Corporate leaders worldwide are rapidly adopting artificial intelligence (AI) to accelerate daily decision-making yet only a small fraction is using the technology to fundamentally reinvent their organizations, according to a global survey released by Deloitte.
The study, based on responses from 3,500 senior executives, paints a picture of an enterprise landscape caught between rapid short-term efficiency gains and the harder structural work required for lasting transformation.
While 81% of respondents reported that AI has noticeably accelerated decisions and execution, only 25% said they are leveraging the technology for deep organizational redesign — such as developing novel products, overhauling core operations, or altering fundamental business models.
Financial returns are nonetheless gaining momentum: 28% of organizations now report measurable revenue growth driven by AI implementations, up from 20% earlier this year. As corporate expectations evolve, executive priorities are broadening from mere productivity gains and cost cuts toward customer relationship enhancements and operational innovation.
However, financial oversight remains incomplete. While 71% of companies actively monitor their aggregate spending on AI inference and APIs, visibility drops significantly beneath the top-line ledger. About half track actual token consumption, and just 44% calculate granular per-unit inference costs — leaving corporate finance departments with little insight into the specific drivers behind escalating technology bills.
Workforce structures and corporate governance are similarly lagging adoption. Some 60% of executives surveyed noted their firms have redesigned employee roles around AI capabilities to some extent. Yet only 15% categorized those workforce changes as extensive, and just 15% reported significant shifts in decision-making authority.
At the same time, systems are growing increasingly autonomous: 40% of firms now permit AI to execute tasks or make decisions without mandatory human sign-off. Although most organizations have designated owners for technological failures, 23% acknowledged that accountability for negative outcomes remains ambiguous.
Looking ahead, emerging frontiers like physical and sovereign AI are reshaping strategic roadmaps. Enterprise utilization of physical AI, systems interacting with physical environments, is projected to nearly double from 25% today to 48% within two years. Yet specialized talent remains critically scarce, with only 19% of leaders rating their internal physical AI capabilities as high.
Geopolitics and regulatory pressure are also driving architectural shifts. Nearly three-quarters of respondents said that data residency and domestic computing infrastructure are central to their strategic planning, while 62% cited growing concerns over dependence on foreign AI platforms.
As enterprise adoption matures, industry observers note that winning organizations will be those that pair deployment speed with rigorous unit economics, operational redesign, and strict internal accountability.
AI does not have to reinvent a company to be useful. Helping employees solve problems faster or produce better work has value, points out Stephanie Walter, practice leader for AI Stack & Enterprise Application Development at HyperFRAME Research. But the pressure to show quick returns can also limit what companies attempt, she quickly added.
“For publicly traded companies, improving an existing process can be easier to justify to investors than funding a larger change whose payoff is uncertain and may take years,” Walter said. “Deeper changes require more than technology spending. Companies have to rethink workflows, decision authority, and who is responsible when something goes wrong. My concern is that responsibility may not be keeping pace with automation. Giving AI more work to do is easier than establishing who answers for the result.”
