Article

IDC forecasts that the world's 1,000 largest companies will underestimate their AI infrastructure costs by an average of 30% by 2027 due to inadequate budgeting methods for resource-intensive projects. As AI workloads grow, complexities in calculating costs for GPUs, inference, and network usage highlight the need for companies to adopt FinOps practices to manage expenses effectively. IBM's CEO warns that investments in data center capacity may not yield profitability, urging firms to prioritize AI projects with high ROI potential.