Infosys research finds one in four executives see AI ROI below expectations
- Infosys survey of 1,000+ US executives finds 25% report AI ROI below expectations
- 80% view AI as critical for revenue, but value is largely in efficiency and speed
- Two-thirds struggle to measure ROI; nearly half lack centralized KPI frameworks
- Fewer than 25% of AI pilots successfully scale to enterprise-wide deployment
- Governance and workforce behavior cited as critical emerging ROI factors

*this image is generated using AI for illustrative purposes only.
Infosys Limited released a new research report highlighting a significant disconnect between executive expectations for artificial intelligence and the actual returns organizations are realizing. The study, titled The AI ROI Gap: Turning Ambition Into Enterprise Value, surveyed over 1,000 senior U.S. executives at large companies.
While 80 percent of respondents view AI as critical for new revenue opportunities, one in four reported that ROI from AI investments has fallen short of expectations. The findings suggest that while leaders often evaluate AI through a revenue lens, the technology’s strongest measurable impact currently lies in accelerating speed to market and improving operational efficiency rather than driving top-line growth.
Key Findings on AI Value and Measurement
The report identified several structural challenges in how enterprises evaluate and scale AI initiatives. Three-quarters of respondents stated that AI’s overall value to their organization has been net positive. However, the data reveals a divergence in where value is being captured versus where it is being sought.
| Metric | Finding |
|---|---|
| Executives viewing AI as critical for revenue | 80% |
| Executives reporting AI ROI below expectations | 25% |
| Organizations struggling to measure AI ROI | 66% |
| Pilots successfully scaled to enterprise-wide deployment | <25% |
Two-thirds of executives indicated their organizations struggle to measure AI ROI effectively. Nearly half lack a centralized key performance indicator framework, and only a quarter formally track speed to market, despite it being one of AI’s strongest areas of impact. This suggests many companies may be undercounting the value AI is already creating by relying too heavily on revenue-based metrics.
Scaling and Governance Barriers
Scaling remains a primary barrier to realizing returns. Nearly three-quarters of respondents noted that fewer than 25 percent of AI pilots have successfully scaled to enterprise-wide deployment while delivering their intended ROI. Executives cited unclear business cases and poorly defined ROI targets as the primary reasons initiatives fail to progress beyond experimentation.
Workforce behavior and governance are emerging as critical factors. As adoption expands, concerns regarding employee overreliance on AI tools, unapproved usage, and security risks are rising. The findings suggest that governance, training, and human oversight are becoming as important to AI success as the technology itself.
What the Numbers Show
The data indicates a measurement mismatch: while 75% of respondents see net positive value from AI, 25% report ROI below expectations. This divergence suggests that companies lacking centralized KPI frameworks (nearly half of respondents) may be failing to capture non-revenue benefits such as cost savings and speed to market, leading to perceived underperformance despite operational gains.
Historical Stock Returns for Infosys
| 1 Day | 5 Days | 1 Month | 6 Months | 1 Year | 5 Years |
|---|---|---|---|---|---|
| +1.24% | +0.36% | +9.90% | -10.31% | -25.33% | -33.52% |
How will the shift from revenue-centric to efficiency-centric AI metrics influence enterprise budget allocations in the upcoming fiscal year?
What specific governance frameworks are emerging to mitigate security risks associated with unapproved AI usage and employee overreliance?
Which industries are most likely to bridge the gap between AI pilots and enterprise-wide deployment given their existing data infrastructure?


































