Half of firms see AI payoffs as CFOs raise ROI scrutiny.
ROI leaps for AI - at last.
After years of hype with scant returns, new research indicates a turning point in AI’s value delivery. Boston Consulting Group’s just-released Applied AI Index for 2026 finds 48.5% of companies are now generating “meaningful value” from AI, a dramatic rise from the mere 5% that reported substantial returns in 2025 ([1]) ([2]). In other words, nearly half of enterprises are finally seeing real ROI from their AI initiatives, overturning the prevailing narrative that artificial intelligence investments rarely pay off. This middle tier of firms has shifted from experimentation to execution - they’re no longer dabbling in pilots, but actually seeing significant efficiency gains and new revenue streams.
This encouraging development comes amid an explosion in AI spending. Corporate AI investment has doubled in a year to about 3.3% of the average company’s revenue, up from roughly 1.7% in late 2025 ([3]). Moreover, more than 80% of that AI spend now happens outside the traditional IT budget, as business units directly fund their own AI-driven projects ([4]). AI has rapidly evolved into a core business expenditure - one that many boards and CFOs may have underestimated - and it’s fuelling a new urgency to ensure these bets actually deliver results ([5]).
Yet even with nearly half of companies finding payoffs, the other half remain in what one analyst called the “ROI gap,” still struggling to translate AI into bottom-line impact. BCG’s data suggests the conversation around AI is evolving: the question is no longer “Can AI create value?” but rather “Which organisations will capture that value - and how will they govern its costs and risks?” ([6]) The stakes for getting it right are growing as AI moves from the lab to the heart of the business.
Boards and CFOs demand results, not experiments.
As AI projects proliferated, corporate leaders have lost patience for initiatives that don’t deliver. This week, a new finance-industry survey underscored a stark shift: only 26% of boards now say “invest now and sort out the returns later,” while 66% will only fund AI with proof of value and 22% have completely frozen new AI spending until ROI is shown ([1]). In other words, the days of blank cheques for AI experiments are over ([2]). Boards and investors are pushing for accountable AI investments - in fact, one global poll found 53% of investors expect positive returns from AI in six months or less, a timeline most technologists consider wildly unrealistic ([3]).
Chief financial officers are feeling this heat directly. In a new survey of 260 finance leaders, 87% said they must tie AI spending to concrete business outcomes within one year to justify the expense - but only 22% can currently make that connection ([4]). This yawning gap leaves roughly four out of five CFOs essentially **unable to prove** whether their AI budget is delivering real value ([5]). Little wonder that three in five finance leaders admit they’re already spending more on AI than they can rationalise to the business ([6]).
Without clearer returns, cost-conscious executives are hitting the brakes. According to the same survey, fully 75% of CFOs who lack ROI evidence have put additional AI investments on hold, and 35% have even cancelled at least one AI initiative that wasn’t paying off ([7]). By contrast, among those who can demonstrate value, these retreat rates drop to 38% and 11%, respectively ([8]). The takeaway for leadership teams: if you can’t prove ROI, don’t expect the funding spigot to stay open.
The high cost of intelligence.
One reason ROI has lagged is that the true costs of advanced AI at scale are only now coming into full view - and they’ve caught many organisations by surprise. Early AI pilot projects were often subsidised or run on free credits, masking the real price tag of large-scale AI computing. But once companies began deploying AI “agents” and large language models in production, the cloud bills spiked. Case in point: when Microsoft switched GitHub Copilot to usage-based billing, one power user found his charges would jump from a flat $10 per month to a projected $180 - after just one day of heavy coding with the AI assistant ([1]). Multiply that across thousands of users or numerous AI-driven workflows, and suddenly a promising pilot can snowball into a budget-busting line item.
These cost overruns are forcing tough choices. In its second-quarter global AI survey, KPMG found that 49% of executives had delayed or scaled back AI agent deployments once it became clear that operating costs were outweighing the benefits ([2]). This isn’t a mass retreat from AI - in fact, the same survey noted that 79% of those leaders still rank AI as a top investment priority ([3]). Instead, it’s a sign of maturation. Companies are reining in the freewheeling experimentation of 2023 - 25 and refocusing on financial sustainability. One industry observer called it “rephasing, not retreat,” as organisations pause or restructure costly projects now in order to invest more wisely later ([4]).
In response, a new financial discipline around AI is emerging. Until recently, barely a quarter of big firms had any real-time visibility into what their various AI applications were spending on cloud resources ([5]), leaving CFOs blindsided by usage-based pricing. Now, 53% of companies report having implemented AI cost-tracking dashboards so they can “see spending as it happens” ([6]). And 54% have embedded cost checks into their project approval processes - meaning no experimental AI system is allowed to scale up without a business case demonstrating a viable cost-to-value ratio ([7]). Finance leaders are also starting to treat AI budgets like cloud computing spend, instituting governance mechanisms (e.g. monthly cost reports, unit-cost forecasts, chargebacks to business units) to ensure AI investments don’t spiral out of control ([8]).
Cracking the code to real ROI.
What distinguishes the minority of companies that are reaping strong AI returns? Research and real-world case studies point to a clear pattern: the winners focus on transformation and tangible outcomes, not technology for technology’s sake. As one analysis found, CEOs who do see financial returns from AI are “two to three times more likely to have embedded AI extensively across decision-making” - they don’t just buy tools, they actually **rewire operations** around AI capabilities ([1]). In short, value comes from redesigning core processes with AI in mind, rather than simply layering AI on top of old workflows.
Leading companies also make sure every AI project starts with well-defined business metrics and rigourous oversight. “CIOs who funded broad experimentation are the ones sweating; CIOs who tied spend to a specific, measured workflow are defending, and often growing, their budgets,” notes one tech advisory executive ([2]). Organisations that attach a clear KPI and owner to each AI initiative - and kill those that aren’t delivering - are managing to avoid the fate of the 95% of pilots that never fully scale or pay back their investment ([3]). They also invest in the less glamorous essentials like data readiness, integration, and training, knowing that AI’s impact comes from changes in processes and people, not just algorithms.
The payoff for this disciplined approach is significant. BCG’s study found the most “future-built” companies (roughly 7.5% of firms) are achieving 2.4× higher revenue growth and 2.8× greater profit growth than their laggard peers, thanks in large part to AI-driven innovations ([4]). And an additional 41% of companies are now actively scaling AI in ways that deliver above-average performance, signalling that the recipe for AI ROI is becoming clearer for those willing to fundamentally change how they operate ([5]). For C-level leaders, the implication is clear: To close the AI value gap, treat AI as a strategic business transformation - with the same accountability, process re-engineering, and cross-functional commitment you’d demand of any major investment.