Opening the black box of AI-enabled green innovation: a PLS-SEM and fsQCA study of manufacturing firms

The manufacturing industry has shifted dramatically due to the digital transformation and the vision for sustainable growth. Digital technologies have been at the forefront of this transformation through artificial intelligence (AI), which enables businesses to improve their ability to innovate, increase operational efficiency, and respond to both the competitive forces of globalization and the need to be environmentally responsible. The use of AI within manufacturing represents a paradigm shift rather than simply a technological update. This new way of thinking about creating, delivering, and capturing value in increasingly complex and unpredictable markets is being adopted by manufacturers (Batwara et al., 2025; Sharma et al., 2025).

Manufacturing firms undergoing digital transformation face increasing expectations from stakeholders that their growth aligns with sustainability objectives. This alignment is particularly relevant to “green innovation,” which encompasses both Circular Business Model Innovation (innovations in the reuse, recycling, and regeneration of resources) and the development as well as deployment of green technologies (environmentally-friendly technologies). The use of AI to support or enhance these innovative processes may provide an opportunity for the creation of new economic value propositions through the reduction of waste and the maximization of resource utilization (Batwara et al., 2025; Yin et al., 2025; Alghamdi, 2023).

AI in Indonesia is currently changing the way goods are manufactured through the integration of new technologies, including computer vision models (e.g., object detection and quality inspection), IoT systems for real-time monitoring, and big data analysis in manufacturing firms. The use of these advanced technologies has led to an increase in productivity, better products, and lower production costs. However, there are still some obstacles that need to be addressed. Some of the main obstacles include infrastructure issues, the high cost of implementing AI technology, and the lack of skilled professionals. Therefore, there is a need to develop the skills required in the job market and to create low-cost, locally developed AI solutions that will meet the needs of both large-scale manufacturers and small-to medium-sized enterprises (MSMEs) (Bangun et al., 2025; Oktavian and Rachmadi, 2023; Rahutomo et al., 2019).

The role of AI Capabilities in enhancing the manufacturing firms' capacity for innovation has been established across the sector. AI Capabilities represent more than simply technical tools. Instead, they represent organizational abilities to connect data, analytics, digital infrastructure, knowledge, and managerial routines to enhance a firm's capacity to pursue sustainability-oriented innovations. Prior research portrays AI Capabilities as an important strategic capability by which manufacturers can identify opportunities, process information, and transform their sustainability goals into new products, processes, and business models rather than treating sustainability as a separate compliance process. AI Capabilities are associated with numerous implementations of green innovation, including Circular Business Model Innovation and Sustainability-Oriented Corporate Performance in manufacturing firms (Renfei and Zhongwen, 2026), green innovation and creativity in manufacturing-oriented SMEs (Alwakid and Dahri, 2025), and stronger sustainable performance when combined with AI-oriented leadership and frugal innovation (Ahmad et al., 2025).

AI Capabilities are shown to help companies develop, refine, and scale new products and services related to sustainability, particularly circular business models, green product or process improvements, and resource-efficient operations. Several studies have found that AI Capabilities for green innovation lead to increased sustainability performance. For example, Sun et al. (2025a, b) demonstrate how AI Capabilities strengthen the positive relationship between a firm's boundary spanning search and its green innovation. Furthermore, Al Koliby et al. (2025) demonstrate how a firm's green AI capability enhances the positive relationship between a firm's green knowledge orientation and its green innovation. Finally, some research has linked AI Capabilities with frugal innovation (Ahmad et al., 2025), disruptive innovation and circular economy outcomes (Al Halbusi et al., 2025a, b), and circular supply-chain practices through innovation ambidexterity and Green Leadership (Hou et al., 2026).

The previous literature has shown that AI Capabilities do not automatically lead to sustainable innovation. AI Capabilities have the potential to be a major driver of sustainability-related innovation in manufacturing; however, this only occurs when AI Capabilities are supported by leadership, strategies, organizational flexibility, and external institutional support. For instance, Renfei and Zhongwen (2026) emphasize that dynamic sustainable capabilities and managerial cognition are critical. Hou et al. (2025) and Ahmad et al. (2025) mention that leadership is an important tool for converting AI resources into sustainability results. Further, Al Halbusi et al. (2025a, b) find that organizational change capacity and financial stability enhance the sustainability payoffs from AI-enabled disruptive innovations. Zhao and Liu (2025) show that the likelihood of AI supporting substantial green innovations was increased by environmental regulations and market competition driving companies towards tangible results rather than superficial results. In addition, Qu and Kim (2025) further emphasize that the adoption of sustainable AI in the manufacturing sector will depend upon the level of support provided by other firms, universities, and industry associations.

In order to achieve the full potential of AI in manufacturing, AI Capabilities will be a critical factor. Although many studies have been conducted, research gaps remain with respect to the role of Green Leadership and the context-dependent nature of the impacts of AI on green innovation. For the development of theory and practice that support sustainable digital manufacturing, addressing these gaps will be essential. In particular, there is limited understanding of how AI Capabilities can lead to green innovation in manufacturing firms. This limitation is evident in several areas. First, there is a lack of clarity regarding the connections between AI Capabilities and different types of green innovation (e.g. Madanaguli et al., 2024; Mikalef and Gupta, 2021). Second, the influence of leadership on AI-enabled green innovation remains underexplored (e.g. Madanaguli et al., 2024; Giraud et al., 2023). Third, there is insufficient evidence on the business value of AI-enabled green innovation (e.g. Madanaguli et al., 2024; Ångström et al., 2023). These gaps indicate that the relationship between AI Capabilities and green innovation remains a “black box”. This study is designed to open the “black box” by disaggregating green innovation into Circular Business Model Innovation and Green Technology Innovation, and then positioning Green Leadership as a mediating mechanism that converts AI Capabilities and their components (Perceptive, Predictive, and Prescriptive Capabilities) into green innovation.

Therefore, the primary purpose of this study is to examine the effect of AI Capabilities on two types of green innovation (Circular Business Model Innovation and Green Technology Innovation) and the role of “Green Leadership” since AI Capabilities have the potential to become powerful enablers of green innovation, and also to explain the impact of those two types of green innovation on Sustainability-Oriented Corporate Performance. To accomplish the research objectives, the study identifies a number of critical questions. These are whether the capabilities of artificial intelligence (AI) contribute to the development of two different forms of green innovation, Circular Business Model Innovation and Green Technology Innovation, whether Green Leadership has a major contribution to the two forms of innovation, and whether both forms of innovation will positively influence corporate performance related to sustainability. In addressing these issues, the research intends to identify the mechanisms through which AI Capabilities serve as drivers of sustainable innovation outcomes. The research incorporates the conceptual framework of dynamic capabilities and sustainable innovation, and integrates empirical evidence of AI Capabilities in green innovation research (Alghamdi, 2023; Renfei and Zhongwen, 2026; Yin et al., 2025).

In doing so, this research refines and elevates its theoretical positioning by bridging a critical gap at the intersection of strategic management and digital sustainability. While classical NRBV posits that green capabilities generate sustained competitive advantage (Hart, 1995), the precise mechanisms to deploy these resources in the digital era remain undertheorized. By opening the “black box” of AI, this study demonstrates that AI-driven predictive and prescriptive capabilities act as catalysts that operationalize dynamic capabilities (Teece, 2018). This interaction ensures that Green Leadership is not just a passive strategic posture, but an active driver of substantive, resource-efficient environmental innovations, thus advancing contemporary extensions of the natural resource-based paradigm (Hart and Dowell, 2011; Hou et al., 2026; Sun et al., 2025a, b).

Therefore, this research has two contributions. First, it opens the “black box” of AI-enabled green innovation by explaining the mechanism of Circular Business Model Innovation and Green Technology Innovation development with AI Capabilities as the antecedent. This effort extends the boundary conditions of the dynamic capabilities perspective from managerial orchestration (Teece, 2018) to semi-autonomous or even autonomous orchestration and refines the Natural Resource-Based View (NRBV) in clarifying which environmental strategies are suitable in the autonomous domain and which remain leadership-dependent (Hart and Dowell, 2011). Second, this research also identifies the digital microfoundations of dynamic capabilities in the AI-enabled green innovation context by aligning Perceptive, Predictive, and Prescriptive Capabilities with the sensing, seizing, and reconfiguring routines (Sjödin et al., 2023; Teece, 2007). Furthermore, in the NRBV domain (Pappas and Woodside, 2021), Prescriptive Capabilities constitute the most important configuration for AI-enabled green innovation, and Green Leadership plays a quasi-necessary role for Green Technology Innovation.

The theoretical framework of this research is deeply rooted in the convergence of the Natural Resource-Based View (NRBV) and the Dynamic Capabilities perspective. The NRBV posits that a firm's long-term competitive advantage in contemporary markets is fundamentally constrained by its relationship with the natural environment, requiring strategic capabilities such as pollution prevention, product stewardship, and clean technologies (Hart, 1995; Hart and Dowell, 2011). However, in highly volatile and digitalized landscapes, deploying these green resources effectively requires dynamic capabilities, defined as the firm's ability to integrate, build, and reconfigure internal and external competences, to address rapidly changing environments (Teece, 2018). Dynamic Capabilities theory is built upon the resource-based view (RBV) and expands it by providing a theoretical lens for understanding how firms can create new capabilities to sense opportunities, seize opportunities, and gradually reconfigure and renew their current resources to remain competitive in changing environments (Silva et al., 2025; Teece, 2018; Ambrosini and Bowman, 2009). This study bridges both lenses by arguing that advanced digital infrastructures do not act in isolation; rather, they serve as the operational engines that translate environmental aspirations into concrete strategic outcomes.

To contextualize these findings, we move beyond treating technology and management as parallel streams by defining how AI Capabilities specifically “reconfigure” Green Leadership itself. Unlike traditional organizational learning, which relies on historical, slow-paced, and cognitively bounded human experience (Eisenhardt and Martin, 2000), AI's predictive and prescriptive microfoundations—specifically through digital sensing, seizing, and transforming (Teece, 2018)—provide real-time, data-driven foresight. This algorithmic intelligence actively reconfigures the strategic posture of green leaders, transforming intuitive environmental intentions into precise, agile, and systemic eco-innovative routines. This aligns with the strategic notion that corporate technological capabilities act as primary antecedents that trigger organizational agility and dynamic reconfigurations (Martín-Rojas et al., 2026). By mathematically and structurally augmenting the cognitive capacity of environmental leadership (Hart and Dowell, 2011), AI Capabilities systematically dismantle organizational inertia, unlocking complex paths toward circular business models and technological eco-innovations.

The manufacturing industry has widely adopted AI in recent years. The previous works have mentioned how AI is used in manufacturing within various aspects beyond automation such as scheduling, quality control, maintenance, process optimization, and sustainability improvement. Zeba et al. (2021) explain how AI is being used in the context of smart manufacturing by means of cyber-physical systems, deep learning, big data, and real time scheduling. Bag et al. (2021) indicate that AI could enhance supply chain performance, sustainable manufacturing practices, and circular economy capability within automotive manufacturing. Lăzăroiu et al. (2022) describe “cognitive manufacturing” wherein AI supports IoT-based real-time production logistics, deep learning-assisted smart process planning, and predictive maintenance. Senoner et al. (2022) illustrate an example from semiconductor manufacturing where explainable AI identified factors that influence process quality and improve production yields. Finally, Vadivel et al. (2025) demonstrate the application of AI-based computer vision technology in jute manufacturing whereby a CNN can detect fabric defects such as stains and tears more effectively than manual inspections.

As AI rapidly evolves as a transformative technology, there is a need to better understand how companies can leverage AI Capabilities. These capabilities can support the development and implementation of dynamic capabilities (Haverila et al., 2025). They also help firms recognize and capitalize on emerging opportunities in highly competitive and rapidly changing markets (Leemann and Kanbach, 2022; Liu et al., 2024a, b; Roy et al., 2025). When applying dynamic capabilities theory to AI Capabilities, AI can be seen as a unique strategic resource for creating value in an organization's activities and operations. In addition, AI improves an organization's ability to respond to environmental changes, such as technological shifts and market changes. This is achieved by enhancing analytical capabilities, improving decision-making, and strengthening organizational learning. In addition, leveraging AI as part of organizations' strategies enables them to effectively recognize environmental changes and take advantage of emerging opportunities and transform their business models as they transition from traditional business formats into a strengthened digital business presence in a more competitive landscape (Chen et al., 2025; Roy et al., 2025; Liu et al., 2024a, b).

AI Capabilities are used throughout the manufacturing industry in many different ways. As an example, a study by Sjödin et al. (2021) describes three primary capabilities that support the scaling of AI for manufacturers to redesign their business models. These include data-pipeline capability, AI algorithm development, and AI democratization. Zhong and Um (2026) find that AI Capabilities enhance both novelty and efficiency in designing new business models when manufacturers use customer and supplier integration in supply chain management. Additionally, Ratanacharoenchai and Jantapoon (2026) highlight the importance of using AI-enabled sensing, seizing, and reconfiguring capabilities as part of building stronger resilience and sustainable systems for small-to medium-sized enterprises (SMEs). On the operational side, Abou-Foul et al. (2023) indicate that having AI Capabilities improves internal processes and resource utilization through better servitization. Also, Sharma and Dang (2026) note improvements in terms of increased accuracy, faster time-to-market, reduced costs, improved product quality, productivity, and efficiency resulting from the adoption of AI into various manufacturing processes. Finally, in terms of sustainability-oriented research efforts, Ruangkanjanases et al. (2025) find that generative AI capabilities can lead to enhanced organizational creativity and green innovation ambidexterity. Additionally, Chung (2022) emphasizes that AI in manufacturing will depend on domain knowledge-based human capital which means that instead of relying on data science skills alone, it will require combining AI Capabilities with deep industrial expertise.

The typical example of Circular Business Model Innovation in the manufacturing sector is Xerox's product-service offering with photocopier products where they use an approach to design their business models and practices so that they do not solely focus on selling new products but also include a process that uses a model built around remanufacturing and reusing photocopier products from customers who have sent back their used copiers (Linder and Williander, 2017). In this case, it is circular in nature since the manufacturer generates revenue by increasing the length of time each product is used, reducing its total amount of materials wasted, and lowering the manufacturers' overall environmental footprint while continuing to generate financial gains due to a service-based approach versus traditional “sell-and-dispose” linear methods. Therefore, Circular Business Model Innovation within the manufacturing sector can be defined as a strategic transformation in which companies transition away from the traditional linear “take-make-dispose” model of thinking and instead create innovative ways to deliver value creation, delivery, and recovery throughout the entire product lifecycle.

The move towards a circular economy for manufacturing is broader than merely recycling and reducing waste; instead, it involves changing how manufacturing firms interact with their customers, their supply chain, their internal capabilities, and their products' designs. Linder and Williander (2017) also suggest that by utilizing reuse- and remanufacturing-based circular business models, there are opportunities for cost savings as well as environmental benefits. However, they have yet to be widely adopted due to the associated risk and significant upfront investment. Circular transition in manufacturing also depends on the levels of readiness across these four areas as identified by Chirumalla et al. (2024) including External Partners, Customer and Business Model Alignment, Organizational Culture/Internal Capabilities, and Design/Product Development.

The literature has shown that AI Capabilities are an enabler for companies to develop their capability to redesign business models in a way that supports circular economy principles. A study by Renfei and Zhongwen (2026) demonstrates empirically how companies can benefit from using AI-driven capabilities to enhance the speed at which they make circular transitions and therefore increase sustainable outcomes. Furthermore, Wang et al. (2026) state that AI provides companies with the means to generate value, exchange value, and recapture value through increased efficiencies in resource utilization, reduced waste, improved recycling opportunities, lower costs, and new forms of value creation. All of these enablement mechanisms are important for transitioning to Circular Business Model Innovation. Additionally, Montes-Pineda and Garrido-Yserte (2024) note that companies developing their AI Capabilities will be better positioned to support circular business models through the development of solutions including reducing waste, reusing materials, recycling products, increasing product life, enhancing the company's supply chain, and increasing its ability to predict customer needs. The overall consensus from these studies indicates that stronger AI Capabilities will provide manufacturing firms with the necessary intellectual capital, flexibility, and innovative capacity required to create more successful Circular Business Model Innovation. Therefore, hypothesis 1 is proposed as follows:

The examples of Green Technology Innovation in the manufacturing industry are the implementation of industrial robots to promote research and development in green technologies in the world's manufacturing sector (Lee et al., 2022), the utilization of digital products and digital manufacturing methods to enhance the ability of a firm to innovate through the green processes (Yin et al., 2024), the Haier COSMOPlat platform, which is an ecosystem that uses digitalized business models to develop green innovation potential through better interaction and exchange of knowledge among its members (Li et al., 2025), smart product platforms applied in automotive manufacturing companies, such as smart factories to enable flexibility, intelligence, and net-zero-oriented operations (Zhang et al., 2024), and finally the application of Artificial Intelligence in the manufacturing industries' processes to improve their Green Technology Innovation performance thanks to better knowledge combination (Tian et al., 2023).

Manufacturing firms use Green Technology Innovation as an opportunity to develop a competitive advantage in their industry while also promoting long-term sustainability objectives. The previous literature indicates that manufacturing firms will only implement Green Innovation if they have an environment in which the organization has the ability to develop and sustain its innovative capabilities. For example, Peng et al. (2020) find that Green Technology Innovation was a key area of development for achieving Sustainable Development Goals for the manufacturing sector. Lee et al. (2022) indicated that industrial robots could play a significant role in stimulating Green Technology Innovation through Green R&D Investment, but only under favorable Environmental Regulation and Industry 4.0 conditions. Likewise, Yin et al. (2024) as well as Liu et al. (2024a, b) demonstrate that digital transformation enhances Green Technology Innovation by providing organizations with better resource allocation, information transparency, market expectations, and R&D investment. Others build on these findings by demonstrating how Business-Model Digitalization (Li et al., 2025), Artificial Intelligence (Tian et al., 2023), ESG Performance (Jinglin et al., 2024), Sustainable Finance (Zhu et al., 2025), and Government Procurement (Li, 2025) can serve as significant precursors to encourage manufacturing companies to pursue Greener Technological Upgrades. AI Capabilities help manufacturing firms develop sensing capabilities and the ability to process and act upon data-based information for the development of clean products, processes, and solutions for sustainability problems. Kumar et al. (2025), as an example, find that AI Capabilities help the creation of green technologies. Likewise, Sun et al. (2025a, b) find that AI Capabilities provide technical assistance throughout all phases of the innovation process, and it improves the relationship between searching for knowledge and creating new green ideas in China's manufacturing sector. Cassânego et al. (2025) report that AI Capabilities facilitate green product and process innovation. Additionally, Zhang et al. (2025a, b) demonstrate that AI Capabilities enhance green innovation through the absorption of new knowledge. Also, Al Halbusi et al. (2025a, b) show that there is a direct impact of strong AI Capabilities on the development of green innovation. In manufacturing environments specifically, Ruangkanjanases et al. (2025) indicate that AI Capabilities enabled resources aid in both explorative and exploitative approaches to green innovation. These studies demonstrate that organizations with strong AI Capabilities experience improved outcomes associated with green innovation due to their enhanced knowledge processing, absorption capacity and innovation implementation. Therefore, hypothesis 2 is proposed as follows:

Green Leadership in manufacturing can be defined as a way of leading that is designed to bring people, strategy, innovation, and operations into alignment with sustainable environmental goals. In doing so, manufacturing firms improve both their environmental performance and their sustainable performance. Green Leadership has been studied in the previous literature in several forms. For example, green transformational leadership provides a stimulus for new green product and process innovation (Ahsan, 2025; Begum et al., 2022), green inclusive leadership promotes pro-environmental behavior and supports environmental strategies (Rafiq and Xiuqing, 2025), green servant leadership enhances tacit knowledge sharing on green issues and green innovation (Hu'an et al., 2025), and green knowledge-oriented leadership links leadership with environmental strategy and green innovation (Al Koliby et al., 2025). Across these studies, the common theme is that Green Leadership in manufacturing is not just about compliance but it is about leaders actively creating an environmentally responsible and sustainable organizational culture, employee behavior, capacity for innovation, and decision-making outcomes.

In the recent literature, AI Capabilities are mentioned to improve a manufacturing firm's ability to identify and respond to its external environment, help to interpret, analyze, and use sustainability-related data to make better decisions, and develop new ways of doing business (new forms of operations). Specifically, in the green manufacturing context, AI Capabilities reconfigure the firm by reshaping the leaders' action and transforming the managerial cognition on which Green Leadership rests (Renfei and Zhongwen, 2026; Sjödin et al., 2023). Unlike dynamic capabilities in traditional organizational learning, where the reconfiguration is incremental and path-dependent in altering what the firm knows (Teece, 2007), the reconfiguration driven by AI Capabilities is real-time and forward-looking in altering how the firm decides (Ibrahim et al., 2026). As a result, manufacturing firms will be able to adopt “greener” operations through AI Capabilities, allowing their leaders to communicate, advocate, and require greater emphasis on environmental stewardship.

Hou et al. (2026) demonstrate that AI-based capabilities enable a relationship between Green Leadership and Innovation Ambidexterity which enables manufacturing companies to pursue sustainability as a goal. They also indicate that Digital Capabilities represent an important means by which manufacturing companies may enact Sustainability-Focused Leadership. Ahmad et al. (2025) similarly find that companies with AI Capabilities were more successful at using those capabilities to achieve sustainable results because their leaders had access to AI resources and were able to create an organizational climate conducive to creativity and innovation. This indicates that a higher level of AI capability has the potential to increase a leader's orientation toward sustainability. Lastly, Al Koliby et al. (2025) demonstrate that Green AI Capability is positively associated with the effect of Green Knowledge-Oriented Leadership on Green Innovation in Manufacturing SMEs, indicating that manufacturing firms' adoption of AI-based capabilities will enhance their leader's ability to advocate for environmental strategy and green practices. Therefore, hypothesis 3 is proposed as follows:

The leaders of firms who hold high environmental values will likely incorporate sustainable practices into their strategic plans. They will also encourage the firm to responsibly manage resources and generate innovative business models that redesign how companies produce, deliver, and capture value from their products or services in more circular ways. As Sukasmanto and Jaya (2025) point out, Green Leadership has been related to the efficient use of resources, as well as social innovation for environmental protection in terms of a circular economic development strategy. Green Leadership was shown by Hou et al. (2026) to have a positive influence on circular practices and enhance Circular Business Model Innovation in manufacturing environments through its positive effect on sustainability-oriented innovation. Furthermore, Green Leadership was demonstrated by Sun et al. (2025a, b) to increase knowledge sharing, digital innovation, and ultimately increase the adoption of the circular economy. Additionally, Bashynska et al. (2024), indicate that Green Leadership creates an organization-wide sustainable culture and supports ethically based decisions which facilitate circular transformations. Consequently, it can be posited that in manufacturing firms, Green Leadership can positively affect Circular Business Model Innovation through the promotion of organizational commitments, abilities, and behaviors that support the creation of circular values. Thus, hypothesis 4 is proposed as follows:

Transformational leaders concerned about the environment create a strategic climate that supports their firms' investments, experiments, and adoption of clean technologies. Specifically, according to research by Waqas et al. (2024), there is a positive and statistically significant relationship between green transformational leadership and Green Technology Innovation. This relationship is mediated by green innovation capabilities and serves as a mechanism through which leadership influences innovation outcomes. Hameed et al. (2023) also demonstrate how leadership characterized as having high levels of ethics and environmental orientation could be used to develop green organizational resources and consequently lead to increases in Green Technology Innovation. This has led to the conclusion that when companies are able to establish leadership-oriented green development, it will assist them in establishing both strategic guidance and company-wide support (Qu et al., 2026). Therefore, because environmental pressures, production efficiency, and regulatory compliance are fundamental issues in manufacturing organizations, Green Leadership will probably encourage the development and use of green technologies. Thus, hypothesis 5 is proposed as follows:

Circular Business Model Innovation enables firms to deliver sustainable development by redesigning value creation, delivery, and capture around circular economy principles such as resource recovery, reuse, and lifecycle extension in an integrated way. Circular Business Models have been shown to provide both economic and social benefits by improving environmental performance and enhancing resource security (Mendoza et al., 2022), and to support improved economic value and environmental performance (Kuzma and Sehnem, 2023). The literature also supports the notion that organizations which implement circular economy-based innovation and sustainable practices will be able to recover resources, reduce their environmental footprint, and deliver greater economic and social value than those who do not adopt these practices. Additionally, the relationship between the implementation of Circular Business Model Innovation and other forms of technological innovation (Islam et al., 2024) suggests that organizations implementing strong Circular Business Model Innovation are better positioned to produce sustainability-related results. The most direct evidence of this relationship exists in the fact that Circular Business Model Innovation positively affects Sustainability-Oriented Corporate Performance in manufacturing firms (Renfei and Zhongwen, 2026). Therefore, hypothesis 6 is proposed as follows:

The implementation of Green Technology Innovation provides the firm with an opportunity to transform its product, process, or system of production to be less wasteful, produce fewer emissions, require fewer resources while increasing competitive advantage and creating long-term value. According to Mukhtar et al. (2025), as Green Technology Innovation is adopted by manufacturing firms in Malaysia, it enhances sustainability performance from an economic perspective, environmental perspective and social perspective. In addition, as stated by Masoudi and Shahin (2025), Green Technology Innovation has a very strong direct relationship to the triple bottom line. Furthermore, the research indicates that Green Technology Innovation also has a direct positive effect on sustainability performance and serves as a means for managerial practices to result in sustainable outcomes. As noted by Wang et al. (2023), Green Technology Innovation may provide the manufacturing company with substantial improvements in its ability to provide better environmental and economic performance. Finally, as indicated by Jie and Jiahui (2023), Green Technology Innovation is critical to industrial enterprises wishing to develop in a sustainable manner. These studies support the idea that if a manufacturing company chooses to utilize Green Technology Innovation, it is much more likely to be able to obtain superior Sustainability-Oriented Corporate Performance. Therefore, hypothesis 7 is proposed as follows:

The overall research model is presented in Figure 1.

In this context, researchers conduct an empirical analysis using quantitative methods by distributing online questionnaires to respondents who are specifically targeted for the survey, manufacturing firms within industrial estates in Indonesia. A purposive sampling approach was conducted by targeting respondents with sufficient knowledge of their firm's technology adoption, leadership, and sustainability practices (Etikan et al., 2016). There were 101 valid participants among the total respondents in the survey. The respondent's profile is presented in Table 1. For the PLS path model (Hair et al., 2016), the researchers calculated the appropriate sample size (minimum = 90 participants) to achieve a statistical power of 80%, which can identify R-squared values of 0.10 or greater (significance level = 0.05) from each of the three independent variables examined within each structural model (Valliant et al., 2018). Nonprobability sampling techniques were utilized because they allow the characteristics of the sample to vary in proportion to how those characteristics would appear in a randomly selected sample from the target population (Valliant et al., 2018). Nonprobability samples were utilized to provide quantifiable results and enable statistical analysis of the data.

In this research, the construct measurements were developed in a quantitative research setting based on the variable definitions discussed in Section 2 as well as validated measurements from previous studies relevant to AI and Green Innovation implementation in the manufacturing context as described in Table 1. There are 5 higher-order constructs and 11 lower-order constructs (components) from the research model as presented in Figure 1. AI Capabilities are reflected by Perceptive Capabilities (continuous monitoring capabilities to identify hidden patterns and anomalies), Predictive Capabilities (the abilities to anticipate and respond to environmental changes based on multidimensional data for efficient resource utilization purposes), and Prescriptive Capabilities (running simulations capabilities to determine the best actions for maximizing sustainable value). Circular Business Model Innovation is reflected by Augmented Business Model (business model to drive additional cyclical value) and Automated Business Model (business models to generate considerable economic and sustainable benefits). Green Technology Innovation is reflected by Green Product Innovation (produce eco-friendly products designed to reduce resource consumption and pollution within the lifecycle), Green Process Innovation (process to minimize emissions through alternative materials and technological upgrades), and End-of-Line Management Innovation (optimize the treatment of “three industrial wastes”). Lastly, Sustainability-Oriented Corporate Performance is reflected by Environmental Performance, Social Performance, and Economic Performance.

This research applied a two-stage analytical framework that combines both Partial Least Squares Structural Equation Modeling (PLS-SEM) and Fuzzy-Set Qualitative Comparative Analysis (fsQCA), to examine the theoretical model. PLS-SEM was used to test the hypothesized net-effect relations in the conceptual model at the level of higher-order constructs (Hair et al., 2016). Hypotheses were tested through non-parametric bootstrapping with 5,000 subsamples using SmartPLS 4, which reports path coefficients and p-values, with a threshold value of less than 0.05 as the criterion for a supported hypothesis. fsQCA, on the other hand, was used to find different combinations of AI Capabilities and Green Leadership lower-order constructs (components) that lead to both Circular Business Model Innovation and Green Technology Innovation. In this regard, the use of both methods provides a complementary approach as it allows PLS-SEM to analyze symmetrical and average effects within the sample, while fsQCA captures equifinality, asymmetric causation, and configurational causality (Pappas and Woodside, 2021).

The result of this research shows that the minimum R-square is 0.353. Meanwhile, the minimum respondents for PLS-SEM method for statistical power of 80% with 5% significance level, two maximum number of arrows pointing at a construct, as well as structural model and minimum R-square above 0.10 is 90 (Hair et al., 2016). Furthermore, the sample sufficiency was also verified through post-hoc power analysis from G*Power 3.1 software with the 101 samples used in this study, a 5% significance level, the smallest f² = 0.131, and two predictors. The achieved statistical power at N = 101 is 90%, exceeding the recommended 80% benchmark. Therefore, the collected sample data with 101 manufacturing firms as respondents are sufficient.

The result of measurement models in this research fulfills all the criteria that Hair et al. (2012) mention as rules of thumb for model evaluation in reflective measurement models. The result of the composite reliability (CR) was above 0.7 (internal consistency reliability), indicator loading factors was above 0.7 (indicator reliability) and average variance extracted (AVE) was above 0.5 (convergent validity). The AVE was also greater than the construct's higher squared correlation between the associated construct and other construct (discriminant validity). The CR and AVE value of each measurement indicators are presented in the Table 2.

In this research, discriminant validity was determined by three criteria. First, the ratio of heterotrait-monotrait (HTMT) should not exceed 0.9 (Henseler et al., 2015). Second, the number of bootstrap samples should exceed 5,000. Third, there should be an acceptable indicator weight and loading (the p-value should be < 0.05), and finally, multicollinearity should be ruled out with the variance inflation factor (VIF) being below five for each indicator. Finally, the use of at least 5,000 samples will allow the researchers to assess the significance of the path coefficients through bootstrapping (the p-value should be < 0.05 or the confidence level is 95%). The number of cases must be similar to that of the original sample of observations.

The findings from the structural model testing using PLS-SEM show that AI Capabilities affect the development of Circular Business Model Innovation and Green Technology Innovation differently, as presented in Figure 2. AI Capabilities directly develop Circular Business Model Innovation. On the other hand, AI Capabilities do not have a direct effect on the development of Green Technology Innovation; instead, AI Capabilities affect the development of Green Technology Innovation through the mediation of Green Leadership. Green Leadership itself has a direct effect on both Circular Business Model Innovation and Green Technology Innovation. Thus, it shows the importance of Green Leadership in connecting the AI Capabilities to Green Innovation in manufacturing firms within industrial estates. Meanwhile, both types of Green Innovation, Circular Business Model Innovation and Green Technology Innovation, have a positive effect on Sustainability-Oriented Corporate Performance. The results of all hypothesis tests are presented in Table 3.

For the robustness check, this study applied the Gaussian copula approach to the full latent path model to exclude endogeneity bias that may occur from self-selection of high performing green firms into AI and Leadership investment. The Shapiro-Wilk test confirmed the non-normality of AI Capabilities and Green Leadership scores (p < 0.01) that fulfill the identification requirement (Becker et al., 2022). Copula terms added one at a time (Hult et al., 2018) were insignificant in the Circular Business Model innovation equation (c_AIC: p = 0.130; c_GL: p = 0.062), the Green Technology Innovation equation (c_AIC: p = 0.278; c_GL: p = 0.315), and the Green Leadership equation (c_AIC: p = 0.540), while all hypothesized paths retained their direction and significance. Thus, the results reflect that endogeneity does not occur and that the reported estimates are robust (Sarstedt et al., 2020).

This asymmetry reflects the different nature of the two AI-enabled Green Innovation outcomes. Circular Business Model Innovation involves the reconfiguration of the manufacturing firm's value architecture, which includes information-intensive and optimization tasks that align with the components of AI Capabilities (Perceptive, Predictive, and Prescriptive Capabilities). In this context, these capabilities can act natively as direct enablers of green outcomes without a leadership intermediary (Chi and Vu, 2026; Gupta et al., 2026). On the other hand, Green Technology Innovation embeds physical products and processes that require R&D and capital allocation, where Artificial Intelligence may provide knowledge spillovers but cannot be authorized on its own without Green Leadership (Boonlua et al., 2026; Al Koliby et al., 2025).

This study also conducted the fsQCA analysis that focuses on the components of AI Capabilities, including Perceptive Capabilities, Predictive Capabilities, and Prescriptive Capabilities, as well as Green Leadership, as predictors of both Circular Business Model Innovation and Green Technology Innovation. The calibration values used in the fsQCA analysis are presented in Table 4. The Likert-scale scores of each construct were adjusted into fuzzy-set membership scores by using three qualitative anchors based on the sample distribution, including the 95th percentile for full membership, the 50th percentile for the crossover point, and the 5th percentile for non-membership (Pappas and Woodside, 2021). The Necessary Condition Analysis in Table 5 shows that high values of both Circular Business Model Innovation and Green Technology Innovation can be achieved when all components of AI Capabilities and Green Leadership are present, as shown by a consistency level above 0.9.

In conducting sufficiency analysis, this research applies a raw consistency threshold of 0.80, a frequency threshold of 1 for moderate sample size, and an overall solution consistency above 0.75 (Pappas and Woodside, 2021). The consistency reflects the degree to which cases sharing a configuration exhibit the outcome, whereas coverage indicates the empirical relevance of each configuration in explaining the outcome. All of the configurations in Table 6 and Table 7 exceed the threshold and support the robustness of the configurational findings. There are four similar configurations predicting high values of both Circular Business Model Innovation and Green Technology Innovation as presented in Table 6. Configurations IIIA and IIIB show that high values can be achieved when all components of AI Capabilities are present. On the other hand, when Predictive Capabilities and Prescriptive Capabilities are absent, high values of Circular Business Model Innovation and Green Technology Innovation can still be achieved when either Perceptive Capabilities or Green Leadership is present (Configurations IA, IB, IIA, IIB). The presence of Prescriptive Capabilities can yield high values of Circular Business Model Innovation and Green Technology Innovation even when the other components of AI Capabilities and Green Leadership are absent (Configurations IVA, IVB).

The importance of Green Leadership can be seen in the configuration to predict the low value of Circular Business Model Innovation and Green Technology Innovation as presented in Table 7. The absence of Green Leadership can result in low values of Green Technology Innovation even when all components of AI Capabilities are present (Configuration IIIB). Furthermore, the absence of Green Leadership can also yield low values of both types of Green Innovation when Perceptive Capabilities and Predictive Capabilities are also absent (Configurations IIA, IIB) or when Predictive Capabilities and Prescriptive Capabilities are absent at the same time (Configurations IA, IB).

The fsQCA findings show that the four sufficient solutions in Table 6 represent equifinal pathways, including comprehensive AI orchestration, prescriptive autonomy, leadership substitution, and sensing-driven pathways, which correspond to different combinations of the sensing, seizing, and reconfiguring microfoundations of dynamic capabilities (Sjödin et al., 2023; Pappas and Woodside, 2021; Teece, 2007). The negation analysis in Table 7 shows that there is causal asymmetry: low Circular Business Model Innovation results from the absence of the AI Capabilities base, whereas low Green Technology Innovation occurs even when all AI Capabilities components are present but Green Leadership is absent (Configuration IIIB; consistency = 0.924). These findings support the PLS-SEM results for the research framework presented in Figure 2 by reflecting the direct effect of AI Capabilities on Circular Business Model Innovation and the quasi-necessary role of Green Leadership for Green Technology Innovation, which converges with the PLS-SEM finding of the full mediation of Green Leadership. The configurational evidence describes the boundary between the autonomous orchestration domain, where AI can act natively on Circular Business Model Innovation, and the Green Leadership-dependent domain of Green Technology Innovation within the Natural Resource-Based View (Boonlua et al., 2026; Chi and Vu, 2026; Hart and Dowell, 2011).

As Prescriptive Capabilities are shown in Table 6 to be the most critical single component of AI Capabilities that can yield high Circular Business Model Innovation and Green Technology Innovation even when the other components of AI Capabilities and Green Leadership are absent, managers of manufacturing firms need to operationalize its three measurement indicators presented in Table 2 (Renfei and Zhongwen, 2026), as elaborated in the managerial implications section.

The manufacturing industry is experiencing a significant transformation as a result of two major influences: Digitalization and Sustainability. In addition to creating new ways for firms to generate value through digital technologies, artificial intelligence (AI) technologies such as machine learning, automation, and data analytics are being incorporated into all aspects of the manufacturing process. At the same time, the manufacturing industry is under increasing pressure from environmental and societal concerns, requiring manufacturers to transition toward sustainable products, green technologies, and green innovations, as well as to adopt circular business models. This study integrates the existing literature to examine the extent to which AI technologies, combined with Green Leadership, can drive green and circular innovation, thereby enhancing Sustainability-Oriented Corporate Performance among manufacturing organizations (Al Koliby et al., 2025; Khan et al., 2025; Ledi et al., 2024; Potharaju et al., 2026).

Circular Business Model Innovation and Green Technology Innovation are two key mechanisms through which manufacturing companies can improve their sustainability performance. Both Circular Business Model Innovation and Green Technology Innovation have been empirically linked to corporate sustainability performance, supporting prior research on Circular Business Models (Renfei and Zhongwen, 2026; Zaidan et al., 2024; Motke et al., 2022), as well as Green Technology Innovation (Koh et al., 2026; Kumar et al., 2025; Zhao and Jin, 2025). The effectiveness of this relationship is further strengthened by factors such as a firm's organizational preparedness, top management commitment, regulatory pressure, and pro-ecological culture. The results are theoretically grounded in two complementary frameworks: the Natural Resource-Based View (NRBV) and Dynamic Capability Theory (Al-Koliby et al., 2025; Potharaju et al., 2026; Ledi et al., 2024; Teece, 2018), both of which emphasize that the development of Green Leadership capabilities is critical for achieving sustainable transitions in manufacturing industries leveraging digital technologies. The NRBV states that companies can create long-term competitive advantages when they develop capabilities that allow them to perform better environmentally. The NRBV suggests that AI Capabilities act as dynamic capabilities because they can transform AI-based resources into green innovations. Advanced Analytics and Automation are examples of AI Capabilities in manufacturing and do not produce green innovation outcomes on their own. Rather, these AI Capabilities contribute to green innovation outcomes through Green Leadership, which serves as an organizational catalyst for change and innovation. Green Leadership also contributes to improving a company's environmental strategy, enhancing its organizational culture to become more environmentally conscious, and facilitating the exchange of information among employees (Al Koliby et al., 2025; Khan et al., 2025; Ledi et al., 2024), all of which are necessary for successful green innovation. The Dynamic Capability Theory also explains how firms can sense, seize, and reconfigure resources, such as AI and leadership, to successfully innovate under sustainability challenges (Siddique et al., 2026; Al Koliby et al., 2025; Badar and Siddiquei, 2025; Teece, 2018).

AI Capabilities enhance both green knowledge and Green Leadership to increase the effectiveness of an organization's green innovative activities. The empirical data support this view with regard to both a company's economic and environmental outcomes (Keskin et al., 2025; Al Koliby et al., 2025; Zhang et al., 2025a, b). In terms of the type of innovation involved, both Circular Business Model Innovation and Green Technology Innovation have a significant positive effect on a firm's long-term, sustainability-oriented performance. The primary focus of Circular Business Model Innovation is to create long-term competitive advantages by minimizing waste and optimizing the utilization of materials (Krassnitzer et al., 2025; Potharaju et al., 2026). Green Technology Innovation has the greatest influence on short-term improvements in environmental and economic performance through increased efficiency in products and processes (Ye et al., 2022; Hussain et al., 2025).

This study examined whether AI Capabilities in manufacturing organizations are positively associated with two forms of green innovation, Circular Business Model Innovation and indirectly with Green Technology Innovation, and evaluated how Green Leadership influences the relationship between AI Capabilities and both types of Green Innovation, while also evaluating their collective effects on the sustainable performance of manufacturing firms. In terms of our first question, we found that AI Capabilities have a direct positive effect on Circular Business Model Innovation, while their effect on Green Technology Innovation operates fully through Green Leadership; specifically, AI Capabilities enhance a firm's ability to sense changes in its environment, learn from those changes, and reconfigure its resources to capitalize on new opportunities. We also determined that adopting AI-driven technologies does more than improve operational efficiency. It also facilitates the development of innovative green processes. These benefits were particularly evident in highly technological-oriented industries and dynamic environments. Additionally, our results indicate that Green Leadership plays a critical role in stimulating innovation in firms. Our research conceptualizes Green Leadership as encompassing knowledge-oriented and entrepreneurial characteristics. As such, green leaders are better able to develop environmental strategies and promote the integration of AI-driven innovations. Furthermore, because they could align organizational goals with environmental objectives, green leaders strengthen the impact of green innovation on sustainable performance. Overall, our results support the importance of effective Green Leadership in successfully leveraging AI Capabilities to achieve meaningful transformation in manufacturing firms. Circular Business Model Innovation and Green Technology Innovation are positively related to Sustainability-Oriented Corporate Performance. Both forms of innovation contribute to improving resource efficiency, reducing environmental impact, and promoting long-term growth. The integration of Circular Economy Practices highlights the importance of adopting holistic approaches that combine innovation and sustainability strategies.

To clearly delineate the precise theoretical novelty of our model, this study moves beyond the familiar, linear boundaries that often restrict both the Dynamic Capabilities perspective and the NRBV. The true originality of this research lies in its dual conceptual-methodological architecture. While mainstream sustainability literature generally assumes symmetric, one-size-fits-all pathways to eco-innovation, our model conceptualizes and tests how digital dynamic capabilities (AI) and environmental strategy (Green Leadership) operate through asymmetric, equifinal configurations. By combining PLS-SEM and fsQCA, we do not merely replicate established frameworks; we expose the real causal complexity and the diverse recipes needed to activate clean technologies and circular designs. This shifts the academic conversation away from simply asking ‘if’ these strategic perspectives interact, focusing instead on uncovering the specific, non-linear configurations through which digital microfoundations reshape corporate environmental capabilities in the smart manufacturing era.

The theoretical contributions of the current study are as follows: it contributes to both the Dynamic Capability Theory (Teece, 2018) and Natural Resource-Based View (NRBV) (Wang et al., 2026). It demonstrates how AI Capabilities can be combined with Green Leadership to drive green innovation and sustainability. Thus, this study provides an integrative framework for facilitating sustainability transitions across industries. We advance both lenses by illustrating how emerging artificial intelligence capabilities transmute static green resources into dynamic environmental outcomes. Specifically, we clarify how data-driven capabilities renew the firm's asset orchestration (Teece, 2018), reinforcing the perspective that corporate technology and knowledge management support the continuous renewal of organizational capabilities needed to navigate complex environments (Martín-Rojas et al., 2026). This provides empirical evidence on how digitalized dynamic capabilities are essential to deploy the clean technologies conceptualized by the NRBV framework (Hart and Dowell, 2011). The positioning of this study relative to the existing literature is presented in Table 8.

The results of this study show that Artificial Intelligence Capabilities and Green Leadership are two of the most important factors driving both Circular Business Model Innovation and Green Technology Innovation in the pursuit of enhancing Sustainability-Oriented Corporate Performance. The most important components of AI Capabilities are Perceptive Capabilities and Prescriptive Capabilities, which alone can produce high values of both Circular Business Model Innovation and Green Technology Innovation. In addition, Green Leadership is also important in this context, as it alone can yield both Circular Business Model Innovation and Green Technology Innovation even when all AI Capabilities components are absent. Conversely, the absence of Green Leadership can cause low values of Green Technology Innovation even when all AI Capabilities components are present.

Furthermore, these findings offer critical insights by contrasting this successful synergy against the performance of firms that fall into what could be termed the “AI adoption trap”. Organizations that rush to invest heavily in advanced artificial intelligence without a corresponding development of a Green Leadership culture often suffer from structural misalignments. In these firms, AI remains a decoupled technological asset. This aligns with the strategic notion that digital and technological tools act as primary antecedents that trigger organizational agility, but their capacity to foster environmental outcomes is entirely neutralized without a core leadership philosophy capable of directing data toward sustainability (Martín-Rojas et al., 2026). When contrasting this with literature on eco-innovation, we find that technology alone cannot overcome organizational inertia; complex environmental innovations require specific managerial capabilities to orchestrate external and internal knowledge effectively (De Marchi, 2012). Adopting cutting-edge technology without the strategic alignment of green leaders merely creates a veneer of digital modernization, trapping the firm in high capital expenditures with zero circular economy returns (Wong et al., 2020).

Since the findings of this study show that the most important Dynamic Capabilities microfoundation in AI-enabled green innovation is Prescriptive Capabilities, manufacturing firms need to adopt several measures. First, for real-time problem solving, as operational layer, manufacturing firms may combine the sensors with explainable AI (XAI) that enable the operators to adopt prescriptive recommend-and-act loop in standard daily operation (Sjödin et al., 2023). Second, for proactive risk management, as anticipatory layer, the firms may implement early warning systems based on the predictive models on enterprise-risk cases that also can conduct pre-stages mitigations (Tian et al., 2023; Li et al., 2025). Lastly, for augmented decision making with AI prescriptions, as strategic layer, the manufacturing firms may build decision-support cockpit for the leaders that presents prescribed options with explainable rationale (Renfei and Zhongwen, 2026).

Managerially, these results warn against the “AI adoption trap” discussed above: firms that acquire Predictive and Prescriptive Capabilities without a robust Green Leadership culture fall into an illusion of green digitalization, leaving AI Capabilities underutilized and failing to bridge the gap between digital capacity and the resource-efficient demands of cleaner production (Chen et al., 2006; Ghisetti et al., 2015; Martín-Rojas et al., 2026).

Therefore, the results of this study indicate that companies should prioritize the adoption of AI technologies to improve their green innovation abilities and ensure that their investments in AI align with circular economy principles. Furthermore, leadership development programs should focus on cultivating green leaders who can effectively drive environmental initiatives and innovation. In the coming years, researchers will need to evaluate how AI-driven green innovations influence both environmental outcomes and financial performance across diverse industries and geographic contexts.

The limitations of this research relate to the geographical context of Indonesian industrial estates, which may affect the generalizability of the findings. Indonesia has an emerging market status marked by institutional void where market supporting institutions are weak or missing (Khanna and Palepu, 2010). This situation has an impact on AI strategies since AI Capabilities require adequate infrastructure, talent, and financing, which are unequally distributed in emerging markets (Bangun et al., 2025; Kshetri, 2020). Industrial estates partly offset those voids through streamlined governance and shared infrastructure; furthermore, some industrial estates have special economic zone status with various government incentives (Aritenang and Chandramidi, 2020). Thus, the results of this study may not generalize to resource-constrained manufacturing firms outside industrial estates. Future research may adopt an institution-based view that examines regulatory stringency, industrial estate-level institutional completeness, as well as AI-governance maturity (Mhlanga, 2021; Peng et al., 2008). In addition, future research can also help advance understanding of how environmental regulations and other external factors shape the relationship between AI, green innovation, and sustainability. In addition, an analysis of the relationships among these emerging technologies (e.g., AI and the Internet of Things [IoT]) and Green Leadership particularly under conditions of high environmental and market volatility, may provide deeper insights into achieving sustainability objectives.

The authors express their sincere gratitude to Universitas Indonesia for their generous financial support.