A TechCrunch review of the growing “AI graveyard” (Nov 14) reveals that many much-hyped AI projects have been shut down after falling short on results ([1]). One data point from S&P Global Market Intelligence shows 42% of companies have now abandoned the majority of their AI initiatives, double the share from a year ago ([2]). Crucially, the top reasons were not flawed algorithms but business factors: high costs, data privacy and security risks, and a failure to scale or find user demand ([3]). In other words, many enterprises invested heavily in AI without a clear path to measurable ROI, prompting them to pull the plug when returns didn’t materialise.
This wave of aborted projects underscores a hard truth for leaders: AI’s value must be proven in concrete terms or investments will be written off. The S&P data highlights how quickly the narrative has shifted from excitement to scrutiny. Only a few years ago, companies raced to roll out AI pilots; now, nearly half report scrapping most of those initiatives because they couldn’t justify their impact on the bottom line ([4]). For CFOs, the lesson is that hype alone can’t pay the bills, AI efforts need clearly defined outcomes and cost controls from day one, or they risk ending up as expensive experiments on the corporate scrapheap.
Senior executives’ patience for elusive returns is wearing thin. In PwC’s Global CEO Survey (released in January), only 12% of 4,454 chief executives said that AI had delivered both higher revenues and lower costs in their businesses ([1]). A further 33% reported at least one of those benefits, but the majority (56%) admitted they had seen no significant financial gains from AI so far ([2]). These findings, discussed at this year’s World Economic Forum, highlight a widening gap between AI’s promise and its tangible impact on profit. Despite widespread adoption of AI tools, most CEOs have yet to witness a visible improvement in their financial statements.
Other research paints a similar picture. McKinsey’s latest global AI survey (August) found just 37% of respondents could attribute even “some” EBIT growth to AI, roughly unchanged from the previous year ([3]). Only 6% of organisations qualify as “AI high performers” that generate at least 5% of EBIT from AI with a significant business impact ([4]). Even sectors at the forefront of AI report modest outcomes; for example, an Apollo Global Management analysis in October concluded that outside big tech firms, AI had delivered “zero” boost to corporate earnings ([5]) ([6]). The prevailing message from the C-suite is clear: for most companies, AI has yet to move the dial on profits or productivity in a meaningful way.
With budget season underway, CFOs are intensifying their scrutiny of AI investments. Finance leaders who once green-lit ambitious AI projects on promise alone are now insisting on hard metrics and controlled experimentation. Many organisations are building formal “tokenomics” governance to monitor cloud AI usage and avoid surprise bills, reflecting concerns that open-ended use of large language models can lead to cost overruns ([1]). The latest Protiviti Global Finance Trends survey (902 finance leaders worldwide) found that 77% of finance teams have deployed AI in some form, yet only 35% say they can effectively measure its ROI ([2]). This measurement gap makes it difficult for CFOs to defend continued AI spending without evidence of tangible returns.
Boardrooms are also applying pressure. According to tax software firm Avalara, 92% of CFOs feel the need to prove that their AI spending is delivering a decent return, a sentiment fuelling more rigorous oversight of AI projects. Only 7% of finance chiefs say their organisations emphasise AI governance over speed of adoption, indicating that most have been pushing to deploy quickly and are now reckoning with the results ([3]). To regain control, CFOs are establishing clearer objectives and ‘kill switch’ criteria for AI initiatives. Some are exploring new ways to quantify value: in a widely noted “scorecard for the AI age,” OpenAI’s CFO Sarah Friar has advocated measuring AI by “useful work accomplished per dollar spent” rather than vanity metrics like user counts ([4]) ([5]). As financial stewards, CFOs are making it known that AI projects must demonstrate real business impact, either in cost savings, improved margins or revenue growth, or risk being scaled back or stopped.