Only 28% of AI Infrastructure Projects Deliver ROI, Gartner Survey Reveals
Key Takeaways
- ▸Only 28% of AI infrastructure projects achieve full success with positive ROI, while 20% fail entirely
- ▸Unrealistic expectations and poor project scoping are the primary drivers of AI initiative failures in IT infrastructure
- ▸Skill gaps and data quality issues each affect 38% of organizations that experienced AI failures
Summary
A new Gartner survey of 782 IT infrastructure and operations (I&O) managers has found that only 28% of AI infrastructure projects fully succeed and deliver return on investment, with one in five projects failing outright. The research, conducted in November and December, revealed that 57% of I&O leaders have experienced at least one failure when applying AI to their operations. Key obstacles include unrealistic expectations about AI's speed and capabilities, skill gaps, poor data quality, and overly ambitious or poorly scoped initiatives that don't fit organizational needs.
Gartner research director Melanie Freeze attributed many failures to inflated expectations, noting that tech leaders often assume AI will immediately automate complex tasks and cut costs. The most common failures occur in auto-remediation, self-healing infrastructure, and agent-led workflow management. However, the survey found higher success rates—around 53%—in more mature AI applications like generative AI for IT service management and cloud operations. The findings underscore broader challenges companies face in justifying AI spending, with recent surveys showing that over 80% of corporate executives detect no discernible impact on employment or productivity despite widespread AI adoption.
- Mature AI applications in ITSM and cloud operations show higher success rates (53%) compared to emerging use cases
- Board pressure to demonstrate ROI is intensifying, with 98% of CIOs reporting increasing pressure and 71% facing potential budget cuts if targets aren't met
