What happened
Observer reports that Chris Bedi, ServiceNow's chief customer officer, told attendees at the 2026 ServiceNow Knowledge event, "I think this [tokenmaxxing] will be a short-lived hype cycle," arguing that organisations face "a bill to pay for those tokens." The article frames "tokenmaxxing" as employees pushing large language model usage and agentic workflows to the limit to increase token consumption.
Technical details
Editorial analysis - technical context: Token usage rises quickly with agentic workflows, continuous code generation, and long-context prompts because tokens are the unit LLMs process. Usage volume is a usage metric, not a direct measure of business value; volume-heavy workflows increase compute, latency, and cost without automatically improving downstream outcomes.
Context and significance
The Observer article notes ServiceNow generated $3.67 billion in subscription revenue in Q1 2026, a 19% year-over-year increase, and reports that roughly 90% of Fortune 500 companies use ServiceNow. ServiceNow also offers A.I. Control Tower, a product for overseeing AI deployments, including tracking agentic behaviour and measuring ROI. Industry context: Enterprise platforms and vendors are increasingly packaging operational controls and cost-visibility features as customers demand outcome-level governance.
What to watch
Watch whether organisations shift procurement and SLOs from usage-based KPIs to outcome-based metrics, adoption of tooling that maps tokens to business value, and how vendors expose cost and ROI signals in deployment dashboards.
Key Points
- 1Executives publicly critique "tokenmaxxing," reframing token counts as a poor proxy for business value.
- 2High token volumes often come from agentic workflows and code generation, raising compute costs without guaranteed ROI.
- 3Enterprises and vendors increasingly emphasise tooling for ROI tracking and governance over raw usage metrics.
Scoring Rationale
This is a notable practitioner-oriented reminder about AI measurement and cost control from a major enterprise vendor executive. It is relevant for teams designing KPIs and governance but does not introduce a new technology or policy shift.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
