28% of AI Projects in Operations Deliver Expected Returns. The Rest Are Stuck.
Most AI deployments in infrastructure and operations sit in a gray zone. They work. They don't break. But they don't justify their cost either, and teams can't decide whether to expand them or shut them down.
Gartner's latest data quantifies the problem: only 28% of AI use cases in I&O fully meet ROI expectations. Another 20% fail outright. The remaining 52% operate in that middle ground-functional but not strong enough to scale.
That middle tier is now consuming most of the evaluation effort across organizations.
Why Ambiguity Kills Scaling
The AI projects that consistently deliver share one trait: they operate inside defined systems. IT service management and cloud operations produce the clearest returns because workflows are structured and inputs are predictable. AI has less ambiguity to resolve, which leads to more stable performance.
High-autonomy scenarios struggle. Self-healing infrastructure and full remediation initiatives still can't stabilize because real systems are too variable. The vision exists. The execution doesn't.
Data Quality Determines Survival
Environmental constraints filter which use cases mature and which stall. Data quality, accessibility, and fragmentation continue to determine outcomes. Internal capability matters equally-teams that can operate and adapt these systems remain uneven across organizations.
Organizations getting consistent returns make deliberate structural choices. They embed AI into existing workflows, align ownership across functions, and manage use cases as a portfolio with clear prioritization criteria.
The CIO's Challenge
As AI spending moves into core IT budgets, tolerance for ambiguity drops. Every use case competes for survival, measured by whether it consistently improves operations over time.
For operations leaders, that means enforcing discipline across a growing set of initiatives and concentrating resources on what actually performs. Portfolio management is no longer optional.
Related: AI for Operations covers the fundamentals of deploying AI in operational environments. AI Learning Path for CIOs addresses the portfolio management and ROI challenges specific to chief information officers.
Your membership also unlocks:
