Artificial intelligence spending has become a standard budget line item at the majority of large enterprises for the first time in 2026, a structural shift that reflects both the scale of AI expenditure and a CFO-driven push for formal accountability. According to a survey of 800 CFOs at companies with over $100 million in annual revenue, 68% now have a dedicated AI budget line — up from 31% in 2025 — with median AI spend reaching 4.2% of total IT budget.
The formalization of AI budgets is driving a parallel demand for AI ROI measurement frameworks. In the early adoption phase, AI spend was typically absorbed into departmental budgets, classified as software or professional services, and evaluated informally. As the spend becomes material — often running into millions of dollars annually for mid-market enterprises — CFOs are applying the same accountability standards used for other capital allocation decisions.
The challenge is that AI ROI is genuinely difficult to measure with standard accounting tools. The productivity benefits of AI coding assistants, writing tools, and data analysis capabilities often manifest as freed-up employee time rather than direct revenue or cost reduction, and attributing that time savings to business outcomes requires assumption chains that CFOs find unsatisfying. Companies that have solved this problem typically did so by defining specific workflows where AI replaced or accelerated manual work, measuring output volume before and after, and applying a cost-per-unit-of-output metric.
Cost containment is becoming a significant theme. Several enterprises that deployed AI tools broadly in 2024-2025 without per-user or per-department cost tracking are now discovering that actual AI spend significantly exceeded budgeted amounts, driven by token consumption rates that were underestimated during procurement. The retroactive installation of usage monitoring and cost allocation tools is a common remediation project in enterprise IT teams this year.
The positive signal is that companies that built formal ROI measurement early are reporting consistently positive returns and are increasing AI budgets with confidence, while companies that invested without measurement are often pausing or restructuring their AI spend pending clearer accountability data.