AI spending is accelerating, but organizations struggle to measure whether the investments actually pay off. A new Apptio survey reveals that 90 percent of technology leaders cite ROI uncertainty as a moderate to major obstacle in tech investment decisions, up 5 percentage points year-over-year. The problem intensifies with AI, where costs remain unpredictable and governance frameworks are still emerging.
The core issue: companies deploy AI aggressively without clear visibility into how the technology translates to business value. Traditional ROI measurement breaks down when dealing with AI workloads that scale unpredictably, consume variable compute resources, and operate across hybrid and cloud environments. Infrastructure costs spike. Attribution becomes murky. Business leaders demand accountability. Technology teams lack the tools to provide it.
This measurement gap creates three operational problems. First, overspending goes undetected because nobody tracks AI compute consumption with precision. Second, teams cannot identify which AI initiatives drive revenue versus which ones drain budgets. Third, executives cannot make informed decisions about which AI projects deserve continued funding.
Closing this gap requires three actions. Organizations need governance frameworks that establish who approves AI projects and how. They need measurement systems that tie AI spending to specific business outcomes, not just technology metrics. They need chargeback models that allocate costs transparently across departments.
The stakes are high. Companies that master AI cost visibility gain competitive advantage. They shift from reactive spending to strategic allocation. They fund high-return AI projects while killing low-impact experiments faster. They justify continued investment to boards and shareholders.
The window to establish these practices is narrowing. As AI spending continues its trajectory, the cost of not measuring ROI compounds. Organizations that implement proper governance, measurement, and attribution now will position themselves to scale AI profitably. Those that defer the problem will face mounting pressure to justify increasingly large budgets with increasingly unclear returns.