How AI Resources Translate into Supply-Chain Efficiency: Parallel Dynamic Capability Mechanisms in Manufacturing SMEs

Artificial intelligence (AI) offers manufacturing small and medium-sized enterprises (SMEs) new opportunities to improve supply-chain operations, but AI resources do not automatically generate operational value. This study examines how three AI resource-capability foundations-AI Infrastructure Flexibility (AIIF), AI Management Capabilities (AIMC), and AI Personnel Expertise Capability (AIPEC)-are translated into Supply-Chain Efficiency (SCE) through three parallel dynamic capability mechanisms: sensing, seizing, and reconfiguring. Survey data from 604 manufacturing SMEs in Zhejiang Province, China, were analyzed using a two-stage partial least squares structural equation modeling (PLS-SEM) approach. All three dynamic capabilities were positively associated with SCE and jointly explained 31.2% of its variance. All nine specific indirect effects were statistically significant, while the total indirect effects for AIIF (.143), AIMC (.145), and AIPEC (.134) were closely grouped. The findings indicate that AI value realization is distributed across multiple complementary capability-conversion channels rather than concentrated in a single dominant pathway. The study contributes by separating upstream AI resource foundations from the organizational mechanisms through which those resources become operationally valuable.