Gartner: AI's High Cost Demands I&O Leadership for Infrastructure Optimization

Gartner's latest insights highlight a significant challenge facing enterprises: the high cost of AI infrastructure. A recent survey revealed that 38% of companies are currently exceeding their AI budgets, with an overwhelming 85% anticipating even greater expenditures in the coming year. This financial pressure is forcing a re-evaluation of how AI initiatives are managed and underscores the critical need for Infrastructure & Operations (I&O) teams to take a leading role. Jonathan Forest, VP Analyst at Gartner, emphasizes that I&O must elevate its capabilities to proactively enable enterprise AI, moving beyond traditional reactive operations. This trend matters immensely to practitioners because the success of AI adoption hinges on efficient and cost-effective infrastructure. Without strong I&O leadership in designing and supporting AI infrastructure, organizations are more than twice as likely to fail in their AI endeavors. The shift towards AI agents, which perform complex multi-step tasks requiring substantial token usage, introduces new cost management complexities. This directly impacts DevOps and cloud engineers responsible for deploying and maintaining these systems, as uncapped variable pricing models tied to token usage can lead to initial cost expectations being exceeded by over 50%. This development fits into the broader, well-established trend of FinOps and cost optimization becoming central to cloud and AI strategies. As cloud spending continues to grow, with global public cloud spending projected to surpass $1 trillion in 2026, the imperative to manage costs effectively has never been greater. The emergence of AI workloads, with their dynamic scaling and unpredictable consumption patterns, further complicates traditional budgeting and forecasting models. The industry has seen a rise in estimated wasted cloud spend, reaching 29% in 2026, indicating that despite increased FinOps adoption, the challenge persists. In practice, this means I&O and DevOps teams must adopt a "no telemetry, no deployment" policy for AI systems to gain visibility into spending and token usage. They should prioritize predictable subscription and bundled pricing models over variable ones for AI agents. Furthermore, the increasing adoption of multi-cloud and hybrid environments for AI workloads—with 80% of large enterprises expected to operate production AI across at least three infrastructure environments by 2029—demands robust strategies for cross-platform cost management. Practitioners should focus on strengthening governance from the outset of AI projects, investing in data quality, and modernizing infrastructure proactively to avoid costly rework and ensure that AI investments deliver their promised value.