KPMG’s Q2/2026 Global AI Pulse survey of 2,145 senior executives across 20 countries, from enterprises with over $50 million in annual revenue, shows that 49% have delayed or scaled back AI agent deployments because operating costs exceeded the benefits generated.
However, AI remains a top investment priority. 79% of leaders stated that AI continues to be a strategic investment area, up from 74% in the previous quarter, with an average investment of about $188 million per enterprise.
The proportion of businesses considering AI as part of daily work increased from 13% to 22% in just one quarter, the largest increase since KPMG began tracking the AI maturity index.
In the Asia-Pacific region, 81% of businesses reported that AI has generated clear business value, up from 69% three months ago.
The article suggests that the main reason for the sharp rise in AI costs is that providers are shifting from fixed subscription models to token-based pricing, while AI agents consume many more tokens than chatbots because they perform prolonged tasks, call multiple tools, and continuously self-check results.
For example, after GitHub Copilot transitioned to usage-based pricing, one user estimated that costs could reach about $180 per month, instead of the fixed $10 per month plan, after just a single long coding session utilizing multiple tools.
Another issue is that enterprises lack the ability to track costs. According to KPMG’s survey in the US of 204 executives from billion-dollar enterprises, only 26% have full real-time visibility into AI costs.
Globally, about one-third of leaders stated that they do not yet fully understand the AI cost structure and token-based pricing mechanisms, hindering the expansion of AI agents.
To control costs, 53% of enterprises have deployed AI cost-tracking dashboards, while 54% have integrated cost assessments into pre-approval processes before scaling AI agents.
KPMG believes the current trend is not enterprises abandoning AI, but rather restructuring investments, focusing resources on applications that prove business efficiency and maintain good cost control.
📌 The enterprise AI market is transitioning from the experimentation phase to financial optimization. Although 49% of businesses have reduced the scale of AI agents due to high costs, 79% still consider AI an investment priority. As token-based pricing models become the standard, the ability to measure costs, forecast budgets, and evaluate investment efficiency will become the decisive factor for the success of enterprise AI programs.
