AI investments are soaring, but where's the ROI? a reality check for business leaders.

The financial side of AI is coming under as much scrutiny as the technical side. Many companies have been caught off guard by the true cost of scaling up AI initiatives – and new reports show why. Despite assumptions that automation would reduce expenses, a late-2025 survey found 85% of organizations had underestimated their AI project costs by more than 10%, with nearly a quarter underestimating by over 50% ([1]). Such budgeting miscalculations hurt the bottom line: in the same study, more than 8 in 10 firms said their AI spending ended up eroding gross margins by over 6%, and 26% saw margin hits above 16% ([2]).

It’s not just about data scientists’ salaries or pilot programs overrunning – it’s the rising price of compute and third-party AI services. Industry insiders note a new tipping point: some companies are now funding more for AI computing power than for the human staff using it ([3]). One global tech firm’s CTO even admitted they blew through the entire annual AI budget by March ([4]), illustrating how fast costs can spiral. And just weeks ago, OpenAI jolted enterprises by raising its top-tier API prices by 350% – from $60 to $270 per million tokens ([5]). These kinds of cost spikes are forcing CIOs to rethink the "build vs. buy" equation and push for more efficient AI usage.

This surge in spending is drawing new oversight. CFOs and procurement leaders – historically hands-off with R&D projects – are now imposing financial discipline on AI initiatives. There’s growing interest in cloud cost governance or 'FinOps' as finance chiefs grapple with runaway AI bills. Nearly 88% of organizations report that their cloud costs are rising due to AI workloads, creating a heavy drag on profitability ([6]). The takeaway for executives: treat AI like any other major investment that must earn its keep. Without proactive cost management and a clear business case, even promising AI projects can turn into expensive experiments that erode margins.

If internal budget alarms weren’t enough, external stakeholders are now piling on. Top executives are hearing a clear message from investors and directors: show us real results – and show them fast.

In a global CEO-and-investor survey released this week, 53% of investors said they expect to see positive returns from new AI projects in six months or less ([1]). This extraordinarily tight timeline – essentially demanding an immediate payoff – conflicts with the reality that transformative AI initiatives often require years of process changes and learning to bear fruit. Such pressure is putting many CEOs in a bind: they know that true ROI on AI requires patience and strategic alignment, yet they must scramble to deliver early wins to satisfy the market.

Boards are also insisting on clear accountability for AI outcomes. Nearly nine in ten financial services CEOs now say they have designated board-level or C-suite oversight of AI value delivery ([2]) – a sign that governance is catching up with the hype. Similarly, 61% of senior business leaders report that the pressure to demonstrate AI ROI has increased compared to a year ago ([3]). For many companies, 2026 is shaping up to be the 'show me the money' year for AI: initiatives that can’t justify themselves with tangible business value will face tough questions – and potentially the chopping block.

Another counterintuitive finding emerged about the workforce impact of AI: cutting employees hasn’t been the quick fix many hoped. Gartner’s new data shows about 80% of organizations implementing 'autonomous' or AI-driven systems have also reduced staff to save costs ([1]). Yet those that trimmed headcount saw no better returns – workforce reduction rates were nearly identical among companies with high AI ROI and those with minimal or negative results ([2]). In short, layoffs may create some budget breathing room, but they don’t automatically translate into business value. As one Gartner VP analyst put it, many CEOs who resort to job cuts to boost AI returns are misguided; 'workforce reductions may create budget room, but they do not create return' ([3]).

Instead, the organizations getting more value from AI are doing the opposite – doubling down on their people. Companies with successful AI outcomes invest in re-skilling and rethink job roles rather than eliminating them. Even in efficiency-driven sectors like finance, 60% of CEOs now expect their AI investments to support stable or growing headcount in 2026, not workforce cuts ([4]). These leaders view AI as an augmentation tool for employees, automating routine tasks so staff can focus on higher-value work instead of being replaced. Notably, analysts predict that by 2027, half of the companies that cut jobs due to AI will end up rehiring for those same positions ([5]). And more than half of employers already regret laying off workers in anticipation of AI gains that haven’t materialized ([6]). The takeaway for executives: redeploying and upskilling your workforce to work alongside AI yields more reliable ROI than chopping headcount for short-term savings.

If there’s a silver lining in this flood of new data, it’s a clearer picture of how the AI frontrunners are extracting value. PwC’s 2026 AI Performance Study reveals that leading companies take a different approach on two fronts ([1]). First, they treat AI as a growth engine, not just a cost-cutter. These top 20% of firms are 2–3 times more likely to use AI for developing new products, reaching new customers, or reinventing their business models – moves that create new revenue streams – instead of focusing AI solely on efficiency tweaks ([2]). By aiming AI at top-line opportunities (which show up clearly in financial results) rather than minor cost savings (which often get absorbed and go unnoticed), these leaders can credibly demonstrate ROI.

Second, the leaders re-engineer workflows and roles to fully integrate AI. PwC found that successful organizations are about twice as likely to redesign core processes to incorporate AI, rather than simply layering new tools onto existing processes ([3]). They don’t just automate tasks in isolation; they reorganize work so that AI and employees can collaborate seamlessly. This often entails significant change management, but it pays off. Boston Consulting Group reports that trailblazer firms allocate roughly 60% of their AI budgets to upskilling and reskilling their workforce to ensure people can leverage new AI tools effectively ([4]). This kind of investment in human capital helps prevent the low adoption rates that undermine many AI projects.

Finally, effective governance and accountability are critical. A recent Gartner study concluded that ROI from AI projects isn’t determined by having the flashiest algorithms, but by how well the technology is integrated, governed, and aligned to real business needs ([5]). Successful enterprises manage AI initiatives as they would any core product – with clear ownership, cross-functional buy-in, and agreed-upon success metrics from the outset. Not coincidentally, Gartner found that 77% of I&O leaders who achieved at least one meaningful AI success attributed it to embedding AI into existing workflows and securing full support from business executives ([6]). In practice, that means technical and business teams jointly own AI projects, ensuring the solutions address strategic priorities and have the necessary executive sponsorship. By focusing on these fundamentals – targeting growth, rethinking processes, investing in skills, and enforcing accountability – the leading few are finally translating AI investments into tangible business value.