PurposeThis study constructs an efficiency evaluation framework for assessing the human, structural and relational capital in the semiconductor industry of Taiwan. Furthermore, we analyze whether there are significant differences in efficiency across different levels concerning the industry supply chain (upstream, midstream and downstream), employee service tenure, capital scale and company establishment years.Design/methodology/approachThis study focuses on Taiwanese semiconductor companies, utilizing data sourced from the Taiwan Economic Journal (TEJ) Database for the period spanning 2017 to 2021, encompassing a total of five years. Due to the nondisclosure of intangible asset values by all companies, an effort was made to ensure a comparable baseline by excluding companies with incomplete or missing data. Finally, empirical analysis was conducted on a sample of 64 companies using the dynamic network data envelopment analysis method.Findings(1) Overall efficiency demonstrates structural capital as the most prominent, followed by relational capital, while human capital shows relatively poorer efficiency. (2) To enhance the efficiency of intellectual capital, priority should be given to improving the efficiency of outputs related to intellectual property rights such as patents. (3) The midstream segment exhibits the best efficiency in both structural and relational capital. (4) Companies with longer employee service tenures exhibit superior efficiency in human capital in the long run. (5) Companies with extended establishment years and larger capital scales demonstrate superior efficiency in both human and structural capital.Originality/valueReflecting on past literature, scholars have primarily focused on the relationship between intellectual capital and firm efficiency, often emphasizing the overall efficiency of intellectual capital. However, within organizations, human capital, structural capital, and relational capital are interrelated. This study, for the first time, assesses the efficiency of these three components within an organization. The research addresses the challenges in analyzing the efficiency of intellectual capital and introduces a highly contemporary approach – dynamic network data envelopment analysis (DNDEA). Using the semiconductor industry in Taiwan as a case study, this paper conducts empirical analysis in a captivating and worthy industry. Therefore, the ideas presented in this paper are original.