ทำไมผู้บริหารถึงควร

The reality of missed AI savings

Bain’s survey reveals a stark disconnect between expectations and results. Forty percent of firms reported cost reductions of 10% or less, a figure that contrasts sharply with the more ambitious projections that typically underpin AI business cases . In the broader market, only 23% of companies can directly tie their generative AI deployments to new revenue or lower costs, leaving the majority unable to quantify a bottom-line impact .

This underperformance is not confined to a single industry. The survey spanned retail, technology, advanced manufacturing, and several other sectors, all showing similar patterns of cost-reduction deficits . The problem is particularly acute given that 72% of companies track cost savings as their key automation metric, yet many are not seeing those numbers materialize at scale .

The uncomfortable structural problem

The core structural problem the survey uncovers is that companies are funding further AI investment on the basis of savings that have not yet arrived. Many executives are approving increased AI spending with the explicit expectation that automation will deliver offsetting cost reductions; when those savings fall short, the funding model starts to look precarious .

This dynamic is part of a larger pattern Bain has observed in its B2B Growth Agenda research. While companies expect significantly higher revenue growth rates — 20% higher in 2026 compared to the prior year — many lack the AI capabilities and data foundations needed to meet those goals. Sixty percent of surveyed leaders admit they don’t have the data infrastructure or technology required to scale AI effectively .

The numbers support that anxiety. CFOs are planning to increase enterprise-wide AI spending significantly: 83% expect to boost AI budgets by more than 15% over the next two years, and 42% plan increases of 30% or more during that period . Yet only about 23% of surveyed executives can currently point to measurable revenue or cost outcomes from generative AI . This asymmetry — aggressive spending growth against tepid realized returns — is what Bain is flagging as the uncomfortable structural reality.

Bain’s recommendations for fixing the gap

Bain’s research and prior analysis suggest that the solution is not to pull back on AI investment, but to fundamentally change how companies pursue savings from it. The firm recommends that companies stop treating AI as a standalone bolt-on that will deliver automatic cost reductions. In a separate analysis, Bain argues that productivity gains alone won’t create ROI for companies investing in generative AI, and that “leading companies are on course to achieve cost savings of up to 25% by combining end-to-end process redesign with generative AI tool deployment” .

In practice, this means embedding AI inside broader operational overhauls rather than layering it onto existing workflows. Bain’s research on automation reinforces this: companies that invested most heavily in automation — defined as spending at least 20% of their IT budget on it — achieved an average 22% in cost savings, compared to just under 8% for companies that invested less than 5% . The firms seeing the best results don’t just deploy more AI; they deploy it within redesigned processes that eliminate work rather than just speed up existing steps .

Bain also emphasizes that growth winners deploy significantly more use cases than laggards — an average of 4.5 compared to just 3.3 — and achieve nearly twice the cost efficiencies for any given use case . The firm’s broader recommendation is to stop measuring AI’s success by deployment alone and instead tie it to clearly defined process and financial outcomes, supported by the underlying data and technology architecture that 60% of companies currently lack .

What this means for AI budgets going forward

The Bain survey, released June 1, 2026, arrives at a moment when corporate AI spending is accelerating even as ROI remains elusive for most companies. Bain itself has separately calculated that the global compute buildout needed for AI demand by 2030 could require $500 billion in annual capital investment and $2 trillion in new annual revenue to fund sustainably — a figure that even optimistic savings projections don’t reach . The current savings miss, then, is not just a tactical disappointment; it’s a warning that the funding model underpinning enterprise AI is on an unsustainable trajectory for many companies.

The most urgent takeaway from the survey is that companies cannot afford to treat AI investment as a leap of faith backed by spreadsheet savings that never arrive. Instead, they need to rebuild the bridge between ambition and operational reality, starting with process redesign and a clear-eyed measurement of whether the promised cost reductions are actually showing up in the P&L.