EY finds 4 in 5 firms alarmed by AI token costs

Enterprises are confronting a new financial reality as AI systems consume tokens at rates that have blown past forecasts, forcing companies to rethink deployment strategies and adopt cost-optimization measures even as per-token prices continue to fall.

An EY report released this week, surveying 500 U.S.-based senior leaders in SVP roles and above, found that more than four in five businesses investing in AI have internal concerns over token usage and related implementation costs. The findings arrive amid growing evidence that agentic AI systems — which reason, plan, and call tools in multi-step loops — are driving token consumption far beyond what organizations budgeted for.ciodive

The Hidden Token Tax

The disconnect between falling per-token prices and rising total AI bills has emerged as one of the defining economic challenges of enterprise AI. Goldman Sachs The Goldman Sachs Group, Inc. ↗ 4.50% Research estimates a 24-fold increase in token consumption by 2030, reaching approximately 120 quadrillion tokens per month as agentic technology scales.techradar+1

The pattern has already produced high-profile budget blowouts. Uber ↘ 1.17% reportedly exhausted its entire 2026 AI budget by April — just four months into the year — after rapid adoption of AI coding tools across its engineering organization, according to Forbes. Amazon Amazon.com, Inc. ↗ 3.90% shut down an internal AI-usage leaderboard called "KiroRank" after concerns that it encouraged tokenmaxxing, with executives urging employees not to use AI merely to climb usage rankings.forbes+2

Peter van der Putten, director of the AI Lab at Pegasystems, described the dynamic in a TechRadar analysis published Thursday: "The visible output you receive is only a small part of what is happening behind the scenes," noting that agents interpret requests, plan, call tools, and loop repeatedly before producing a single answer.techradar

From Experimentation to Discipline

Despite the cost pressures, most enterprises are not pulling back. The EY report found 37% of businesses are expanding the scope of their planned AI implementations, compared to just 15% shifting to narrower deployments. "Companies are showing signs of reckoning with setting priorities rather than merely driving adoption," said EY Global AI Consulting Leader Dan Diasio.ciodive

Organizations are turning to tactical cost-reduction strategies. Data from AICC's 2026 AI API Infrastructure Report, drawn from 2.4 billion API calls processed between January and April, shows enterprises implementing multi-model routing, prompt caching, and open-source model integration have achieved cost reductions of 30 to 80 percent. Open-source models captured 38% of enterprise token volume in the first quarter of 2026, up from 11% a year earlier.openpr

Architecting for Efficiency

Industry voices are converging on a distinction between "design-time" and "run-time" use of AI agents. Van der Putten argued that organizations should use expensive reasoning models once to design workflows, then deploy lighter-weight systems to execute them repeatedly — "like paying a five-star gourmet chef to invent a recipe once, then using skillful but much cheaper chefs to prepare the meal".techradar

The FinOps discipline, initially targeting cloud costs, has expanded to encompass AI spending, and the Tokenomics Foundation — an offshoot of the Linux Foundation unveiled last month — aims to help businesses bring enterprisewide AI costs under control. "The era of 'AI at any cost' is over," one enterprise AI strategist wrote this week. "Success now belongs to the organizations that treat observability not as a technical requirement, but as a strategic imperative".electronicsmedia+1