Driven by ambitious climate targets, the rapid growth of electric vehicles (EV) is creating complex global supply chains that are increasingly exposed to disruptions. While supply risks, such as political instability and resource scarcity, are widely recognized, systematic assessments of disruption events across the entire EV supply chain remain limited. This study employs a Large language model (LLM) with prompt engineering to identify disruption events from unstructured textual data and further analyze risk dynamics. Applying this approach to 3125 news articles from 2006 to 2025, we extract 2843 documented events and further categorize them into 50 disruption event types and 8 supply risk types, spanning 49 supply chain stages and 78 countries and regions. We find that geopolitical risks have emerged as the dominant concern in recent years. Vehicle assembly, battery manufacturing, and lithium ore extraction emerged as the most vulnerable stages. Geographic heterogeneity in risk distribution reveals the distinct roles countries play in the global EV supply chain. This proposed LLM-based framework provides actionable insights for enhancing supply chain resilience during the accelerating EV transition. • Establish a comprehensive, stage-explicit supply risk taxonomy and event database for the global EV supply chain. • Propose a scalable LLM-based methodology for comprehensive supply chain risk monitoring and assessment. • Identify 50 disruption event types and 8 supply risk types, spanning 49 supply chain stages and 78 countries and regions. • Provide data-driven insights to support targeted policy and industrial strategies for enhancing supply chain resilience.
Zhang et al. (2026) studied this question.