A recent survey by the consulting firm KPMG among more than 2,000 executives across 20 countries highlights the growing challenge of controlling AI costs. While 79 percent of companies view AI as a priority for new investments, many lack the necessary transparency: around one-third of executives do not understand the complex token cost structures, and 42 percent have only a partial overview of total expenditures. Companies that implement targeted monitoring measures, such as dashboards or cost audits, achieve a significantly higher ROI of 15 percent compared to firms without such controls. Due to cost uncertainty, 49 percent of respondents have already slowed down or scaled back projects. To reduce costs, companies are increasingly relying on cheaper, open-source models or strict token budgets to ensure the economic efficiency of the technology.