IntroductionIn the era of artificial intelligence (AI), computing power infrastructure (CPI) has become a foundational pillar of the digital economy. At the same time, the security and resilience of mineral resource supply chains particularly for critical minerals such as gallium, germanium and rare earth elements face unprecedented challenges from geopolitical tensions and surging demand. Prior research has examined these two domains separately, leaving the potential of CPI to enhance mineral resource security resilience (MRSR) unexplored.MethodsWe construct a provincial CPI development index and an MRSR composite index based on a resistanceadaptationtransformation framework, using panel data from 30 Chinese provinces over 20122023. We employ a hybrid methodological approach integrating two-way fixed effects (TWFE), mediation analysis, moderation analysis, spatial Durbin models (SDM), panel threshold models, instrumental variable (IV) estimation and double machine learning (DML) for causal inference, supplemented by LASSO, random forest, XGBoost, LightGBM, support vector regression (SVR) and SHAP-based interpretability analysis.ResultsWe document five key findings: (1) CPI significantly enhances MRSR (β = 0.218, p < 0.01), a result robust to IV (0.276) and DML (0.243) corrections for endogeneity; (2) the effect operates through dual mediating pathways of technological innovation (24.8% mediation share) and information empowerment (18.4%); (3) digital governance capacity and regional openness positively moderate the CPIMRSR relationship; (4) the SDM reveals significant positive spatial spillover effects of CPI on neighbouring provinces MRSR; (5) panel threshold regression identifies a non-linear, marginally increasing pattern across two threshold values (γ1 = 0.215, γ2 = 0.478). Machine learning ensemble analysis and SHAP interpretation provide methodological triangulation that corroborates and extends the econometric findings.DiscussionThe results establish a novel “digital infrastructureresource security” analytical framework, showing that computing power is not only a consumer of critical minerals but also an enabler of their security. Because the returns are threshold-dependent and conditional on complementary institutions, concentrated and scaled-up computing power development should be integrated explicitly into national mineral security strategies.