Finance in the Real World
Why this is in the vault
CJ Gustafson's launch essay for a new report argues that SaaS metrics (ARR, NRR, CAC payback, Rule of 40) became the default lens for "how business works" despite SaaS being ~0.4% of global output, and that a growing wave of capital-intensive companies — chip fabs, data centers, robotics, EVs, defense — need a fundamentally different finance operating model; worth keeping as a direct corroboration of RDCO's own capital-intensive-company investing thesis.
Summary
This is the announcement/teaser email for a forthcoming Mostly Metrics report (gated behind a Typeform signup, not the full report itself). CJ's core thesis: SaaS's clean, teachable metrics colonized how the business world talks about operating a company, but that toolkit was built for one narrow business type (near-zero marginal cost, sells by the month, location-agnostic). A large and growing slice of the economy — chip makers, humanoid-robot builders, data-center operators, EV makers, defense companies — runs on a different operating system entirely, and finance content has mostly ignored it.
The report is built from ten interviews with CFOs at capital-intensive companies. Three get quoted directly in this teaser:
- Scott Haralson, former CFO of Spirit Airlines (now at Hertz): described airline finance as needing to "sell every seat on every plane, over and over" — perishable inventory that reprices daily and is worthless the moment the gate closes.
- Dustin Pederson, CFO of Locus Robotics: framed idle warehouse robots as "inventory is cash sitting on the shelf" — the asset only earns once deployed and running, so every day between manufacture and deployment is stranded capital.
- Nitin Agrawal, CFO of CoreWeave: described his accounting team as "forming precedents, not following them" on AI-infrastructure contract structures that have no existing accounting-standard template.
CJ frames the moment as structural, not cyclical: data-center construction pace compared to the Eisenhower interstate build-out, multi-year/multi-billion-dollar fab timelines, and 600-mile-range EVs as signals that "the physical world, supercharged by AI, is having a moment." The report is explicitly not an anti-SaaS-metrics argument — it's a claim that a second operating system now needs its own toolkit, and that the two are increasingly converging (SaaS companies moving into infra, infra companies adopting recurring-revenue structures).
Mapping against Ray Data Co
Direct corroboration of Ray's own investing thesis (chip-fab/memory capital-cycle, [[2026-05-17-memory-cycle-v1]]): CJ is independently observing, from the CFO-interview seat, the same capital-intensive-company wave (fabs, data centers, EV/robotics manufacturing) that anchors Ray's Phase 2 capital-cycle positioning. This is a finance-practitioner-side data point, not a markets-side one — it corroborates the why now of the thesis (real capital is being committed at unprecedented pace and the accounting/finance function is visibly improvising to keep up, per the CoreWeave quote) without adding market-timing signal. Also useful as a Sanity Check voice/structure reference: CJ builds an essay-format teaser around three tight, concrete practitioner quotes rather than abstract synthesis — a pattern worth studying for how RDCO frames its own report-style content.
Related
- [[2026-07-24-physical-ai-capital-cycle-phase2-analogs]] — the vault's existing Phase 2 capital-cycle research this report corroborates
- [[2026-06-16-mostly-metrics-rivian-capacity-planning]] — same sender, same capital-intensive-manufacturing theme (EV capacity planning)
- [[2026-07-07-mostly-metrics-multi-plan-budgeting-hyperscale]] — same sender, hyperscale/data-center capex budgeting angle
- [[2026-06-28-chip-memory-cycle-phase2-phase3-indicators]] — vault's memory-cycle phase-indicator tracking, same capital-intensive-economy thread