by Ujjwal Singla
ABSTRACT
Data center operators need to make sourcing decisions that consider energy efficiency, reliability, technical suitability, sustainability and exposure to operating costs. This study proposes an AI-based strategic sourcing model for data center uninterruptible power supply (UPS) procurement based on the publicly available data of UPS products certified by the ENERGY STAR. The framework transforms product certification data into sourcing intelligence by performing data screening, preprocessing, feature engineering, energy-loss cost analysis, multi-criteria AI-assisted scoring, product and supplier ranking, recommendation classification, clustering, explainable AI, traditional AI versus AI comparison, and robustness analysis. The final analytical sample consisted of 465 UPS models relevant to the data center from 15 suppliers. The findings indicate that a more balanced procurement decision is made with integrated AI-assisted sourcing than with single factor evaluation. Traditional efficiency-only and cost-only approaches chose CyberPower CP3000PFCRM1U, but the AI-based approach chose Eaton 5P1550IRT2UG2-L for its better overall efficiency, runtime, warranty and cost performance. The supplier robustness analysis revealed that CyberPower was the strongest supplier if suppliers with less than 10 products were not included. Explainable AI also found that energy cost per kW, efficiency, sustainability, and technical suitability were the primary factors influencing the outcomes of recommendations. The study provides a clear decision support process for converting public certification information to strategic sourcing intelligence and helps to make more balanced, sustainability-driven procurement decisions in data center ecosystems
