At a Fortune CFO dinner in New York this week, finance expert Mark Williams warned that companies may need to "rethink how much they spend on AI as financing costs rise". With borrowing costs increasing after the US midterm elections, he expects AI budgets "to be reduced" as investment decisions face greater scrutiny. Williams, a Boston University lecturer and former Federal Reserve bank examiner, told the gathering of finance chiefs that they must be "more disciplined about how they fund AI initiatives and evaluate their returns," with "explicit measures of success or failure" to connect project spending to financial results.
He cautioned that cutting back too little or too late on costly AI ventures could leave firms at a "competitive disadvantage", but doing nothing is not an option. Amid persistent inflation and higher rates, CFOs should assume "the cost of capital is going up" and factor that into their plans. In practice, that means only green-lighting AI projects with clear owners and metrics, and preparing to pause or re-prioritise initiatives that cannot demonstrate near-term value.
A new survey highlights how one area of AI investment, AI-powered search tools in marketing, is surging ahead of its evidence of return. On November 11, Berlin-based startup Peec AI launched “Primary Sources”, a quarterly research series on AI in marketing. Its first report, based on a survey of 700 marketing decision-makers in the US, UK and Germany conducted by Sapio Research, found that "89% of companies raised AI search spend in the past year and 81% plan to raise it again, yet three in four that raised it can’t show what it returned" ([1]). In other words, only about a quarter of marketing leaders can prove any ROI from their new AI search investments ([2]).
Even among firms that significantly boosted this spending, 64% "cannot prove a return" on it ([3]). Peec AI’s data suggests many organisations still lack clear success metrics: only 47% of respondents said they can trace a lead, deal or revenue outcome back to AI-driven search on at least a weekly basis ([4]). The top obstacles to demonstrating value were the rapid pace of change in AI platforms (cited by 33% of respondents), concerns over brand safety (30%) and unreliable measurement (28%) ([5]). Marketing teams, Peec AI concludes, must not only invest in AI tools but also in better tracking and attribution, if they want to convincingly justify these initiatives to the C-suite.
Recent research underscores how rare it is for established companies to see measurable bottom-line impact from AI so far. Last month, Apollo Global Management’s chief economist Torsten Slok analysed productivity data for the S&P 500 and found that AI has delivered "zero financial payoff to corporate America outside of tech" ([1]). In a similar vein, Gartner reported that while "three-quarters of board members expect AI will deliver…productivity and cost savings," a full 97% of organisations have failed to meet their AI-related cost savings goals through productivity gains ([2]).
Fresh data from the first half of this year bolster these warnings. According to a new Gartner analysis shared at an October event, only about 4% of the 1,000-plus companies in the S&P Global 1200 index reported any material cost savings from their AI initiatives in H1 2026 ([3]). The reality, for now, is that even as boards and executives demand tangible results from AI projects, very few non-tech enterprises can point to significant cost reductions or revenue gains directly attributable to AI. This ROI shortfall is pushing finance leaders to intensify oversight of AI spending and focus on projects with clearly measurable outcomes.