Computing power and data elements have become new drivers of China’s economic growth, and cross-regional deployment of computing power has emerged as a key measure to facilitate the development and utilization (D&U) of enterprise data elements.

In order to examine the impact of computing infrastructure construction on the D&U of enterprise data elements, this paper uses the BERT and DeepSeek large language models to conduct text analysis on policy documents related to data elements and the MD&A sections of corporate annual reports, and constructs an indicator system for the D&U of enterprise data elements. Based on data from China’s A-share listed companies from 2014 to 2023, this paper treats the “East-to-West Computing Resource Transfer” project as a quasi-natural experiment and empirically examines its impact on the D&U of enterprise data elements. The results show that the project enhances the D&U of enterprise data elements, driven by four mechanisms: improved computing power supply, coordinated computing power transportation, inclusive penetration of computing power, and application-enabled computing power. Further analysis indicates that this effect is more pronounced among firms with stronger data governance capabilities, larger firms, and industries with higher sensitivity to computing power demand; it is stronger in regions with higher levels of marketization, higher levels of data openness, and better business environments. Additionally, the project fosters regional economic growth, alleviates development imbalances, and reduces carbon emission intensity, thereby facilitating coordinated green development.

The marginal contributions of this paper are reflected in the following aspects: First, by leveraging the policy shock of the “East-to-West Computing Resource Transfer”, it examines the impact of computing power infrastructure on the D&U of enterprise data elements and proposes the micro-level enabling mechanisms. Second, by integrating BERT’s semantic understanding capabilities with DeepSeek’s technical advantages, it constructs an indicator system for the firm-level D&U of data elements and offers a new methodology for constructing data element indicators. Third, it constructs a “elementization–marketization–servitization” framework for data and aligns the micro-level enabling mechanisms of the project with the process of the D&U of enterprise data elements, providing a new theoretical explanation for how the project promotes the D&U of enterprise data elements.