IBM Institute for Business Value, 2025 CEO Study
TL;DR
The IBM-sponsored CEO survey for 2025, run by the IBM Institute for Business Value (IBV) with Oxford Economics. 2,000 CEOs, 33 geographies, February to April 2025. Launched 6 May 2025.
- A quarter of AI initiatives paid off. “Over the past three years, CEOs say only 25% of AI initiatives have delivered expected ROI—and only 16% have scaled enterprise-wide” (PDF p. 27, printed p. 50). The unit is initiatives, not organisations.
- Most CEOs are still piloting generative AI. Asked for the main aim of their generative-AI investment, 60% say piloting and experimentation, 22% efficiency and cost savings, 7% growth and expansion, 11% not investing (PDF p. 19, printed p. 35). A year earlier, CEOs expected only 20% to still be piloting by now.
- Value beyond cost is a coin toss. Only 52% say their organisation is realising value from generative AI beyond cost reduction (PDF pp. 27-28, printed pp. 51-53).
- Investment is speeding up anyway. Expected two-year growth in generative-AI investment rises from 15% (2024-2026) to 31% (2025-2027); traditional AI from 13% to 31% (PDF p. 18, printed p. 32). 61% say they are actively adopting AI agents and preparing to scale them (PDF p. 5, printed p. 7).
- Data is the weak spot. 50% say the pace of recent investment has left them with “disconnected, piecemeal technology”; 68% call an integrated enterprise-wide data architecture critical (PDF p. 22, printed pp. 40-41).
Key data
| Measure | Figure | Unit | Page (PDF / printed) |
|---|---|---|---|
| AI initiatives that delivered expected ROI, past three years | 25% | initiatives (CEO estimate) | 27 / 50 |
| AI initiatives scaled enterprise-wide | 16% | initiatives (CEO estimate) | 27 / 50 |
| Main aim of generative-AI investment today: piloting and experimentation | 60% | CEOs | 19 / 35 |
| … efficiency and cost savings / growth and expansion | 22% / 7% | CEOs | 19 / 35 |
| Realising value from generative AI beyond cost reduction | 52% | CEOs | 27-28 / 51-53 |
| Prioritise AI use cases based on ROI | 65% | CEOs | 27 / 51 |
| Have clear metrics to measure innovation ROI | 68% | CEOs | 27 / 51 |
| Actively adopting AI agents, preparing to scale | 61% | CEOs | 5 / 7 |
| Recent investment left disconnected, piecemeal technology | 50% | CEOs | 22 / 41 |
| Expect positive ROI by 2027: scaled AI efficiency / scaled AI growth | 85% / 77% | CEOs | 28 / 53 |
Method and definitions
- Sample: 2,000 “CEOs or the most senior executives” from 33 geographies and 24 industries, surveyed in the first quarter of 2025, with CEO interviews from February to April 2025 (PDF pp. 2, 37). At least 500 respondents come from companies also in IBM’s parallel COO and CMO surveys. The sample is 68% publicly traded and 32% privately held, excluding government (PDF p. 38).
- Geography: the United States has 240 respondents (12%), the largest country group; the Netherlands has 40 (2%). Europe is 470 (24%) (PDF pp. 37-38).
- Questions: more than 150, on innovation, technology, leadership, partnerships and talent (PDF p. 4). The question behind the 25% and 16% is not printed.
- Unit of the headline figures. The 25% and 16% are shares of AI initiatives, as estimated by CEOs over three years. They are not shares of CEOs or of organisations. The report does not say whether they are averages of each CEO’s estimate or something else.
- Analysis: descriptive statistics, cross-tabulations, regression, and a two-step clustering of organisations on six variables tied to digital transformation (PDF p. 37). The clusters are not reported as a maturity share.
- The 14% average AI ROI on the same page comes from a separate IBM study of Chief AI Officers (endnote 4), not from this survey.
Scope and reliability
Self-report by CEOs. IBM sells AI, hybrid cloud and consulting, and its CEO and Vice Chairman write the foreword; the five “mindshifts” are advice from a vendor. Oxford Economics ran the fieldwork. The sample is large, the country and industry breakdown is published, and repeated questions are tracked year on year. The report is candid about slow progress: it reports that CEOs’ own forecasts from 2024 were far too optimistic about leaving the pilot phase.
A newer edition exists. The 2026 CEO Study (IBM newsroom release of 4 May 2026; 2,000 CEOs and equivalent leaders, 33 geographies, 21 industries, February to April 2026) reports that 76% of organisations have a Chief AI Officer, up from 26% in 2025, and that “AI-first” organisations have scaled 10% more AI initiatives enterprise-wide. Its release does not repeat the 25% and 16%. It was not acquired.
Dynamic-capabilities reading
digital-seizing/balancing-digital-portfolios: 65% of CEOs say they choose AI use cases on ROI, and only 16% of initiatives scaled. The report’s advice is “Focus less on quantity and more on quality” (PDF p. 29).digital-transforming/improving-digital-maturity: the integrated data architecture that 68% call critical, against the piecemeal technology that 50% report.contextual/internal-barriers: CEOs rank organisational silos, aversion to risk and disruption, and lack of expertise as the top three barriers to innovation (PDF p. 18, printed p. 32).
How it connects
- enterprise-ai-adoption and the thread ai-maturity-measurement-comparison: the main instrument in the corpus that counts initiatives. It should not sit on the same axis as instruments that count organisations or respondents.
- Deloitte State of AI in the Enterprise 2026: the pilot-to-scale gap from a 3,235-leader survey that counts respondents, not initiatives.
- Krishna on Bain’s Winning with AI: IBM’s own CEO argues for a few scalable bets instead of 100 experiments, the qualitative side of the 16%.
- micro-productivity-trap: 52% see generative-AI value beyond cost reduction, and only 7% name growth as its main aim.
- ai-agents: 61% say they are adopting agents.
- Accenture, front-runners’ guide: released the same day, it counts firms rather than initiatives: 8% front-runners scaling AI at enterprise level.
What was actually ingested
The full PDF, fetched from IBM’s own download link, converted with pdftotext. The AI, ROI, innovation, data and methodology spreads were read in full; the leadership and talent chapters were skimmed. The generative-AI investment chart (PDF p. 19) was read from a page render. The IBM newsroom release of 6 May 2025 was read in full; its figures match the report. The PDF metadata shows a build of 24 October 2025, later than the launch.
Linked entities and concepts
- IBM — publisher (IBM Institute for Business Value).
- IBM, IBM Institute for Business Value: publisher, no entity pages yet (see the batch’s entity proposals).
- enterprise-ai-adoption, micro-productivity-trap, ai-agents: concepts.
Dangling (single-source mention, deferred): Gary Cohn (IBM Vice Chairman, foreword), Mohamad Ali (SVP, IBM Consulting, quoted in the release), Oxford Economics (fieldwork partner).