Artificial intelligence has been deeply embedded in the network of social operations, serving as the underlying architecture and infrastructure of the digital society. Academia has conducted in-depth typological analyses of data risks in generative artificial intelligence from various perspectives and methodologies, including the life cycle perspective, rights theory perspective, and data security perspective, and has proposed corresponding measures to address these data risks. However, a common limitation of the aforementioned life cycle perspective, rights theory perspective, and data security perspective lies in the disconnection between theoretical logic and practical dilemmas, specifically, they fail to fully recognize the unique characteristics of data risks in generative AI and systematically examine the complex feedback mechanism between training data and generated content. Consequently, these perspectives are trapped in an internal perspective of data risk governance, neglecting the spillover effects, transmissibility, and interconnection inherent to generative AI data risks themselves. Overall, academic analyses of generative AI data risks have long been confined to the cognitive limitation of "centering on governance objects",that is, overemphasizing the attribute characteristics of "data" and the morphological manifestations of "data risks" while ignoring the core attributes and process mechanisms of "governance" itself. This has rendered existing analytical approaches and models unable to respond to the complexity of governance practices, ultimately reducing governance schemes to mere "paper compliance" that cannot be effectively implemented. Therefore, it is necessary to take governance theory as the starting point, analyze the practical dilemmas in generative artificial intelligence data risk governance, and propose targeted optimization schemes for such governance.
Professor Gerry Stoker has further refined a "governance" theory that can provide an organizational framework. "Governance as theory" emphasizes four key aspects: the clarity of governance goals, the synergy of governance subjects, the completeness of governance basis, and the flexibility of governance means. The governance dilemmas in generative AI data risks can thus be categorized into four dimensions: ambiguity of governance goals, fragmentation of governance subjects, absence of governance basis, and rigidity of governance means. Accordingly, the governance dilemmas in generative artificial intelligence data risks can thus be categorized into four major types: the dilemma of misaligned governance goals, which includes difficulties in balancing security and development, reconciling national and corporate goals, and aligning short-term and long-term objectives; the dilemma of fragmented governance subjects, which involves complex games among public authorities, the binary opposition between the state and enterprises, and international competition for rule-making power; the absence of governance basis, which covers the lack of laws and regulations, industry standards, and ethical norms; and the rigidity of governance means, which includes the path dependence on traditional administrative tools, institutional obstacles to technological empowerment, and the breakdown of synergy among multiple tools.
In response, system synergy, dynamic adaptation, risk communication, and multi-stakeholder co-governance shall serve as core concepts. Goal calibration is to be achieved through a risk-classified dynamic balance mechanism, a synergistic coupling mechanism of government-enterprise interests, and a temporal cohesive mechanism for short-term and long-term goals; subject synergy is to be realized through a collaborative linkage mechanism among public authorities, a co-governance operation mechanism for government-enterprise symbiosis, and a collaborative mechanism for international rule-making; the governance basis is to be improved through a hierarchical legislative legal guarantee mechanism, a multi-stakeholder co-governance standard-setting mechanism, and an institutionally embedded ethical constraint mechanism; and governance means are to be innovated through a precision-oriented reform mechanism for administrative tools, an institutional activation mechanism for technological tools, and a synergistic coupling mechanism for multiple tools. Furthermore, through in-depth coupling of governance goals, subjects, foundations, and means, as well as systematic reconstruction of interrelated mechanisms, continuous innovation in the generative artificial intelligence data risk governance system and a qualitative leap in governance mechanisms can be ultimately promoted, laying a solid foundation for the healthy and sustainable development of generative artificial intelligence.

Against the backdrop of intensifying global technological competition and the increasingly urgent need for self-reliance in core technologies across critical sectors, how enterprises can effectively leverage emerging technologies like artificial intelligence to drive breakthrough innovation has become a focal point for both academia and industry. Existing research predominantly examines AI's innovation effects from a technological determinism perspective, yet it fails to systematically uncover the complex internal and external organizational synergies essential for its efficacy. This study constructs an integrated analytical framework based on socio-technical systems theory. It examines how corporate AI applications (exploratory/exploitative AI applications) synergize with characteristics of the social subsystem,including the organization's digital transformation level, dynamic capabilities, executives' digital background, and the digital focus of the local government to collectively influence breakthrough innovation performance.
Using a sample of Chinese A-share listed manufacturing companies from 2020 to 2023, this study employs machine learning methods for empirical analysis. First, the K-Means clustering algorithm was applied to objectively classify enterprises based on the aforementioned socio-technical system characteristics, thereby identifying heterogeneous enterprise clusters that share similar intrinsic socio-technical configurations. Subsequently, the CART decision tree algorithm was employed to uncover the specific socio-technical system element combinations and decision rules that each cluster relies on to achieve high breakthrough innovation performance.
Key findings encompass two dimensions. First, according to the characteristics of socio-technical systems, sample enterprises can be categorized into three distinct groups: (1)Laggard enterprises exhibit low levels of both exploratory and exploitative AI applications, accompanied by slow progress in digital transformation, marked deficiencies in dynamic capabilities, inadequate internal and external organizational coordination, and thus overall stagnant development. (2)Pioneer enterprises demonstrate high levels of exploratory AI applications, digital transformation, and dynamic capabilities, along with strong forward-looking strategic awareness. However, their exploitative AI applications remain relatively weak, resulting in an unbalanced state characterized by “strong exploration, weak exploitation”. (3)Efficiency-driven enterprises demonstrate high levels of exploitative AI applications and actively respond to and leverage government digital policy support. Nevertheless, they tend to adopt a conservative stance in exploratory AI deployment, digital transformation deepening, and dynamic capability cultivation, thus following a development path centered on efficiency optimization and steady advancement.
Second, the attainment of high breakthrough innovation performance hinges on a socio-technical system configuration that matches the enterprise type. For laggard enterprises, if they possess a solid digital transformation foundation, active deployment of exploratory AI applications can effectively drive breakthrough innovation. Conversely, if their digital foundation is weak, the advancement of exploratory AI applications will instead inhibit breakthrough innovation. For pioneer enterprises, when their digital transformation foundation is robust, strengthening exploitative AI applications constitutes the key to overcoming breakthrough technological bottlenecks. If their digital capabilities are insufficient, they must rely on robust dynamic capabilities to advance exploratory AI applications. For efficiency-driven enterprises, when their digital foundation is weak but they receive strong government support, the synergistic advancement of both exploratory and exploitative AI applications can effectively foster innovation breakthroughs. In contrast, if they overemphasize exploitative applications to comply with short-term policy mandates despite having a mature digital foundation, they will risk hindering breakthrough innovation due to path dependency.
This paper makes four key theoretical contributions: refining the AI application classification framework by distinguishing the distinct roles of exploratory and exploitative AI in innovation;developing a novel enterprise classification paradigm based on socio-technical system synergies, offering a more nuanced theoretical lens for understanding context-dependent AI-enabled innovation; uncovering the multi-factor nonlinear configuration mechanism driving AI-enabled breakthrough innovation, thereby enriching research on multi-level interactive mechanisms of breakthrough innovation antecedents through a socio-technical systems perspective; and building an integrated analytical framework connecting "AI application-socio-technical system co-evolution-breakthrough innovation" , which extends socio-technical systems theory to the study of AI adoption and organizational innovation relationships.

The upgrading of enterprise value chains through artificial intelligence (AI) applications has become a pivotal concern for high-quality development. Viewed through the “smile curve”, this upgrading entails enhancing value-added capabilities in R&D/design, manufacturing, and marketing/service. However, two gaps persist: firm-level AI applications lack consistent measurement, and the micro-level mechanisms through which AI reshapes value creation remain underexplored. Therefore, this study proposes that AI applications can elevate the “smile curve” by strengthening value creation in upstream innovation, midstream operations, and downstream market activities, thereby facilitating the upgrading of enterprise value chains.
Using a panel dataset of Chinese A-share listed firms (2012-2024), the study constructs a novel AI application index via large language model (LLM)-based text analysis of annual reports. After rigorous data cleaning, the sample comprises 33 534 firm-year observations.Enterprise value chain upgrading is measured by firms' value-added performance (proxied by the value-added rate). AI application is designated as the core explanatory variable and measured through large language model (LLM)-based text analysis of annual reports.
Specifically, sentence-level narratives related to AI application are identified using a prompt-assisted annotation strategy; a classification model is then fine-tuned and validated before being applied to the full corpus to extract AI-related application sentences, which are subsequently aggregated at the firm level to construct an AI application index. Empirically, fixed-effects models are employed with comprehensive controls for firm characteristics and governance factors, while incorporating firm and year fixed effects and further controlling for industry-by-year effects. To enhance identification, this study additionally conducts instrumental-variable estimation and multiple matching-based approaches, and performs robustness checks with alternative specifications.
The empirical results indicate that AI application significantly promotes enterprise value chain upgrading, and the positive effect remains stable across different model specifications and robustness tests. Mechanism analyses further show that AI promotes upgrading through three channels: (1) improving innovation quality by expanding and deepening firms’ knowledge base, thereby strengthening upstream value creation;(2) enhancing production flexibility by facilitating digitalized, modular, and responsive operations and by increasing the effective use of IT and skilled human capital, thereby improving midstream efficiency and adaptability;and (3) enabling customized marketing through better demand prediction and customer profiling, strengthening downstream value capture and feeding market insights back into R&D and production decisions. Heterogeneity analyses based on the technology-organization-environment (TOE) framework suggest that the upgrading effect of AI is stronger for firms with richer data assets, stronger managerial incentives, and those located in regions with higher marketization levels, implying that complementary resources and institutional environments amplify AI-driven transformation. In addition, an extension analysis finds that AI application significantly enhances total factor productivity (TFP), providing evidence that the micro-level productivity effect of AI has begun to materialize.
This study contributes in four main innovative aspects. First, it develops an LLM-based measurement framework for firm AI application using unstructured annual report text, providing a scalable and replicable approach for capturing firms’ AI deployment intensity over a long time horizon. Second, it integrates the “smile curve” logic with firm-level empirical identification, clarifying how AI reshapes value creation across R&D, production, and marketing, and empirically verifying the “technology-process-market” linkage mechanism. Third, it systematically identifies key boundary conditions (data assets, incentive mechanisms, and market-oriented institutions), enriching the understanding of when and why AI more effectively translates into value chain upgrading. Fourth, by linking AI application to TFP improvement, it offers micro-level evidence supporting the realization of AI productivity gains, and provides practical implications for enterprise data governance, talent strategies, incentive alignment, and institutional support.

Technological convergence has emerged as a fundamental driver of breakthrough innovation and economic growth in contemporary technological development.Against the backdrop of China's deepening implementation of its innovation-driven development strategy, the trend of technological cross-integration has become increasingly prominent, promoting the integration and reconstruction of knowledge from different disciplines in technological innovation.However, China started late in many high-end manufacturing fields, resulting in a relatively shallow overall technological foundation.Moreover, multinational corporations often restrict access to core module technologies to maintain their competitive advantages, leaving resource-constrained firms trapped in the predicament of being "integrated but not interconnected, converged but not strong".Achieving innovation breakthroughs through technological convergence is not only crucial for the survival and development of individual firms but also vital for enhancing the overall competitiveness of the industrial chain, making it a key issue of common concern for both academia and industry.
This paper focuses on the core question: How do resource-constrained firms achieve innovation breakthroughs through the dynamic evolutionary mechanisms of technological convergence ? The investigation explores how technological convergence evolves in stages within a firm, the conditions, characteristics, models, and trends of convergence at each stage, how capabilities improve in a stepwise manner, and the factors that drive and constrain this evolutionary process.Existing research has demonstrated that technological convergence plays a significant role in promoting innovation breakthroughs.However, little research has revealed the dynamic evolution process of technological convergence at the micro-level of the firm, with a particular lack of in-depth discussion on the boundary conditions for the transformation of convergence models and the role of endogenous capability evolution.
Adopting the perspective of technological convergence, this paper employs a longitudinal single-case study method, selecting a leading Chinese automotive enterprise, Changan Auto.To ensure robustness, the research data were triangulated through multiple channels: 12 semi-structured interviews (totaling over 1 400 minutes) with senior executives, technical core members, and partners, supplemented by extensive secondary data including corporate annual reports, public speeches from the Changan Auto Technology Ecosystem Conference, academic literature, and news reports.The analysis process strictly followed the Gioia structured data analysis method, organizing the data into first-order concepts, second-order themes, and aggregated theoretical dimensions to construct a process model of the technological convergence evolution at Changan Auto across three distinct stages from 2006 to 2025.
The analysis reveals that the evolution of a firm's technological convergence exhibits a three-stage dynamic transition.Heterogeneous contexts in each stage give rise to differentiated convergence drivers, which in turn lead to convergence models with distinct characteristics.The activities of technological convergence evolve to form a dynamic capability system encompassing "convergence knowledge management capability", "convergence technology integration capability",and "convergence industrial network capability", demonstrating a progressive evolutionary trend from "single-point absorptive convergence" to "multi-dimensional integrative convergence" and then to "ecosystem penetrating convergence".The synergistic interaction of technological context constraints (comprising technology maturity and boundary characteristics), the firm's strategic intent, and endogenous capability evolution drives the transformation of convergence models, which is subject to boundary conditions.
This study makes three primary contributions.First, it constructs a dynamic evolutionary model of technological convergence encompassing drivers, conditions, models, characteristics, capabilities, and trends, shifting the perspective from a static strategy to a dynamic, multi-stage capability-building process, thereby offering a new process-oriented theoretical explanation for the asymmetric catch-up of latecomer firms.Second, it reveals the core driving role of endogenous capability evolution in the transformation of convergence models.By elucidating the synergistic driving factors and boundary conditions, it extends dynamic capability theory to the specific context of technological convergence.Third, it delineates a stepwise capability-building chain from convergence knowledge management capability to convergence technology integration capability and then to convergence industrial network capability, deepening the theoretical understanding of how firms gradually build capabilities through technological convergence.The research findings provide managers with a phased pathway reference for achieving innovation breakthroughs and offer policy insights for governments to foster a favorable technological convergence environment through targeted incentives, platform construction, and intellectual property protection.

The digital transformation of entrepreneurial firms constitutes a pivotal impetus for the high-quality advancement of private enterprises. However, they are shackled by resource constraints across technology, talent, and capital dimensions,and confront dual predicaments: the paralysis of wanting to transform yet being unable to execute and the recklessness of blind trend mimicry, both of which stem from the absence of tailored strategies. This necessitates the development of clear, actionable digital transformation frameworks. Extant studies focus on how the introduction of digital technology affects the economic performance of mature enterprises, and little is known about how entrepreneurial firms conduct digitalization to achieve sustainable growth under the dual pressure of “innovation championship” and resource endowment constraints. Based on text mining and identification, this study proposes two types of digitalization strategies: radical digitalization and incremental digitalization. Radical digitalization aims to rapidly achieve deep integration and breakthroughs in efficiency, experience, business operations, and business models through intensive, cross-boundary resource inputs. Its essence lies in seeking to disrupt existing market rules or technological pathways, with the goal of establishing new competitive moats. In contrast, incremental digitalization emphasizes applying digital technologies in a phased and controllable manner within the framework of existing business operations and resource endowments. It prioritizes the optimization and improvement of production, operations, management, and other processes. The core of this strategy is to maintain operational stability while gradually accumulating digital capabilities and adapting to market changes. Drawing upon optimal distinctiveness theory, this study explores the influence of two kinds of digitalization strategies on the sustainable growth of entrepreneurial firms from the perspective of external environment and dynamic capability.
By screening the samples of cross-year entrepreneurial firms on GEM and SME board, the study finds that radical and incremental digitalization have no significant impact on short-term profits, but they could improve market value. It indicates that the impact of digitalization on the sustainable growth of entrepreneurial firms is long-term oriented. With the enhancement of market dynamics, the implementation of radical digitalization has a negative impact on the short-term profits of entrepreneurial firms. The higher the degree of regional digitalization is, the stronger the role of incremental digitalization in enhancing the long-term value of entrepreneurial firms. Additional analysis shows that both radical and incremental digitalization can enhance market value of entrepreneurial enterprises by enhancing innovation and absorption capacity. The mediating effect of innovation capability is moderated by market dynamics and regional digitalization degree. The effect of innovation ability on market value is more prominent under high regional digitalization degree and low market dynamics. Heterogeneity analysis reveals that the moderating effect of market dynamism on the mediating role of innovation capability is significantly stronger in non-technology-intensive industries; furthermore, the analysis indicates that the moderating influence of regional digitalization on this mediation is more pronounced under the condition of Chair-CIO duality.
Its potential marginal contributions are manifold. First, it breaks through the traditional paradigm focused solely on the holistic effects of digitalization. By being the first to systematically compare the differential impacts of radical versus incremental digitalization strategies on both the short-term profitability and long-term value of entrepreneurial firms, this research reveals the time-sequenced accumulation characteristic inherent in digital value creation. This provides a novel explanatory pathway for firm growth research from a behavioral perspective. Second, through uncovering the distinct moderating effects of market dynamism and regional digitalization levels on these two types of digitalization strategies, it clarifies the "environment-strategy" alignment imperative in the digital implementation of entrepreneurial ventures. This offers a theoretical foundation for selecting contextually appropriate digital pathways in specific entrepreneurial situations. Moreover, it advances the understanding of the dynamic capabilities' action pathways. By elucidating the dual mediating mechanisms of innovation capability and absorptive capacity, the study reveals the intrinsic pathways through which digitalization strategies affect entrepreneurial firm growth. This effectively unpacks the "capability-performance" black box within dynamic capabilities theory, providing new empirical evidence for research on the microfoundations of dynamic capabilities. Finally, the study constructs a tripartite alignment decision-making framework.By dissecting the "capability-environment-strategy" alignment mechanism, it not only extends the application of the organization-environment fit theory into the digital context but also provides entrepreneurial firms with an actionable decision logic framework to navigate digitalization dilemmas.

In the context of innovation-driven development and the construction of an intellectual property powerhouse, the cultivation and transformation of high-value patents have emerged as critical drivers of the deep integration between industrial chain and innovation chains. While a limited number of scholars have conducted firm-level research showing that high-value patents can enhance enterprises' technological innovation capacity, help break the path dependence of low-end lock-in, and promote high-quality development by alleviating information asymmetry, there remains a significant gap at the regional level. Specifically, systematic analysis and mechanism identification are still lacking regarding how high-value patents exert their economic effects by optimizing the allocation of innovation factors and activating regional synergy mechanisms.
This study fills this gap by constructing a comprehensive analytical framework that connects high-value patents to the coupling and coordination of industrial chain and innovation chain via resource reallocation, knowledge spillovers, and institutional mechanisms. The theoretical foundation outlines two primary pathways through which high-value patents facilitate integration. First, by embedding core technologies in upstream production processes, high-value patents help enterprises overcome bottlenecks in key industrial segments, thereby enhancing self-reliance and technological control. Aligning innovation activities with industrial needs strengthens the synergy between R&D and production, fosters collaborative innovation networks, and facilitates value chain upgrading. Second, drawing on knowledge spillover theory, high-value patents spread through inter-firm cooperation, talent mobility, and innovation networks, thereby promoting the spatial and functional expansion of the industrial chain around innovation hubs. Compared with ordinary patents, high-value patents demonstrate greater scalability and integration potential, fostering systemic innovation synergy and mitigating the “island effect” in technology development. Furthermore, the effectiveness of high-value patents is conditioned by two critical moderating factors: regional absorptive capacity and the strength of intellectual property protection. These institutional and cognitive factors shape the extent to which external knowledge is absorbed, protected, and transformed into integrated innovation and industrial outcomes.
Empirically, the study utilizes panel data from 30 provinces in China spanning 2012 to 2023. High-value patents are identified using a machine learning-enhanced income method, while dual-chain integration is assessed through a coupling coordination model. Fixed-effects regression results, supported by robustness checks and instrumental variable methods, confirm the significant positive impact of high-value patents on dual-chain integration. Heterogeneity analysis further reveals that these effects are more prominent in economically advanced and innovation-intensive regions in eastern and central China. Mediation analysis further demonstrates that high-value patents influence dual-chain integration not only directly, but also indirectly by attracting and concentrating innovation resources, which in turn optimizes the spatial and structural alignment between the industrial chain and innovation chain.
This study makes the following contributions. Theoretically, it advances the understanding of high-value patents by framing them as systemic enablers of industrial–innovation synergy and highlights, for the first time, the mediating role of innovation factor agglomeration. Methodologically, it introduces a scalable and data-driven method for identifying patent value and evaluating dual-chain integration. To promote the economic effects of high-value patents, targeted measures should be taken from both enterprise and government perspectives across four key areas. First, enterprises should focus on core technologies to enhance the cultivation and market adaptability of high-value patents, while governments need to implement collaborative intellectual property (IP) application mechanisms and advance the integration of such patents with key industries. Second, enterprises ought to strengthen their capabilities in patent management and transformation, and governments should invest in innovation infrastructure and improve the judicial protection system for IP to foster a sound innovation ecosystem. Third, enterprises should develop patent layout strategies based on local industrial strengths, and governments should provide differentiated policies—accelerating the development of high-value patent-intensive industrial clusters in eastern and central regions, and supporting technology-industry collaboration to tap patent potential in northeastern and western regions. Fourth, efforts should be made to improve IP financial support mechanisms, such as developing diversified financial products, and build a talent team for IP services and transformation, thereby constructing a system of high-value patent transformation clusters.

The Chinese government has endeavored to accelerate the popularization and application of new-generation information technologies, promoting the development of industrial Internet, and positioning it as a key engine for new-type industrialization and a strategic foundation for the creation of new productive forces. Industrial Internet platforms, such as AeroCloud Network and Haier COSMOPlat, have empowered enterprises with digital intelligence, helping them achieve lean management, transformation and upgrading, as well as innovative development. However, with the rapid emergence of industrial Internet platforms, how to fully leverage their digital intelligence empowerment value and explore the underlying mechanisms of its realization has become a key issue to be addressed. Existing research has shown that digital intelligence empowerment can create new business logics and value paradigms, innovate manufacturing service models, and promote intelligent manufacturing and high-quality enterprise development. However, current theoretical exploration lags behind practical applications, focusing mostly on single-factor aspects. The systematic and complex nature of platform ecosystems is overlooked, with insufficient analysis of the combined effects and synergies of multiple factors. Additionally, there is a lack of quantitative assessment and systematic elaboration of the empowerment effects.
From a configurational perspective, this study develops an integrated theoretical framework that captures the essence of the digital intelligence era and the developmental requirements of platforms. Using fuzzy-set qualitative comparative analysis, the study examines 49 national “dual-cross” industrial Internet platforms. It focuses on the core question of how these platforms can achieve high digital intelligence empowerment and explores the digital intelligence empowerment of industrial Internet platforms influenced by the interplay of the platforms themselves, platform-enabling enterprises, and the operational environment. By addressing these questions, this study reveals the dynamic and complex nature of the unique driving mechanisms and realization pathways of digital intelligence empowerment. It also offers precise theoretical guidance and practical solutions for formulating strategies and optimizing operations to enhance the digital intelligence empowerment of industrial Internet platforms.
The study reaches the following conclusions: (1) The valuable, rare, inimitable, and non-substitutable resources (hereinafter referred to as “VRIN resources”) are essential for achieving the digital intelligence empowerment of industrial Internet platforms. Optimizing these resources significantly enhances the platforms' empowerment effects and drives the digital intelligence transformation and upgrading of the manufacturing industry. (2) The digital intelligence empowerment of industrial Internet platforms follows multiple pathways and complex mechanisms. Three key factors of“ecological synergy and linkage”, “market stabilization and collaboration” and “resource decision-driven”are critical in achieving high digital intelligence empowerment. These three factors represent significant types of high digital intelligence empowerment among industrial Internet platforms. Each mode reflects the strategic actions of different platforms to achieve high digital intelligence empowerment based on their unique ecological characteristics and strengths, collectively forming a multi-dimensional strategic framework for high digital intelligence empowerment. “Ecological looseness and imbalance” and “weak market competition” are typical types of low digital intelligence empowerment among industrial Internet platforms, exhibiting asymmetric relationships with the high empowerment patterns. (3) Under specific configurations, the elements of “platform subject-platform object-platform environment” can achieve high digital intelligence empowerment through complementary interactions and equivalent substitutions. This process follows a “diverse pathways to a common goal” approach.
This study develops a theoretical framework for the digital intelligence empowerment of industrial Internet platforms from a configurational perspective, thereby enriching the theoretical understanding of digital intelligence empowerment within the digital intelligence context. Meanwhile, this study develops an evaluation framework for assessing the digital intelligence empowerment effect of industrial Internet platforms, thereby expanding the quantitative evaluation of digital intelligence empowerment. In addition, it examines the realization of digital intelligence empowerment based on platform ecosystem theory, and extends the application of platform ecosystem theory. The study emphasizes that industrial Internet platform enterprises must strategically prioritize the cultivation of VRIN resources, focus on their development, and utilize them to enhance digital intelligence empowerment. Achieving high digital intelligence empowerment is not a one-size-fits-all process. Instead, enterprises should comprehensively analyze the characteristics of platform ecosystem elements and external environmental conditions, conduct effective strategic planning, and choose appropriate empowerment pathways.

The digital economy has driven the emergence of new fields and given rise to various new industries. Firms in emerging fields strive to explore, acquire, and integrate innovative resources beyond their existing technological areas in order to succeed in disruptive technological innovation. By doing this, firms in the emerging fields reshape their competitive advantages. Prior research has provided strong evidence on the relationship between cross-boundary integration and innovative performance; however, they have neglected the mechanism of how firms' cross-boundary integration gives rise to disruptive creations. Moreover, although flexible routine replication has been proven to be a key factor in solving the "convention replication dilemma" and promoting enterprise innovation under complex, changing, and unstable scenarios, current studies pay little attention to the relationship between cross-boundary integration and flexible routine replication, as well as the role of digital capability, which has emerged as a core competence for firms in emerging fields.
Thus, this study aims to address two questions: (1) how cross-boundary integration affects disruptive technological innovation via flexible routine replication and (2) how the digital capability of firms in the emerging fields enhances the effect of cross-boundary integration on flexible routine replication. Combining the theory of resource dependence and organizational routines, it constructs an analytical framework of "cross-border integration flexible-flexible routine replication-disruptive technological innovation" and tests hypotheses proposed through hierarchical regression analysis based on a sample of 128 enterprises in the emerging field . The data of this study was collected by a two-stage questionnaire survey. In the first stage, on-site interviews were conducted to ensure that all interviewees have experience with disruptive technological innovation. Then, in the next stage, the questionnaire survey was conducted via phone or email based on the list of emerging field firms provided by on-site interviewees, such as R&D managers and firm leaders.
The results show that (1) both cross-boundary resource identification and resource allocation and utilization positively impact disruptive technological innovation in emerging field firms. The findings suggest that by recombining and recreating heterogeneous technological resources from multiple fields, firms can acquire new innovation opportunities which could subvert mainstream market technologies. (2) By replicating routine, firms can effectively transform external innovation resources and explore new technologies, thereby promoting the disruptive technological innovation. (3) Digital capabilities can positively moderate the relationship between cross-boundary integration and flexible replicating routine. In other words, emerging field firms' digital transformation and usage of digital technology facilitate knowledge internalization and absorption which in turn benefit the updating of organizational practices. Unlike previous studies that focused on the relationship between digital capabilities and innovation performance or innovation capabilities, this article provides empirical evidence of the impact of digital capabilities on flexible routine replication. The results prove the positive role of digital capabilities in addressing the replication dilemma under cross-boundary contexts.
This study contributes to the present theories in three ways. First, by constructing and verifying the theoretical model of "cross-boundary integration-flexible routine replication-disruptive technological innovation", this study, from the perspective of organizational change, reveals the mechanism of how firms in the emerging fields successfully conduct disruptive technological innovation through cross-boundary integration, and also expands the application scenarios of disruptive innovation theory. The paper discloses the mechanism for firms to achieve disruptive technological innovation based on the exploration, acquisition, and reorganization of diverse external technological knowledge, i.e., cross-boundary integration forms new knowledge structures and task execution contexts for firms' routine replication which trigger disruptive technological innovation. Second, it establishes theoretical connection between cross-boundary integration and flexible routine replication. Flexible routine replication is the dynamic updating of operational processes, cognitive norms, and behavioral norms in order to suit new organizational environment and avoid replication dilemmas. Third, the study highlights the role of digital capabilities in routine replication processes. It is verified that the application of digital technology can facilitate knowledge transfer cross distinct business units so as to improve the evolution of organizational routines.

Amid the dual challenges of global climate change and energy transition, green technology innovation has become a key driver for enterprises to enhance their competitiveness and achieve sustainable development. Compared with traditional innovation, green technology innovation is characterized by high investment, long cycles, and strong externalities. These features pose greater technological uncertainty and resource constraints for enterprises during the innovation process, and they highlight the increasing importance of external support mechanisms. Against this backdrop, venture capital, as an important link between capital and innovation, has network characteristics that significantly impact enterprise green technology innovation. The venture capital network is a complex system formed by the investment relationships between venture capital institutions and the enterprises they invest in. Through mechanisms such as knowledge integration, resource allocation, and governance optimization, the venture capital network may become an important external force in promoting enterprise green technology innovation.
Therefore, this study investigates how venture capital network characteristics influence corporate green technology innovation. Drawing on resource-based view and social capital theory, the study develops a theoretical framework centered on "network centrality-value creation-innovation enhancement" to examine the potential impact mechanisms of venture capital networks on green innovation breakthroughs. This study addresses several critical gaps in existing literature. First, while previous studies have largely examined venture capital from traditional financing perspectives, this study proposes that the network position characteristics of venture capital may generate broader value creation effects beyond capital provision. Second, it investigates how venture capital networks facilitate the integration of innovation resources and knowledge transfer in green technology domains. Third, it explores the underlying mechanisms through which venture capital networks enhance corporate green innovation capabilities.
Using a comprehensive dataset of China's ChiNext-listed companies from 2013-2023, the study constructs measures for venture capital network centrality and green innovation based on investment relationship and patent data. The empirical analysis yields several key findings. First, venture capital network centrality significantly promotes corporate green technology innovation, with the effect being more pronounced in private enterprises and high-tech industries. Second, this positive impact is mediated through three primary mechanisms: knowledge spillover effects, reputational certification functions, and supervisory pressure transmission. The knowledge spillover mechanism reveals how centrally positioned venture capital helps firms optimize knowledge acquisition and integration. Through extensive network connections, venture capital facilitates the flow of green technology information and expertise across portfolio companies. The reputational certification mechanism demonstrates how high-centrality venture capital enhances firms′ credibility in green innovation domains, improving their access to external resources. The supervisory pressure mechanism shows how venture capital networks create monitoring and incentive systems that drive sustained green innovation efforts.
This study contributes to the existing theoretical framework from three aspects. First, it introduces a novel network-based value creation framework that extends beyond traditional venture capital studies. Second, by proposing a three-dimensional theoretical model, it deepens the understanding of the micro-level formation mechanisms of green innovation. Third, it expands the boundaries of venture capital network theory by revealing how network positions influence sustainable innovation outcomes. For practitioners, the findings suggest strategies for leveraging venture capital networks to overcome green innovation bottlenecks and achieve sustainable development goals. Enterprises are advised to build a network-oriented green innovation capability system,and deepen strategic collaboration with core-positioned venture capital, treating them as partners rather than funders; moreover, it is necessary to optimize response strategies for the three mechanisms of venture capital networks: knowledge spillover, reputation certification, and supervision pressure, and create a green innovation ecosystem based on networks.In this context, leading companies should create open innovation alliances, green venture funds, incubation centers, and digital platforms to strengthen collaborative innovation and overall competitiveness.
This study also suggests several directions for future research, including expanding the investigation of network structure characteristics, developing more comprehensive measures of green innovation impact, and exploring additional mechanisms in the venture capital-innovation relationship. Such research would further advance the understanding of how venture capital networks can promote the development of new productive forces and facilitate green technology transformation.

Entity List sanctions serve as a pivotal tool for the United States in its crackdown on China. A notable instance was in April 2018 when the U.S. government imposed a technology embargo on ZTE, marking the beginning of a broader technology blockade against China. Under these sanctions,targeted enterprises face restrictions on purchasing U.S.-controlled technologies, products, and services, and even with license applications, they often face strict scrutiny and rejections, leading to operational instability and higher uncertainties. As economic globalization and specialization deepen, enterprises form close supply chain networks. However, the sanctions on midstream enterprises can trigger a chain reaction, transmitting risks to upstream and downstream enterprises, increasing supply chain risks, and weakening overall competitiveness. This situation raises the question of what strategies upstream and downstream enterprises should adopt to deal with the supply chain risks caused by the technology blockade.
From the perspective of supply chain, this paper examines the two-way spillover effect of midstream enterprises on the collaborative innovation of upstream and downstream enterprises after the U.S. technology blockade. It draws on the CNRDS database to compute patent and supply chain matching data to gauge enterprise collaborative innovation. After rigorous sample processing, 2 679 valid samples emerge, comprising 1 466 "year-midstream-upstream enterprise" matches and 1213 "year- midstream-downstream enterprise" matches. Subsequently, a benchmark model is constructed, followed by in-depth empirical analysis. In the mechanism analysis, the study respectively measures the operational risk of upstream and downstream enterprises based on the annual stock daily return volatility of these enterprises, and the higher the value, the greater the operational risk. It then adopts the proportion of the sum of the top 3 executive compensations to all executive compensations to measure managerial overconfidence, and the higher this value is, the higher the degree of managerial overconfidence.
The study finds that after the midstream enterprises are subject to the U.S. technology blockade, the supply chain spillover effect is generated, which not only promotes the collaborative innovation of upstream enterprises, but also promotes the collaborative innovation of downstream enterprises, thus highlighting the symmetry of the U.S. technology blockade supply chain spillover. Operational risk and managerial overconfidence are the internal impact mechanism of U.S. technology blockade on the collaborative innovation of upstream and downstream enterprises. The heterogeneity analysis shows that the spillover effect of the U.S. technology blockade on the supply chain is more obvious in the samples with closer supply chain distance, higher supply chain stability, lower supply chain concentration and higher market status.
According to the empirical results, the following countermeasures and suggestions are proposed respectively. For enterprises, on the one hand, when making collaborative innovation decisions, they not only need to assess the risks they face directly, but also need to comprehensively assess the potential threats in the entire supply chain network, especially the potential risks brought by other enterprises in the supply chain under the sanctions of the U.S. Entity List, so as to adjust the collaborative innovation strategy in a timely manner. On the other hand, enterprises should set up a full-time risk management department, strengthen the voice of independent directors and set up a special audit committee to conduct a third-party evaluation of managers ' major decisions, so as to curb managers ' overconfidence tendency and unblock the channels through which U.S. technology blockade affect the collaborative innovation of upstream and downstream enterprises. For the government, on the one hand, it should take into account the spillover effects on the supply chain caused by the U.S. technology blockade,and support not only the midstream enterprises that have been directly sanctioned but also provide appropriate resource allocation to the upstream and downstream enterprises within the supply chain, so as to maximize the effect of policy support by improving the overall competitiveness of the supply chain ; on the other hand, the government should tailor innovation support policies to different enterprises based on their unique circumstances, so as to enhance the level of enterprise collaborative innovation.

Policy pilots have become a key mechanism through which China advances major reforms under conditions of uncertainty, yet existing studies have focused largely on descriptive accounts of pilot operation and have paid insufficient attention to how central–local interaction shapes the iterative optimization of pilot schemes across different stages of the policy process. Against the backdrop of deepening the comprehensive innovation reform in China, this study examines how policy resources and policy knowledge circulate between the central and local governments during pilot implementation, and how such interaction affects the effectiveness and evolution of reform measures.
From the policy process perspective, the study develops a "W+N" dual-type model of iterative innovation in policy pilot schemes. The "W" and "N" denote the directional vector of initiative: W-type represents a winding-down pathway where central mandates cascade to local implementation; N-type represents a nurturingup pathway where local experiments feed into central authorization. In this framework, the W-type applies when policy goals and reform measures are both relatively clear, with the central government providing concrete arrangements and local governments responsible for implementation and verification; the N-type applies when policy goals are clear but specific instruments remain open, with local governments designing and testing concrete measures under central authorization. By integrating these pathways, the model moves beyond the simple dichotomy between topdown design and bottom-up experimentation, highlighting how different configurations of central guidance and local initiative jointly support policy learning and institutional refinement.
In practice, the study employs policy text analysis and case study research, taking Guangdong Province's Comprehensive Innovation Reform Pilot Zone (which participated in two rounds of national reform since its designation in May 2015) as a typical longitudinal case. Guangdong has treated the pilot as a leading project for implementing the innovation-driven development strategy and has achieved notable results in areas such as intellectual property protection, the commercialization of scientific and technological achievements, and scitech finance. Through an examination of major policy documents, the study both validates the proposed “W+N” dual-type model of iterative innovation in policy pilot schemes and identifies a fourstage pattern in the evolution of policy pilots, namely incubation, experimentation, evaluation, and promotion. These stages correspond, in broad terms, to the formation and authorization of pilot schemes, the localized implementation and adjustment of reform measures, the assessment and feedbackbased revision of pilot outcomes, and the subsequent diffusion of mature and replicable policy arrangements. Across the two rounds of comprehensive innovation reform, the iterative mechanism of Guangdong’s policy pilot schemes gradually shifted from W-type dominance to the N-type pattern, reflecting the increasingly prominent role of local governments in policy innovation.
The research findings reveal that (1) Guangdong's two-round reform demonstrates a shift from the Wtype to the N-type iteration. Initially, the central government provided specific measures; subsequently, local governments autonomously designed proposals through a "bidding" mechanism. The combination of these two paths can enhance the effectiveness of policy pilots and fully leverage the creativity of local governments. (2) Policy pilots exhibit distinct stage characteristics, with the iterative process successively progressing through the stages of incubation, experimentation, evaluation, and promotion. Each stage presents different characteristics in terms of the interaction mode between central and local governments, the direction of resource flow and knowledge iteration. (3) Success depends heavily on organizational coordination and monitoring-evaluation capabilities, with professional talent teams as key to execution and innovation. This study contributes by linking policy process theory with Chinese pilot analysis, revealing the internal relationship between goal clarity, central–local interaction and iterative pathway choice. It provides a practical framework for designing pilot schemes, optimizing intergovernmental coordination and converting pilot experience into scalable reforms, offering implications for deepening innovation reform and advancing governance modernization.

China has long adhered to a single-track administrative model for patent confirmation of rights, and the reform of this model has always been a hot topic in academic circles. Various scholars have proposed four typical reform paths. Existing research also points out the 'problems' in China′s current single-track administrative model for patent validation, primarily the cumbersome procedures and their negative impact on the efficiency of infringement litigation, which serve as the impetus for reform. The aforementioned problems all point to the law and economics issues of litigation benefits and efficiency. From the perspective of law and economics, the Coase Theorem, which is standardized, considers the comparison of plans before and after the reform. The plans after the reform should have lower costs and higher benefits compared to those before the reform. From a law and economics perspective, the normative interpretation of the Coase Theorem provides a benchmark for evaluating pre- and post-reform institutional arrangements. Specifically, the post-reform regime should yield higher net benefits (i.e., lower costs relative to benefits) than the pre-reform one. By comparing the paths before and after the reform, it is found that the plans after the reform do not reduce the costs of trial or improve efficiency; instead, they may even lead to increased costs and decreased efficiency. Through the analysis of the costs of reform and the benefits after the reform, it is evident that either the cost of reform is extremely high and immeasurable, making it impossible to compare with the benefits after the reform, or the benefits after the reform are insufficient to compensate for the costs of reform.
In long-term practice, China′s administrative and judicial organs have closely cooperated to effectively achieve accurate and efficient trials of patent validity. They have continuously explored the matching and connection with patent infringement procedures, forming valuable Chinese experience and establishing a "single-track administrative path for patent confirmation of rights with Chinese characteristics" in practice. Its supporting mechanisms enable effective procedural coordination, and the existing three-tier process, alongside efficiency-enhancing measures, already meets current needs. The establishment and evolution of this model are aimed at reducing social costs and improving the efficiency of enterprises or judicial administration,and there is no necessity to alter the scope of review authority or reduce the number of trial levels. Therefore, this model which is grounded in China’s national conditions and shaped by its practical experience should be preserved and refined incrementally to address existing shortcomings.
The optimization directions for China′s patent rights confirmation model should include three aspects: First, after accepting patent infringement litigation cases, the People′s Courts should file requests for prioritized examination of patent invalidation with the China National Intellectual Property Administration (CNIPA) for patents with uncertain stability. Regarding such requests initiated by the People′s Courts, if CNIPA has already accepted an invalidation request for the patent involved in the case, it should apply the prioritized examination procedure for the review; If the CNIPA has not yet accepted an invalidation request regarding the patent in question, it shall first place the case on record. Upon subsequently accepting an invalidation request filed by a relevant party, the CNIPA shall directly apply the prioritized examination procedure to the review. Second, the Patent Examination Guidelines should clearly specify the time limits for the ordinary procedure of patent invalidation, the prioritized examination procedure, and the retrial procedure after a court revokes an invalidation decision, and the time limit for the prioritized examination of patent invalidation should be set at four months. Third, in the review decisions of patent invalidation declarations, a separate section should be dedicated to a detailed description of the interpretation of patent claims maintained as valid after review. Particularly for claims where the patentee has narrowed the scope using the specification, the narrowed scope should be explicitly recorded.

With the in-depth development of globalization and digitalization, companies face increasingly complex and dynamic competitive environments. Knowledge, as the core resource of modern enterprises, has rendered the innovation of its management model crucial to sustainable development. Technological innovations not only greatly enhance the efficiency of knowledge acquisition, transformation, and application but also drive knowledge management toward intelligence, dynamism, and collaboration. As a result, how companies can drive the innovation of knowledge management models has become a pressing issue for both academia and industry. Existing literature mainly studies knowledge management model innovation from a linear perspective and case studies, but it fails to clarify the causal differences of multiple factors in the coordinated and unified process of achieving knowledge management model innovation, nor does it reveal the driving pathways of knowledge management model innovation. Therefore, exploring the antecedent variables and driving pathways of knowledge management model innovation is of practical value. This paper seeks to answer the following questions: How are the antecedent conditions of knowledge management model innovation interlinked? Which condition among the antecedent factors contributes the most to knowledge management model innovation?
This study adopts actor-network theory (ANT) to construct a research framework for knowledge management model innovation. Taking 202 valid questionnaire responses from high-tech enterprises as the research data, it employs fuzzy-set qualitative comparative analysis (fsQCA), importance-performance map analysis (IPMA), and artificial neural network (ANN) analysis to examine the impact of factors at different levels on enterprises' knowledge management model innovation and the configurational effects among these factors.The FsQCA identifies conditional combinations between variables and reveals different causal paths, making it suitable for exploring multiple causal relationships in complex systems. IPMA helps identify important influencing factors through visual means and quantitatively analyzes the actual performance of these factors. Through training on vast amounts of data, ANNs are capable of capturing non-linear relationships between variables, thereby enabling the prediction of future trends and outcomes.This combination not only enhances the depth of causal analysis, but also improves prediction accuracy, making it suitable for multidimensional research on complex systems.
The results of the study show that (1)a single condition cannot fully explain the driving mechanisms of knowledge management model innovation;(2) six antecedent conditions from the levels of technological actors, human actors, and coordinating actors have a synergistic impact on knowledge management model innovation,the paths leading to complementary high-level knowledge management model innovations can be categorized into three types: technology-coordination dual-core, human-machine synergistic interaction, and technology-led self-driven;(3) AI + knowledge application and the digital capabilities of organizational members can substitute for each other under certain conditions;(4) the key preconditions for knowledge management model innovation mainly lie at the technological actor level, with AI + knowledge transformation being the core condition for achieving high-level knowledge management model innovation.
The contributions of this study are as follows: First, compared to the previous method of linear path analysis, this study identifies multiple implementation paths for enterprises to achieve high-level knowledge management model innovation under different configuration conditions from a configuration perspective, revealing the driving mechanism of multi factor interaction coupling on knowledge management model innovation. Second, different from the TOE framework commonly used in configuration research, the introduction of ANT integrates technical actors, human actors, and collaborative actors into the same analysis network, with a network logic proposed for innovative construction of knowledge management models under human-machine collaboration, providing a new exploration path for the localized application of ANT in management contexts. Finally, this study proposes that technology actors, human actors, and collaborative actors jointly constitute the core driving force for knowledge management model innovation, and further analyzes the substitutability of technology actors and human actors in specific configurations, responding to the lack of in-depth exploration of human-machine collaboration in existing research.

The cornerstone of corporate innovation rests on knowledge collaboration among enterprises, a process that hinges on the accumulation, transformation, and joint creation of core knowledge resources across firms. In the digital economy landscape, platform governance, by fostering the aggregation effect of knowledge flow, incentivizes enterprises to construct open and shared knowledge collaboration frameworks through platform mechanisms. While serving as an effective facilitator for stakeholder interactions, platform governance can exert both positive and negative influences on knowledge collaboration among enterprises. Hence, facilitating such collaboration within platform contexts emerges as a critical and urgent issue.Existing research predominantly delves into the independent impacts of diverse governance models on knowledge collaboration, often overlooking their combined effects. This limitation hinders a comprehensive understanding of how platform governance mechanisms operate. This study, adopting a matching perspective, zeroes in on two quintessential platform governance models—"relational governance" and "contractual governance"—to probe their impacts on knowledge collaboration. It aims to analyze the independent and combined influences of these governance models on knowledge collaboration among enterprises participating in the platform,and examine the pivotal moderating roles played by platform attributes.〖HJ*3〗
Drawing on transaction cost theory and social capital theory, this paper employs nonlinear regression and response surface analysis techniques to dissect how relational governance, contractual governance, and their degree of match affect knowledge collaboration. The research focuses on regions with relatively developed platform economies, including Jiangsu, Shandong, and others, and analyzes a dataset comprising 879 samples using statistical tools like SPSS 27.0 and AMOS 28.0.Key findings include six parts: (1) Relational governance exerts a significantly positive impact on knowledge collaboration, fostering trust and cooperation among enterprises. (2) Contractual governance demonstrates a positive U-shaped relationship with knowledge collaboration, indicating that its effectiveness may increase beyond a certain threshold. (3) Greater consistency between relational and contractual governance enhances knowledge collaboration, suggesting a synergistic effect. (4) The "strong relational-strong contractual" governance model outperforms the "weak relational-weak contractual" model in promoting knowledge collaboration. (5) Interestingly, the "strong relational-weak contractual" governance model yields higher levels of knowledge collaboration compared to the "weak relational-strong contractual" model, highlighting the importance of relational elements. (6) Platform strategic consensus positively moderates the impact of governance consistency on knowledge collaboration, emphasizing the role of shared vision and goals.
In practical terms, innovation-oriented platforms should strive to balance relational and contractual governance, leveraging their respective strengths to foster a collaborative environment. Transaction-oriented platforms, on the other hand, should focus on the cost-efficiency benefits of contractual governance while ensuring adequate relational elements to maintain trust. Enterprises with high external knowledge acquisition needs should prioritize joining platforms that employ a hybrid of both governance models to maximize collaboration opportunities.
This study makes several notable contributions.(1) It clarifies the nonlinear effects of platform relational and contractual governance on knowledge collaboration, addressing gaps in traditional theories by providing a more nuanced understanding of their interplay. (2) By constructing a two-dimensional matrix, it explores the combined mechanisms of governance models on knowledge collaboration, offering a fresh perspective on how these models interact to drive collaboration. (3) It extends strategic consensus research to platform contexts, revealing its unique value in facilitating cross-organizational knowledge collaboration and examining the moderating role of platform openness, thereby broadening the scope of platform strategic consensus and openness research.
The study provides key managerial insights: First, platforms should flexibly choose governance models based on their core objectives (innovation or transaction), while enterprises need to dynamically adjust strategies to adapt to different governance impacts. Second, both platforms and enterprises should prioritize building strategic consensus, incorporating it into cooperation criteria and innovation decisions to enhance collaboration awareness and alignment. Third, improving collaboration efficiency through governance matching and consensus building is crucial, with platforms guiding and enterprises actively adapting to create a stable and efficient ecosystem for innovation and development. However, the study has limitations, and future research should explore more diverse governance models and other influencing factors to enrich understanding.