Under increasing external uncertainty, promoting sustainable transformation and building data-driven supply chain resilience (SCR) supported by frontier technologies has become an important issue. This study uses Chinese A-share listed companies from 2005 to 2024 as the sample. From the perspective of frontier technologies, it employs text analysis and machine learning to identify large-scale unstructured annual report texts and quantify firm circular economy (CE) practices. SCR is measured using an entropy-weighted indicator system, and the relationship between CE practices and SCR is examined using panel regression models. The results show that CE practices significantly improve SCR, and the findings remain robust across multiple robustness and endogeneity tests. Mechanism analysis indicates that CE practices enhance SCR through three channels: resource effects, structural effects, and collaboration effects. Heterogeneity analysis reveals that the positive effect is more pronounced in non-state-owned firms, firms in non-heavy-polluting industries, high factor-intensive firms, and firms with active environmental information disclosure, as well as in regions with higher economic development and stronger governmental attention to green development, and in the period following the signing of the Paris Agreement in 2015. Further analysis shows that CE practices significantly improve the resistance dimension of SCR, while their effect on recovery is insignificant. Economic consequence analysis indicates that SCR enhancement driven by CE practices helps reduce operational risk and increase firm value. This study provides new empirical evidence on how CE practices strengthen SCR, and offers theoretical and practical implications for enabling sustainable supply chain development through frontier technologies and data-driven analytics.