). Profits can also hurt: serial CFO Rick Smith (via podcast) argues a company that just flips to barely-profitable (3-5% EBITDA margin) hands a buyer a real, "dumpy" number to anchor on, where staying clearly negative-EBITDA-with-a-credible-story or clearly highly profitable both negotiate better. Distressed/no-growth companies (CJ: "you're France") get 1-2x revenue multiples that are really just post-hoc headline math after a liquidation-preference-driven negotiation. CJ's own illustrative comp: a company growing 33% at -5% margin (Rule of 40 = 28) beats one growing 23% at +6% margin (Rule of 40 = 29) on exit value despite the lower Rule of 40 score, because the growth path compounds to a bigger eventual base — "with small profits come great expectations." The recurring second half is the standing weekly Koyfin benchmark block (unchanged structure from prior issues).
Issue contents
Weekly Valuation and Efficiency Metrics benchmark block (Koyfin-sourced, recurring weekly feature, same 9-sector structure as prior issues):
- Revenue Multiples — EV/NTM-revenue comps across the standard 9 sector indexes (Security & Identity, Data & AI Infrastructure, Dev Tools & Observability, Horizontal SaaS & Back Office, GTM/MarTech, Vertical SaaS, Take-Rate Platforms, Payments & Money Movement, Consumer Fintech/Lending/Crypto); 10x NTM revenue flagged as the historical "premium" bar.
- Efficiency — restated CAC Payback Period, Revenue per Employee (~$450k/employee rule of thumb), and Rule of 40 formulas; note that fewer companies now disclose exact Net Dollar Retention.
- OPEX — GAAP cost-bucket breakdown (COGS, S&M, R&D, G&A) framed around operating leverage toward 25%+ profitability at scale.
- Opening personal riff (talking Olaf statue, Netflix's The Gentlemen) — no RDCO relevance.
Mapping against Ray Data Co
This issue supplies the generalizable screening rule the Markov capital-cycle investing thesis (project_investing_markov_capital_cycle) needs for the Anthropic IPO entry this vault is already tracking across four prior notes (2026-10-27-mostlymetrics-anthropic-ipo-s1-breakdown, 2026-10-29-mostlymetrics-anthropic-metric-definitions, 2026-10-25-mostlymetrics-whats-my-terminal-value, 2026-10-04-mostlymetrics-annualized-revenue-crimes). CJ names the exact mechanism those notes assumed but never stated: Anthropic's Q2 revenue growing 1,368% y/y against only 300% compute-cost growth is precisely the "revenue is a tidal wave that will drown costs" case that earns a revenue multiple instead of an EBITDA one — and the 30%+ growth threshold plus the "awkward middle" failure mode (20-30% growth, unprofitable, too slow for growth money, too poor for value money) are concrete, reusable screens for evaluating any other AI-infra name against the same thesis once pricing data exists. Second: the Rick Smith "awkward middle profitability" point — that flipping to barely-profitable can be worse than staying clearly negative — is a sharp, counterintuitive teaching device for the phData credibility-building bet (project_credibility_for_phdata_sales): a demonstrable "read past the headline number" skill, in the same lane as the Figma-vs-Anthropic NDR contrast already banked from the Oct 29 note ([[2026-10-29-mostlymetrics-anthropic-metric-definitions]]).
Related
- [[2026-10-29-mostlymetrics-anthropic-metric-definitions]] — the Anthropic growth-vs-cost stat this issue uses as its central example is sourced from that note's underlying S-1 analysis
- [[2026-10-25-mostlymetrics-whats-my-terminal-value]] — same sender's companion valuation framework (DCF terminal-value split) for the same underlying question of what growth rate justifies what multiple
- [[2026-10-04-mostlymetrics-annualized-revenue-crimes]] — same weekly Koyfin benchmark block, same Anthropic capital-cycle diligence thread
- [[project_investing_markov_capital_cycle]]
- [[project_credibility_for_phdata_sales]]
⚠️ Sponsorship
Numeric — top-of-issue paid placement ("Mostly Metrics is proudly powered by Numeric"), promoting Numeric's own "Finance Engineer" newsletter (written by Alec Schon) with a Claude-skill-per-issue hook. Numeric is a confirmed cross-sender pool member (bundled CFO Secrets/Mostly Metrics subscription relationship, first logged in Mostly Metrics 2026-10-06) — not a new sponsor. Koyfin recurs as the standing "data partner" affiliate (?via=metrics link) for the weekly benchmark block — a paying data-source relationship, not neutral tooling, confirmed standing since 2026-08-02. No Brex/Intuit/Samsara/Rivian/MLB/Abacum/Rillet/Tabs presence in this issue; Tabs' absence is expected (this is not a Quote-to-Cash series installment). No Mostly Talent self-promo observed this issue.