Gartner: AI token use to surge 24-fold by 2030, budgets at risk

The discussion around generative AI often focuses on automation and labor savings, but the real problem many companies are grappling with is cost management. AI chatbots and agents consume tokens each time they are used, and charges accumulate based on volume. Unlike software billed by users or licenses, AI's cost structure scales with usage, making budgets harder to predict. Gartner's survey found that 85% of business leaders expect AI to reduce costs, and 87% expect benefits, but this gap between expectation and reality has led to a wait-and-see attitude that is already causing budget overruns.

To address this, Gartner outlines a four-step framework: understanding the billing mechanism, clarifying ROI accountability, ensuring flexibility and predictability at signing, and continuously monitoring post-signing usage. Only 14% of respondents said they were fully prepared for ROI measurement, suggesting most companies still lack the frameworks to evaluate AI's returns properly—not just cost reduction and revenue growth, but also shorter work hours and higher customer satisfaction.

The stakes hinge on whether companies can shift from a narrow focus on model unit costs and token prices to a more holistic view of task-level ROI, usage and performance. How well they internalize this cost discipline will likely determine whether AI becomes a sustainable driver of value or a recurring budget drain.