Analyzing the Efficiency of AI Integration for Project Management Employing Dea: A Case Study for a Data Center

Abstract

As artificial intelligence (AI) workloads continue to expand in the digital economy, data centers must adapt their infrastructure and operational models to support these computationally intensive demands. This thesis evaluates the efficiency of AI workload integration in a 1 MW data center environment by applying a structured project management approach and using Data Envelopment Analysis (DEA) to assess performance across cost, time, quality, and return on investment (ROI) dimensions. Three workload scenarios are analyzed: Option 1 (no AI workload), Option 2 (20–30% AI workload), and Option 3 (50% AI workload). The baseline DEA results show that AI-enhanced configurations slightly improve quality, result in substantially increased costs and extended project timelines, thereby diminishing overall operational efficiency. To further explore this trade-off, two scenario analyses are conducted: (1) evaluating efficiency under increasing AI infrastructure costs, and (2) introducing ROI as an additional output factor. The findings show that AI options remain viable only when ROI is high and breakeven is achieved within a short time frame due to the fast pace of technological change. The study concludes that the most efficient and future-resilient strategy is to allocate 80–90% of capacity to traditional workloads and 10–20% to scalable AI deployments.