Business leaders across industries are voicing a growing frustration with artificial intelligence: They see its revolutionary potential but are slamming the brakes on spending because the technology isn’t delivering tangible results fast enough. A fresh wave of surveys and executive interviews reveals a stark divide between hype and execution, with companies delaying multimillion-dollar investments in enterprise AI tools amid persistent glitches, integration hurdles and underwhelming returns.

At the heart of this tension is a simple truth articulated by executives from Silicon Valley startups to Wall Street boardrooms: AI works in demos, but falters in the messy real world of business operations. Reuters reports that last spring, CellarTracker, a wine-collection app, deployed an AI-powered sommelier meant to deliver candid recommendations based on users’ tastes. Instead, the chatbot proved too polite, undermining its utility and highlighting broader issues in making AI reliable for practical use.

This anecdote captures a pattern playing out in boardrooms everywhere. PwC’s 2026 AI Business Predictions outline how focused strategies and agentic workflows could unlock value, yet many firms remain mired in pilots. McKinsey’s State of AI 2025 survey, released in November, found that while 88% of global enterprises use AI in at least one function, fewer than one in three have scaled it meaningfully, with only about 20% reporting significant financial impact.

Frustration Builds in C-Suites

CEOs are candid about the gap. A recent survey highlighted by Gizmodo shows most executives admit their AI projects aren’t yielding strong returns, yet nearly 70% plan to increase spending next year—a bet on future breakthroughs amid current disappointments. U.S. bank leaders, including those at JPMorgan Chase and Wells Fargo, told Reuters they expect AI to enhance productivity but also trigger job cuts, signaling a cautious push forward despite risks.

For smaller firms, the picture is even murkier. Forbes notes that while large companies have embraced AI widely, smaller ones lag in adoption but express optimism about eventual ROI. This optimism persists even as MIT Technology Review describes a ‘great AI hype correction of 2025,’ where disillusionment sets in after years of soaring expectations fueled by tools like ChatGPT.

Posts on X from tech influencers like Aaron Levie of Box underscore a ‘capability overhang’—AI models can handle many tasks, but enterprises haven’t fully adopted them due to imagination gaps and integration barriers. Levie, after speaking with hundreds of IT leaders, pointed out that current models solve problems not yet exploited at scale.

Pilots Stuck in Neutral

Enterprise adoption stalls at the pilot stage for reasons beyond tech limitations. Fortune’s analysis of 2025 trends reveals companies succeed when prioritizing problems over AI itself, rather than leading with flashy demos. Data challenges loom large: 90% of enterprise data is unstructured, complicating AI deployment, as noted in discussions around Natalie Coughran’s insights on foundational hurdles.

McKinsey emphasizes that the bottleneck isn’t model-building but embedding AI into decision pathways. Perceptron Network on X highlighted how, despite widespread use, verifiable data shortages keep most firms from enterprise-scale rollout. This echoes Chomba Bupe’s 2023 warning—repurposed in ongoing debates—that leaders misjudge AI’s pace, risking premature bets.

Financial services offer a microcosm. Bank executives foresee productivity gains but brace for workforce shifts, per Reuters. Meanwhile, broader investor skepticism brews, with Reuters reporting divided views on whether massive AI investments signal a bubble, as demand growth lags spending.

Skills and Data as Hidden Barriers

The skills gap exacerbates delays. ZDNet predicts 2026 could bring real ROI if firms address unexpected elements like workforce training, with AI shifting from cost-cutter to capability extender. Ethan Mollick has cautioned on X that viewing AI solely as a cost-saver stifles imagination, hurting competitiveness.

JPMorgan’s forthcoming AI stack choices will dictate firm-wide productivity, Levie argues on X, compounding to shape market positions. Yet, as NOCSM notes on X, AI disrupted business spending cycles—flat trends deviate from norms as firms await savings that may take 5-10 years.

Himani Mishra on X cites McKinsey: AI value accrues to the 20% scaling it into decisions. This requires rewiring workflows, dismantling silos and approvals, as JK on X describes, treating AI as an extension of human effort.

2026 as Tipping Point?

PwC envisions agentic AI driving transformation, but success hinges on responsible innovation. Reuters’ broader coverage warns AI reorganizes strategic conditions for investment and competition, impacting security and economic resilience. Fortune stresses problem-led approaches amid three dominant 2025 trends.

Bank leaders’ job-cut predictions signal bolder moves ahead. Forbes’ small-firm bullishness suggests a democratizing wave if barriers fall. Levie’s enterprise AI stack thesis implies winners will emerge from stack decisions compounding productivity.

As 2025 closes, executives balance patience with pressure. Reuters captures the sentiment: Leaders agree AI is the future—they just wish it worked right now. The next year may determine if hype yields heft or more hesitation.