Regarding the production side of energy enterprises, digital technology and intelligent technology help in realizing real-time monitoring of the production process, as well as reducing production costs, energy transportation loss rates, and production failures (Dalla’Ora et al., 2022). For example, artificial intelligence technology can automatically detect and warn of faults, ensure the stability of energy transmission, and prevent safety accidents. Predictive maintenance functions play a crucial role in the energy industry because humans cannot predict every failure, and artificial intelligence technology can effectively identify energy equipment corrosion, cracks, inadequate insulation, and other defects, thus achieving an early warning function. With the deepening of the integration of digital technologies, automatic early warning monitoring and control at the millisecond level will be expected in the future. For example, intelligent solutions help improve supply chain link efficiency, such as enterprise production and operation, and improve the efficiency of energy resource allocation (Fu et al., 2023). Supply chains for specific energy sectors, such as the energy and gas industry, are complex systems that involve sales decisions by oil suppliers/distributors, market prices, refining operations, gantry operations, and product transportation. Digital technology can help managers in the day-to-day production and operation of auxiliary enterprise analysis. Ancillary features include, but are not limited to, management decisions such as optimizing energy selling prices, creating smart warehouses, maintaining inventory, handling transportation operations for replacement assets, risk hedging, and reducing lead times, all of which help managers quickly take relevant actions to reduce overall operating costs and achieve EE goals. From the consumer side, intelligent solutions improve the energy consumption pattern, change the terminal consumption pattern, and save resources (Xue et al., 2022). With the increasing maturity of Internet technology and the growing strength of Internet platforms, workers can choose to work at home, which is conducive to saving energy consumption in office settings and reducing the energy consumption caused by commuting. The wide application of digital technology in public transport, online car booking, and private cars can reduce the empty driving rate of public transportation and online car booking, reduce the waiting time of private cars at traffic lights, and optimize the choices of travel routes through real-time sharing of road information to alleviate traffic jams and reduce unnecessary energy consumption.

Energy enterprises can achieve the real-time collection of production data and the accurate management of production energy consumption, which helps the enterprises to customize energy use solutions based on supply-side demand in order to avoid excessive services and improve personal and household energy utilization. The accelerated development of digital energy and information technology and multi- functional collaborative management platform technologies has gradually broken the barriers between entities in different fields such as coal, oil and gas, electricity, communications, and automobiles, and the information flow between different industries has initially realized greater interconnection. According to the Energy Information Administration (EIA) of USA, nearly half of energy users in USA have smart meters installed in their homes. These meters can provide data about personal energy consumption. The data can predict upcoming energy use levels and help customers better regulate their consumption, such as finding the cheapest time to charge an electric car or run an air conditioner or optimizing energy storage. In summary, this study puts forward the first research hypothesis:

Hypothesis 1 (H1): Digital technology will improve energy efficiency.

The home country company absorbs and transforms the cutting-edge foreign technology brought by OFDI, applies it to the production and manufacturing link and ultimately forms a competitive market advantage with spillover technology as the core (Song and Wang, 2019). Companies participating in global trade activities may have a more vital awareness of new technologies and more up-to-date knowledge. They will be motivated to keep up with foreign trading partners in technological innovation. Through various channels, they can absorb the host country’s advanced technology and management experience in energy conservation and emission reduction, and realize the reverse spillover of the host country's technology (Zhong and Moon, 2023). For example, enterprises in the home country embedded in the R&D resource-intensive areas and related industrial clusters of the host country can absorb the advanced environmental protection technologies of the host country using resource sharing and technology cluster mechanisms; and then, through the flow of talents and the feedback mechanism of advanced technological achievements, a win-win situation of economic development and environmental protection forms in the home country (Kogut and Chang, 1991; Gong and You, 2022; Ma and Gao, 2022).

OFDI can optimize the utilization of resources by accelerating resource sharing and technical cooperation, which accelerate the overall low-carbon process and energy-saving technology innovation in the home country, and will ultimately improve EE. Based on using the host country's advanced scientific and technological resources to improve their own technical levels and innovation abilities, overseas subsidiaries share cutting-edge patents, management experience, and upstream and downstream channels with the parent company through information transmission, the flow of R&D personnel, feedback of R&D results and product flow, to promote the improvement of the parent company's technical level and achieve reverse technology spillover at the enterprise level. After the parent company fully absorbs these advanced technologies to achieve economies of scale, they will be passed on to the upstream and downstream enterprises in the same industry through the demonstration role, resulting in a demonstration effect and competitive behavior in the domestic market. On the one hand, this allows other enterprises in the same industry to learn, imitate and re-innovate the advanced technology and products of the parent company, and at the same time, it encourages the upstream and downstream affiliated enterprises that provide supporting services to the parent company to continuously improve their technical level and the efficiency and quality of production factors supply. On the other hand, the absolute competitiveness of the parent company in the product market will cause competitive pressure on its peers, prompting other enterprises to take the initiative for improving their technological innovation capabilities and even eliminate inefficient enterprises by the law of “survival of the fittest” in order to improve the allocation efficiency of energy factors. The technological upgrade brought by OFDI through reverse technology spillover includes innovation in management, technology, and system, which systematically impacts EE.

Digital technology optimizes the way that enterprises in the home country obtain information, enabling them to make use of online platforms and big data to obtain information, expand their search scope and matching efficiency, break information barriers, weaken information asymmetry, fully grasp overseas market information, reduce search costs and information costs, and effectively reduce their transaction costs. The end result is greater ease of investment. Moreover, the interconnectivity of digital infrastructure and the breadth of digital technology coverage will significantly reduce the cost of commodity transportation, information acquisition, and dialogue exchange in international trade activities, thereby enhancing the host country’s ability to attract China’s OFDI, and ultimately changing the structure of China’s OFDI and promoting the increase of China’s vertical OFDI scale to the host country. Therefore, digital technology improves OFDI by improving the convenience of investment and reducing the channels of trade costs. Based on these considerations, this study puts forward the second research hypothesis:

Hypothesis 2 (H2): Digital technologies can improve EE by increasing access to OFDI in the home country.

Marshall proposed three sources of positive externalities of industrial agglomeration, namely, labor reservoir, intermediate input sharing, and knowledge spillover. The impact of VA on EE is also inseparable from these three channels. The application of digital technology is conducive to expanding the scope of information interaction among enterprises, broadening the geographical spatial pattern of industrial agglomeration, and strengthening the influence of agglomeration on the regional economy. The boundary-free organization theory holds that information, resources, and ideas can quickly cross boundaries between businesses, enabling managers to respond quickly to environmental changes. When the daily operations of enterprises are no longer bound by geography and physics, the production potential will be maximized under the support of digital technologies such as AI, cloud computing, the Internet of Things (IoT), and blockchain. With the acceleration of technological innovation iteration, virtualization features are becoming increasingly prominent, reducing the congestion effect caused by geographical agglomeration and promoting the vertical and horizontal expansion of the energy industry chain.

Advanced technology is fundamentally changing the development paradigm of traditional industrial organizations and promoting the continuous evolution and renewal of industrial development models. Industrial clusters gradually break the barriers of physical boundaries and transform from geographic spatial agglomerations to network VAs with the real-time exchange of data and information as the core (Wang et al., 2018). This model is intended to transform the demand parties and related enterprises from geographic space agglomeration to cloud-level collaborative agglomeration, reduce transaction costs by shortening the information exchange distance of each production link, and finally realize dynamic, flexible production. As a new trend of industrial organization and a new agglomeration economic model in the era of the digital economy, VA promotes the blurring of industry boundaries and the virtualization of industrial clusters (Duan and Zhan, 2023). With the open ecosystem of digital platforms as the carrier, VA can integrate all aspects of social reproduction, such as production, exchange, distribution, and consumption. The spillover effect of agglomeration is no longer limited by geographic proximity, and the information between upstream and downstream enterprises and end consumers can be transmitted and communicated quickly, accurately, and immediately (Tan and Xia, 2022). The spillover effects of closer network connections even outweigh the effects of physical agglomeration in real society (Wang and Liang, 2022).

The positive externality of VA is more from the joint effect of cross-network externality and general network externality, which is regarded as an “invisible community on the Internet”. It not only has the positive externality of traditional geographical agglomeration but also helps to break the geographical space limitation and provides a unique advantage in optimizing production factor allocation, knowledge, and information-sharing linkages (Song and Lu, 2017). Digital networks provide docking channels for the circulation of resources, equipment and waste materials, and provide complementary resource integration information for industrial production networks. The most typical example is that industrial agglomeration will also form a closed material flow between various types of enterprises, which means that the waste “abandoned” by one enterprise may be an essential raw material for another enterprise. At the cyberspace level, company search costs will be reduced, which will not only improve resource recycling but also help improve regional energy efficiency. For example, under the traditional market model, the logistical costs are reflected in the “iceberg cost” of commodities and then reflected in the form of costs in commodity prices. Physical space distance brings transportation costs and information interaction costs due to team cooperation, while the psychological distance from a geographical distance brings a “trust” cost to cooperation. Although VA cannot directly save the cost of goods transportation like traditional industrial clusters, it can reduce the unnecessary costs by optimizing transportation routes, providing goods on demand, improving supply chain efficiency, and helping companies to collect, process and analyze information more effectively through virtual operation, which reduces the costs related to finding and evaluating potential trading partners and improving logistics efficiency and output efficiency, ultimately achieving the goal of improving EE (Chen, 2017; Yang et al., 2023). As another example, VA changes the mode of enterprise organization product innovation and technology research and development. The agglomeration of industrial clusters in cyberspace will compress the “distance” of the division of labor and cooperation, making it possible to specialize in the division of labor nationwide. The energy industry chain under VA pursues a competitive advantage rather than a comparative advantage of resource endowment. The pursuit of internal production optimization is a closed process of value creation, and internal resources easily restrict its value creation ability. In a virtual industry cluster, enterprises can obtain more external resources through network interaction to compensate for the lack of internal resources. The enterprise introduces the non-core business into the professional force through third-party outsourcing and makes up for its shortcomings through the division of labor and cooperation, so that it can focus on developing the leading business. Finally, the tacit knowledge, which was previously difficult to communicate or systematically express, is encoded and decoded in digital media, VR/AR, and other next-generation information technologies, which allows that information to be transmitted in virtual space and transcend the limitations of physical space. This provides a new path for enterprises to use resources better, share knowledge and optimize innovation strategies. By leveraging the cross-border penetration capabilities of the Internet and the IoT in connecting the factors of production, a high degree of integration between offline and online is promoted on a broader scale, which greatly expands the storage space of resources throughout society and constantly changes the efficiency of factor allocation and input mix of business production, which can influence EE (Hanelt et al., 2021).

Hypothesis 3 (H3): Digital technologies can improve EE through channels that promote the virtual clustering of industries.

In order to determine whether digital technology can improve EE, this study constructed the following panel econometrics model combined with the above set of various variables and research H1:

where, the subscripts i, t, and j represent province, time, and the j-th control variable, respectively. a₀ and b₀ represent constant terms, a₁ and b₁ represent regression coefficients for digital technology, Z is the information set of a series of control variables, v_i is the individual fixed effect, δ_t is the time fixed effect, ε_it is the random disturbance term subject to the white noise process, and DT is digital technology. SFEE and TFEE represent single-factor energy efficiency and total factor energy efficiency, respectively.

This study used the more commonly used approach of the two mechanisms in order to validate the intermediate path of digital technology to improve EE. The new intermediary effect model proposed by Jiang (2022) focuses on explaining how institutional variables affect EE in the components of theoretical analysis and research hypothesis, and then the impact of digital technology on institutional variables is tested in the empirical analysis component. The main observation is whether the coefficients and significance of the core explanatory variables in the second part of the equation meet the expectations. Compared with the classical panel fixed effect model, the interactive panel fixed effect can better fit the data and fully consider the impacts of various uncertain factors on the real economy and society (Bai et al., 2009; Petrova and Westerlund, 2020). This method has important applications in controlling for missing variables, capturing time-varying features, and improving goodness of fit. To alleviate the endogeneity problem of the mediation effect model, the following two equations were established in this study.

where, c₀ and d₀ represent constant terms, c₁ and d₁ represent regression coefficients of digital technology, c₂ and d₂ represent the regression coefficient of the control variable, $\text{ }\!\!\nu\!\!\text{ }_{i}^{\text{T}}{{\text{ }\!\!\delta\!\!\text{ }}_{t}}$ is the interactive fixed effect, and Z represents a set of control variables. OFDI and VA represent outward foreign direct investment and virtual agglomeration, respectively.

Following the principles of data availability and comparability, this study selected panel data from 30 provinces in China from 2006 to 2021 as the statistical samples. Note that samples from Tibet Autonomous Region and Hong Kong, Macao, and Taiwan are not included due to missing data values and inconsistent statistical caliber. The primary data of the relevant variables in this study came from the China Statistical Yearbook, China Outbound Direct Investment Statistical Bulletin, China Energy Statistical Yearbook, International Federation of Robotics, EPS data platform, and the statistical yearbooks of provincial and municipal statistics. For the very few missing values, this study used the linear interpolation method to complete the dataset. In order to eliminate the negative effects of outliers and heteroscedasticity, 1% tailing treatment and logarithmic conversion were performed on both ends of all continuous variables. The descriptive statistical analysis of the relevant variables is shown in Table 2.

The results show that both the F test and Hausman test reject the null hypothesis at the 1% level, indicating that the FE model is the most suitable for the sample data in this study. Driscol-Kraay’s (DK) estimation method sets the error structure as heteroscedasticity and a specific order autoregressive. Compared with other estimation methods, this method can obtain consistent standard errors in the control of heteros- cedasticity and autocorrelation. When the time dimension is gradually increased, the standard errors are robust to the general form of sectional correlation and time correlation (Driscoll and Kraay, 1998; Shen et al., 2023b). According to research H1, the two-way fixed effect (TWFE) was used to calculate Eqs. (1) and (2) and the results are shown in Table 3.

The data in Table 3 show that for the results of OLS, which does not include the time-fixed effect or the individual fixed effect, the regression coefficients of DT on TFEE and SFEE are 0.2349 and -0.0833, respectively. Both reject the null hypothesis at the significance level of 1%, which initially confirms research H1, that DT can improve EE. Based on the results of the TWFE, the regression coefficients of DT for the two categories of EE are 0.2267 and -0.2497, respectively. They are significant at 5% and 1%, respectively, indicating that technology can improve TFEE and reduce energy consumption per unit of GDP, that is, DT can improve EE. Thus, H1 is confirmed. DT has significant advantages in reducing the cost of data analysis and improving the speed of information transmission, which helps to improve the optimal combination of production factors such as labor, capital, energy, and technology, accurately allocate factor resources, reduce energy consumption, and ultimately improve EE (Tang et al., 2021).

In order to verify the robustness of the baseline regression results, this study used two methods of replacing the econometric model and the core explanatory variables. First, the comprehensive feasible generalized least squares (FGLS) method was used to correct the potential autocorrelation, heteroscedasticity, and cross-sectional correlation of short panel data to obtain more effective estimators. Secondly, the spatial location information of different provinces was incorporated into the model using the spatial Durbin model. The rapid development of DT is based on accelerating the flow of factors and optimizing the input-output combination. Due to the many economic connections among Chinese cities, the impact of DT on cities is not independent but has a strong spatial correlation. Similarly, advances in energy-using technology in a region or industry may improve the overall energy efficiency in the region or adjacent regions through technology spillover, the industrial chain, or market imitation. The third method is replacing the proxy indicators of DT. In order to implement the national big data strategy and accelerate the creation of a new economic and social development platform supported by big data, the “Action Outline for Promoting Big Data Development” issued by The State Council in 2015 proposed that pilot work related to big data should be carried out in-depth to achieve the integration of big data-related infrastructure and the convergence and utilization of data resources. Promoting the use and sharing of data resources is the primary task of establishing big data pilot zones (Guo et al., 2023; Shen et al., 2023a).

As seen in Table 4, the test results of these three methods all show that DT can significantly improve the TFEE and reduce the EE per unit of GDP. The signs and directions of these result coefficients are consistent with the results of the bidirectional fixed-effect model. That is, the conclusion that DT can improve EE is robust.

Although this study assumes that DT is exogenous to EE and tries to control the external variables affecting EE as much as possible, the model design still needs to address the problem that missing essential variables, such as EE, may be affected by other factors such as environmental regulations, resource endowments, and consumer preferences. In addition, one cannot rule out that DT may be endogenous; that is, there could be a reverse causality endogenous relationship, such as provinces with higher EE may have a complete digital infrastructure and more advanced management experience, pay more attention to the digital transformation of enterprises and flexible production of products, and their DT research and development and application are more convenient. To eliminate the potential endogenous problem, this study used the instrumental variable method to deal with it. Referring to the ideas of previous studies, this study used the number of post offices per 10000 people in each region in 1984 as the instrumental variable (Zhao et al., 2020; Shen et al., 2023b). The instrumental variable selected in this study was the cross-section data for 1984, but that data does not change with time. Therefore, following the solution of the existing literature, this study introduced a variable (the number of Internet broadband access ports) that changes with time and interacts with the historical data of cross-section to form the panel data, and then the instrumental variable of this study was constructed (Nunn and Qian, 2014; Zhao et al., 2020). In addition, to verify the robustness of the test results of the instrumental variable method, this study also used the generalized space two-stage least square method (GS2SLS) to estimate the sample data. This method takes the higher-order spatial lag term of the explanatory variable as the tool variable and estimates the spatial panel model based on the 2SLS method, which can control the spatial spillover effect and endogeneity of DT and EE simultaneously (Baltagi and Liu, 2014).

As seen in Table 5, the LM and Wald F test results of the 2SLS method show that the instrumental variables selected in this study are influential through the under-recognized and weak instrumental variable tests. The results of both 2SLS and GS2SLS show that DT can significantly improve TFEE and SFEE, and the estimation results using the instrumental variables are consistent with the baseline regression results, which verifies the robustness of the baseline regression results.

In order to examine the channel mechanism by which DT improves EE, combined with research hypotheses 2 and 3, this study used the interactive fixed effect model for verification.

The data in Table 6 show that the regression coefficient of DT for VA is 0.1121, and it has passed the significance test at the 10% level. These results suggest digital technologies can improve EE through channels facilitating VA. Thus, H2 is tested. The integration and application of big data, the industrial Internet of Things, and 5G digital technologies help to accelerate the agglomeration of new production factors (data) in virtual cyberspace and the entire flow among various subjects, breaking the dependence of traditional industries on geographical space and promoting the formation of close connections between enterprises and enterprises and between enterprises and consumers in the network information space. On the one hand, the flow and agglomeration of production factors of digital infrastructure in the virtual space network provide a material carrier, which helps the real-time circulation and exchange of factors on the network at low cost and high efficiency, overcoming the problem of information asymmetry, driving the flow of factors to areas with more development space, and optimizing the allocation of factors across the core of the division of labor structure. On the other hand, DT enables VA entities not only to publish output information quickly but also to instantly obtain any number of intermediate inputs that exist in the market. At this time, the market effect of intermediate inputs will be amplified infinitely, reducing external transaction costs (Ru and Liu, 2022). Therefore, DT integrates the resources of various participants in the service ecosystem, realizing the dynamic allocation of production, services, and resources in the virtual space, as well as the value co-creation between service providers and users, ultimately contributing to the improvement of EE.

The data in Table 6 also show that the regression coefficient of the influence of DT on OFDI is 0.6555, and it is significant at the 5% level. These results suggest that digital technologies can improve EE by increasing the channels for home-country companies to invest overseas. Thus, H3 is verified. First of all, with the help of DT, it is easier for enterprise management to obtain the marketing information of subsidiaries and departments in different countries and information that accurately captures idle resources, thus reducing the stickiness of enterprise costs (Warren et al., 2015). The application of various digital technologies accelerates the exchange of information elements in the supply chain and the connections of resources. Enterprises can make timely and reasonable responses according to the information fed back by digital technologies, which can improve the efficiency of capital utilization, reduce the possibility of idle resources and reduce the stickiness of financial costs. The automatic control of the business processes promoted by DT reduces the probability of management's self-interested manipulation and helps curb the cost risk caused by opportunism. Secondly, DT can improve business efficiency and promote OFDI. Digital platforms represented by Amazon, Dunhuang, eBay, Twitter, and Zoom can assist management in actively pushing information directly to various market participants scattered around the world, including investors, creditors, and suppliers, under the condition of efficiently and accurately processing and outputting adequate information, thus improving the matching efficiency between the demand side and the supply side of platform-based service enterprises (Liu et al., 2015; Warren et al., 2015). The de-intermediation operation mode of DT can reduce the transaction costs of overseas mergers and acquisitions and help enterprises sort out existing resources and redistribute them, improve resource utilization efficiency, and reduce management costs. Finally, DT can enhance the adaptability of enterprises to the international market and promote OFDI. DT can not only improve the internationalization tendency of enterprises by reducing the transaction costs of information exploration, international communication, and logistics, but it can also improve the correlations between enterprises themselves and the upstream and downstream enterprises in the supply chain through information sharing and encourage enterprises to implement internationalization strategies. Enterprises can use digital infrastructure to improve their information processing capabilities, enhance their understanding of international markets, enhance their ability to perceive opportunities in dynamic and complex international markets and enhance the flexibility of global supply chains (Elia et al., 2015).

Science and technology determine the future of energy, and science and technology create future energy. Based on a statistical sample of 30 provinces in China from 2006 to 2021, this study used the solid bidirectional fixed effect model and the Driscol-Kraay method to adjust the standard error, and it objectively evaluated the role of DT represented by industrial robots in improving EE and its potential pathway mechanism. Similar to the existing literature that analyzes how technological advances affect EE, the results show that DT can significantly improve the TFEE and reduce the energy consumption per unit of GDP. This conclusion is still robust after the regression of FGLS and SDM models, the use of the national big data total pilot area as the proxy variable of DT and the test of the DID model. In addition, the instrumental variable method and GS2SLS method used in this study have solved the endogenous problems, and the results show that DT can still significantly improve EE. The mechanism test results show that DT can promote VA and increase OFDI to improve EE. This research provides a valuable discussion on the role of DT in the energy field against the background of the new wave of technological revolution. The conclusions obtained are helpful for enterprises and city managers by providing experience for reference in the digital transformation of energy.

The results of this study are useful for promoting the digital transformation of the energy sector, focusing on building a high-quality digital grid and improving the digital capabilities of the energy industry. The digitalization of the energy industry is a digital upgrade of the energy industry chain and supply chain, the crux of which is a highly intelligent digital grid. This effort should continue to deepen digital grid technology, adhere to the needs of the energy industry as a guide, organize resources from all parties in the digital grid field, continue to improve the architecture of the digital grid, and strengthen the standard leading and compilation to achieve synergy between continued innovation, standard creation, and industrial application. Enterprises and government departments should focus on researching the integration of the digital grid into the national integrated arithmetic network and accelerating the construction of the national arithmetic network infrastructure. Promote the in-depth integration of new energy technologies and information technology, strengthen cloud computing services, and layout and build national hub nodes of the national integrated computing network. Form a distributed and open sharing network based on renewable energy, build a national energy internet with extra-high voltage grids as core nodes and coordinate the development of grids at all levels, and change the energy development mode that is now based on over-reliance on coal transportation and the development mode of unbalanced power. Comprehensively improve the intelligent interactive capabilities of the distribution grid and promote the use of distributed energy for widespread access, electric vehicles, energy storage, smart meters, and smart homes.

Relying on the digital economy, integrate and utilize the R&D strength of the whole industry chain. City managers and entrepreneurs should seize the opportunities brought about by the development of the digital economy, promote innovation in production technologies, business models, and industrial formats, closely integrate data advantages with the population advantages, market advantages, and institutional advantages of traditional manufacturing industries, deeply promote action plans based on DT, promote intelligent production and high-end industries, and advocate entrepreneurship and craftsmanship. Leveraging data resources to complete the effective docking of upstream and downstream demand in the manufacturing industry chain, enterprise manager should increase the factors of production to capital. Use the advantages of digitalization and the Internet to change traditional “high investment and low return” production mode of enterprises, integrate manufacturing modules and downstream service modules, and provide personalized and accurate products and services. Focusing on the essential positioning of energy security, administrative departments and various types of companies should promote the integrated application of intelligent manufacturing key technology and equipment, core support software, the industrial Internet and other systems, promote manufacturing service cloud platforms, intelligent connected products and enabling tools and systems, and improve TFEE by enhancing industrial VA.

Education department should strengthen the construction of a digital talent team and accelerate the construction of a conforming talent team. By combining absorption and training, we will fill the gap of composite digital talents with multidisciplinary knowledge of oil and gas, economics, law, industrial policy, etc., as well as excellent practical abilities and management technology. In the process of digital transformation, it is necessary to fully mobilize the enthusiasm, initiative, and creativity of talents in all aspects and actively participate in and lead this change. It is necessary to continuously absorb talents from various fields and adopt a multi-professional integration organization model, that is, artificial intelligence experts, mathematicians, software engineers, and oil and gas professional engineers should be closely combined to establish a multi-professional collaborative working group, so that DT and the oilfield business can be seamlessly connected. It is necessary to formulate a corresponding talent training strategy, organize DT training extensively, and build a composite talent team that is proficient in the energy business and understands DT. Industrial enterprises should increase the re-education of digital skills for existing talents, improve their digital thinking and management ability, innovate the talent management mode, stimulate talent potential and vitality, and tap into relevant digital technical talents, so as to ensure that modern information technology can play a better role in improving EE.

We thank the reviewers for their valuable comments on earlier drafts of this manuscript that helped us improve its quality. The authors are thankful for the financial support of Fundamental Research Funds from the Central Universities in Huaqiao University.