Boards demand ROI proof as CFOs tighten AI budgets.
CFOs clamp down on AI overspending.
Finance leaders are reining in AI initiatives as real-world costs overshoot initial plans. A recent U.S. Bank CFO survey found 51% of senior finance executives spent more on AI projects than they had budgeted in the past year ([1]). Likewise, a global CFO index by Pigment, a financial planning platform, reported 83% of firms using usage-based AI services saw their costs come in above expectations ([2]). Usage-based pricing models have made AI expenditures unpredictable, with one-quarter of CFOs admitting they lacked full visibility into their live AI costs ([3]).
In response, companies are tightening control over how AI budgets are allocated. According to Bloomberg, many CFOs have begun centralising AI spending oversight and cutting back on unproven projects ([4]). Some early heavy spenders have imposed drastic measures: Uber’s finance team reportedly burned through its entire 2026 AI budget by April, prompting rationing, and sources say Amazon and others have slapped new caps on AI trials ([5]). The message is clear, with AI bills mounting, CFOs are no longer signing blank cheques.
Boards and CFOs insist on ROI evidence.
Amid growing scrutiny, boards and finance chiefs are making return on investment a precondition for future AI funding. A CloudZero survey of 252 finance leaders revealed that 66% of boards now demand proof of value before approving new AI spend ([1]). Many CFOs are being pressed to produce ROI metrics for AI projects that they currently cannot provide ([2]), reflecting a persistent measurement gap in tying AI deployments to business results. For nearly half of finance leaders, managing AI spend, and explaining its benefits, has become the most stressful part of the job ([3]).
CFOs are responding by embedding formal ROI checkpoints into their governance of technology projects. In Deloitte’s latest finance survey, 66% of CFOs said they employ an in-house approval process that puts measurement at the forefront of major AI investments ([4]). More than a quarter have adopted a “stage-gate” approach that releases funding only when pilot projects meet specific goals ([5]). A further 25% now require a quantified ROI and business case before greenlighting big new AI initiatives ([6]). In practice, that means AI projects must demonstrate clear benefits, whether cost savings, productivity gains or revenue uplifts, or risk losing funding.
CFO surveys point to imminent payoffs.
Despite current concerns, many finance chiefs remain optimistic that AI will deliver bottom-line improvements soon. Deloitte’s third-quarter UK CFO Survey found that over a third (37%) of finance directors expect AI to boost their profit margins within the next 12 months ([1]). Nearly half believe these technologies will reduce operating costs, and 52% foresee headcount savings over the same period ([2]). This optimism comes even as two-thirds of UK CFOs admit their organisations are only about a quarter of the way through the AI-driven changes they ultimately envisage ([3]), a sign that most firms are still in early stages of adoption.
Other research offers a mixed picture on realised returns. A Gartner survey of senior finance leaders found 55% reporting positive overall ROI from their 2025 AI projects, yet 57% of the most common finance AI use cases had no clear return at all ([4]). This suggests that a few big wins are boosting overall confidence even as many smaller initiatives struggle to show value. Meanwhile, a survey of CFOs at investor-backed companies by Consero Global indicated that more than 75% of finance AI investments were already delivering positive returns within a year ([5]). The outperformance by this subset highlights how disciplined execution and accountability can turn AI from a cost centre into a source of measurable value.