Nearly Half of 2,145 Firms Cut AI Agent Use as Usage-Based Pricing Lifts Bills
Updated
Updated · Financial Times · Jun 30
Nearly Half of 2,145 Firms Cut AI Agent Use as Usage-Based Pricing Lifts Bills
3 articles · Updated · Financial Times · Jun 30
Summary
Nearly half of 2,145 global business leaders surveyed by KPMG in May said they had scaled back AI agents because costs now outweigh the benefits.
Since early 2026, major AI providers have shifted companies from flat fees to usage-based pricing, while more capable AI agents consume far more credits than basic chatbots.
Uber said it had exhausted its 2026 AI budget by April and now limits employees to $1,500 a month per AI coding tool; Atlassian has also capped staff token use and requires manager approval for more.
Companies are also trying to avoid vendor lock-in by routing tasks to cheaper or better-suited models, while open-source alternatives could deliver about 90% of closed-model performance at up to 70% lower cost.
Even with tighter controls, overall AI spending is still rising, and Goldman Sachs last month forecast a 24-fold increase in global token consumption by 2030, driven by AI agents.
As AI costs soar, are companies sacrificing groundbreaking innovation for short-term budget relief?
Will the corporate pivot to free, open-source models ultimately dethrone AI giants like OpenAI?
The 2026 AI Agent Cost Crisis: Why Usage-Based Pricing Triggered a $2 Trillion SaaS Market Correction
Overview
In 2026, major corporations rapidly expanded AI agent use, encouraging employees to maximize AI token consumption through internal competitions. This widespread push, combined with usage-based pricing models where every AI interaction consumed costly tokens, led to a surge in operational expenses far beyond expectations. As companies scaled up deployments, the cumulative costs triggered a severe market correction in the enterprise software sector. This financial shock forced organizations to halt AI expansion and rethink their strategies, highlighting the risks of unpredictable costs and the need for more sustainable, value-aligned approaches to AI adoption.