Supply chain resilience development in the defence industry: an AI-assisted content analysis of annual reports
研究分析了七家上市国防公司2014-2024年年度报告，借助AI提取内容，揭示供应链韧性如何通过桥接和缓冲策略应对不同时期的供需与内部风险，对管理者和学者理解国防供应链韧性演变有参考价值。
Purpose This study aims to examine supply chain resilience (SCR) in the defence industry by mapping supply-side, demand-side and internal risk dimensions and corresponding mitigation strategies disclosed by seven publicly listed defence firms over 2014–2024. Drawing on Resource Dependence Theory (RDT), the study introduces the Strategic Risk and Resilience Roadmap (S3R) to explain how firms use bridging and buffering to manage evolving resource dependencies. Design/methodology/approach The study uses a longitudinal qualitative content analysis of annual reports. It combines artificial intelligence (AI)-assisted extraction using large language models (LLMs), iterative prompt engineering and manual cross-validation against source reports to categorise risks, trace mitigation strategies and track shifts in disclosed mitigation postures across three disruption eras: Baseline (2014–2019), Pandemic (2020–2021) and Emerging Risk (2022–2024). Findings Firms adapt SCR by rebalancing bridging and buffering as resource dependencies change across disruption eras. In the Baseline era, firms rely mainly on bridging to stabilise external dependencies; during the Pandemic, they scale buffering to protect continuity under acute shocks; from 2022, they combine to adopt a hybrid posture as constraints persist and multiply. Supplier capacity and availability dominate post-2022 and customer-side risk reorients from funding uncertainty to delivery–performance exposure during ramp-ups. S3R traces a path-dependent process in which shifts in resource dependence drive risk reassessment, posture recalibration and learning that strengthens SCR over time. Research limitations/implications While based on publicly available reports and AI-extracted content, the analysis may omit informal or undisclosed practices. Future research should triangulate findings with interviews and firm-level performance metrics to deepen the insights. Originality/value This paper applies a longitudinal, AI-assisted qualitative content analysis of annual reports across supply-side, demand-side and internal risk dimensions, and it extends RDT into SCR by revealing dynamic bridging and buffering cycles across three disruption eras. It offers theoretical insight into the temporal evolution of defence SCR and a practical, multi-dimensional diagnostic framework.