A machine-learning approach to supplier selection for resilient and sustainable semiconductor manufacturing

Cathy Liu Wenhua

Abstract

Global disruptions such as pandemics and logistics crises have underscored the importance of resilience and sustainability in semiconductor supply chains. Supplier selection and order allocation are pivotal decisions that influence not only cost efficiency but also operational robustness and environmental responsibility. Conventional frameworks often focus on cost minimization, neglecting preparedness and sustainable resource use, which leads to systemic vulnerabilities.

This study proposes a machine-learning–based decision-support model that integrates sustainability and resilience into supplier evaluation and order allocation. Using K-Means and Agglomerative Hierarchical Clustering, the framework identifies natural supplier clusters based on economic, environmental, and social (ESG) criteria, enabling differentiated sourcing strategies and risk mitigation. Linear and stochastic programming models further optimize order allocation under uncertainty to maintain resilience without compromising sustainability.

Numerical case studies calibrated for semiconductor manufacturing show that the model performs effectively under disruptions such as demand surges, supplier failures, and carbon regulations. The framework provides a practical, data-driven tool for identifying sustainable suppliers, optimizing orders, and achieving balanced trade-offs between cost, risk, and environmental goals—offering valuable insights for practitioners, policymakers, and researchers pursuing resilient and sustainable supply chains.