Introduction

The rapid advancement of Artificial Intelligence (AI) technology has triggered a global transformation across multiple industries, driving economic growth and reshaping innovation ecosystems(Babina et al., 2024; Polak and Anshari, 2024). In the manufacturing sector, AI plays an increasingly pivotal role, particularly within AI-enabled manufacturing clusters—geographically concentrated networks where AI technologies are deeply embedded within manufacturing enterprises, resource infrastructures, and collaborative institutions (Wan et al., 2020). These clusters serve as innovation hubs, facilitating knowledge exchange, resource optimization, and collective learning, thereby accelerating manufacturing transformation and enhancing regional competitiveness (Cappiello et al., 2020).

Extensive literature examines the role of industrial clusters in fostering innovation, highlighting key drivers such as cluster resources, inter-firm networks, and regulatory environments (Lai et al., 2014; Qiu et al., 2024; Speldekamp et al., 2020; Yuan and hai Zhang, 2020). Michael Porter’s cluster theory emphasizes how clusters enhance productivity through shared resources, stimulate innovation via dynamic local networks, and facilitate the emergence of new businesses, thereby contributing to regional economic vitality and sustainability (Connell et al., 2014; Porter, 1998). However, the impact of clusters on firm growth is multifaceted. Critics, including Yasuyuki Motoyama, argue that Porter’s theory is overly descriptive and difficult to replicate in practice, due to its insufficient focus on cluster origins and public sector involvement (Motoyama, 2008). Peter Sunley further critiques the theory’s conceptual vagueness, which complicates its application and measurement. He underscores the necessity of considering factors such as culture, institutions, and market dynamics in understanding the success of clusters (Martin and Sunley, 2003).

Despite these critiques, the literature on technological innovation within industrial clusters provides valuable insights into the potential of AI to transform manufacturing processes. Research has shown that AI adoption can lead to substantial improvements in efficiency, quality control, and decision-making (Hao and Demir, 2023; Putra et al., 2024; Quispe et al., 2023).

Nevertheless, while previous studies have investigated the innovation dynamics within traditional industrial clusters, the specific interactions of these driving factors within AI-enabled manufacturing clusters remain underexplored.

One critical gap in current research is the collective dynamics of AI-driven innovation within manufacturing clusters, particularly the role of inter-firm networks in facilitating such innovations. AI innovation in these clusters is shaped by a complex interplay of economic, technological, and social factors, encountering both challenges and opportunities throughout its evolution (Lazzeretti et al., 2023). While literature acknowledges AI’s potential to enhance efficiency (Waltersmann et al., 2021; Wan et al., 2020), improve quality control (Arinez et al., 2020; Escobar et al., 2023), and optimize decision-making (Chien et al., 2020; González-Castañé et al., 2022), insufficient insights exist regarding how cluster resources, inter-firm networks, and regulatory environments collectively influence the emergence and diffusion of AI-driven manufacturing innovations. Addressing this gap requires a comprehensive exploration of how these factors interact and shape AI diffusion patterns.

Recent years have seen the rise of simulation-based methodologies, such as Cellular Automata (CA), for studying these complex interactions (Gregorio and Serra, 1999; Hegselmann and Flache, 1998). CA models effectively capture the dynamic characteristics of resource sharing, collaboration, and competition within clusters, offering a novel computational perspective on how AI innovations diffuse in manufacturing ecosystems (Kotyrba et al., 2015; Shan et al., 2024).

This study seeks to bridge this gap by explicitly examining the evolutionary mechanisms and driving forces of AI-enabled manufacturing clusters. Grounded in evolutionary economic geography theory and complex systems theory, we develop a theoretical framework and construct a Cellular Automata-based simulation model to investigate how cluster resources, inter-firm networks, and cluster environments collectively shape the adoption and diffusion of AI technologies within manufacturing clusters.

The findings of this research contribute to a deeper understanding of clustered innovation ecosystems and the theoretical foundations of collective learning and competitive advantage in the context of AI. Additionally, this study provides practical insights for policymakers and industry leaders, offering targeted interventions and strategic planning recommendations to enhance the development of AI-enabled manufacturing clusters. By conceptualizing AI-enabled manufacturing clusters as complex adaptive systems, this research highlights the interplay between micro-level drivers (e.g., firm-level AI adoption, inter-firm collaboration) and macro-level outcomes (e.g., industrial transformation, regional economic growth). This systems-based approach paves the way for future research on AI-driven innovation ecosystems.

The remainder of this paper is structured as follows: section “Theoretical framework and evolutionary model” details the Cellular Automata model design; section “Simulation analysis” presents simulation outcomes and analysis; and section “Discussion and conclusion” discusses theoretical contributions and policy implications for optimizing AI cluster development.

Theoretical framework and evolutionary model

The theoretical framework of this study integrates two complementary theories: evolutionary economic geography theory and complex systems theory, with a particular focus on Cellular Automata (CA) modeling and simulation. This integrated framework seeks to offer a comprehensive understanding of the dynamics within clustered innovation ecosystems, while also establishing the theoretical foundations for collective learning and competitive advantage in the context of AI-driven manufacturing innovation.

Theory of evolutionary economic geography

Evolutionary economic geography is a theoretical approach that views the economic landscape as a complex adaptive system. It highlights the roles of competition, spatial environment, technological change, and historical time in shaping economic regions through processes such as variety, selection, retention, and the spatial dynamics of firms and industries. This theory synthesizes Darwinian principles with additional concepts, such as plasticity, robustness, and self-organization (Martin and Sunley, 2006).In this theory, the spatial concentration of related economic activities is seen as a key driver of innovation. The clustering of firms and industries facilitates innovation through mechanisms such as knowledge spillovers, collaborative learning, and the sharing of resources, all of which contribute to a regional competitive advantage. Within the context of AI-driven manufacturing innovations, this theory suggests that the resources within the cluster, inter-firm networks, and the regulatory environment of the cluster collectively influence the emergence and diffusion of innovations. Specifically, the proximity of firms within a cluster enables knowledge exchange and collaborative learning, while the regulatory environment plays a critical role in shaping or limiting incentive structures and knowledge-sharing practices.

Analysis of AI innovation driving force

In this study, we define manufacturing industry clusters that utilize AI technologies as geographically concentrated networks of enterprises, institutions, and resources that integrate artificial intelligence (AI) to foster growth and innovation within the manufacturing sector. These clusters consist of advanced manufacturing enterprises, research institutions, funding mechanisms, and collaborative networks, all dedicated to applying AI solutions aimed at optimizing production processes, streamlining supply chains, and driving product innovation(Lee et al., 2018; Rizvi et al., 2021; Zdravković et al., 2021). The defining characteristics of these AI-enabled manufacturing clusters include: the integration of AI technologies, such as predictive maintenance, smart manufacturing, automation, and data analytics, to enhance operational efficiency and productivity. Access to shared resources, which include specialized knowledge, skilled labor, digital infrastructure, and financial capital tailored to AI applications. Collaborative networks that facilitate knowledge exchange, technology adoption, and co-innovation among manufacturing firms, AI developers, and research institutions. Supportive environments, characterized by favorable policies, market dynamics, and socioeconomic factors, that encourage the adoption and scaling of AI technologies within manufacturing clusters (Haricha et al., 2023; Jagatheesaperumal et al., 2022; Wan et al., 2020).

Drawing on Evolutionary Economic Geography Theory and synthesizing existing research, this paper presents a comprehensive model outlining the primary drivers of artificial intelligence (AI) innovation and its subsequent evolution within manufacturing industry clusters. Figure 1 provides a visual representation of this model. As depicted in Fig. 1, the progress of AI innovation in manufacturing clusters is underpinned by three primary driving forces.

Fig. 1: Evolutionary model of AI innovation and driving forces of evolution in industrial clusters.
figure 1

This figure illustrates the theoretical framework of AI innovation evolution within manufacturing industry clusters. The diagram depicts the interplay among three primary driving forces: cluster resources, cluster networks, and cluster environments, as they influence the emergence and diffusion of AI-driven innovations.

Full size image

Cluster resources

Cluster resources, encompassing human capital, financial capital, digital infrastructure, R&D capabilities, and corporate culture, are vital for fostering AI innovation within manufacturing enterprises. This resource composition, including human capital, financial assets, and digital infrastructure, is visualized in Fig. 2, which illustrates the key components driving cluster innovation potential. Strategic Human Resource Planning (HRP) enhances the value of clusters by focusing on core talent, thereby boosting competitiveness (Shen, 2011). Capital resources impact industrial clusters by shaping market competition and boosting the vitality of financial enterprises through the development of financial industrial clusters (Ali et al., 2014). Digital resources contribute to industrial clusters by integrating companies into innovative cluster structures, which in turn create favorable environments for introducing digital economy elements in urban settings (Ivanenko et al., 2019). The construction of infrastructure positively influences industrial clusters by attracting firms, facilitating industrial transformation, enhancing regional innovation, driving economic growth, and supporting the transition to industrial centers. The effectiveness of this infrastructure depends on local conditions, such as the availability of skilled labor and capital (Wu et al., 2023). R&D capabilities are key to attracting new firms and enhancing both innovation and performance within industrial clusters (Leung and Sharma, 2021). Additionally, corporate culture plays a pivotal role by fostering employee engagement, driving innovation, and enhancing competitiveness. It stimulates entrepreneurship, shapes policy development, and facilitates the introduction and diffusion of technology (Putilova and Shutaleva, 2020).

Fig. 2: Manufacturing cluster resource master composition diagram.
figure 2

This figure presents the key components of cluster resources driving AI innovation in manufacturing clusters. The diagram shows elements such as human capital, financial capital, digital infrastructure, R&D capabilities, and corporate culture.

Full size image

In the context of AI-driven manufacturing innovations, the abundance of resources within clusters positively influences the exploration of new innovations by firms operating within the cluster innovation ecosystem. According to evolutionary economic geography theory, the spatial concentration of related economic activities promotes innovation through mechanisms such as knowledge spillovers, collaborative learning, and the sharing of resources. As a result, firms located in resource-rich clusters are more inclined to pursue new AI-driven manufacturing innovations, thanks to the availability of specialized knowledge, skilled labor, and advanced technological infrastructure. It is evident that manufacturing industry clusters play a critical role in fostering the development of AI innovation.

Cluster network

The industrial cluster network represents a complex, localized system of interconnected enterprises, characterized by free-scale properties. This network fosters innovation, resource integration, and economic development through mechanisms such as corporate risk appetite, collaborative knowledge sharing, and strategic cooperation (Abhari et al., 2019). As depicted in Fig. 3, these components constitute the core framework of the cluster network. Within this context, the structure of the industrial cluster network significantly influences the efficiency of knowledge diffusion and innovation (Schilling and Phelps, 2007). Additionally, enterprise risk preferences shape behaviors and strategic choices within industrial clusters, influencing levels of cooperation and necessitating risk prevention strategies for sustainable development (Liu and Xu, 2018). Knowledge sharing within these clusters enhances economic performance, innovation, and competitiveness by reducing costs and risks, and is influenced by factors such as commitment and leadership (Wang et al., 2017). Furthermore, strategic cooperation within industrial clusters promotes firm clustering, supports industry formation, enhances regional innovation, and increases competitiveness. These strategies vary depending on industry type and regional conditions (Baldassarre et al., 2019).

Fig. 3: Manufacturing cluster network main composition diagram.
figure 3

This figure outlines the main components of inter-firm networks within manufacturing clusters. The diagram illustrates corporate risk appetite, collaborative knowledge sharing, and strategic cooperation.

Full size image

While geographic proximity offers advantages for resource pooling and interaction, it does not, by itself, guarantee the global competitiveness of firms within clusters (Lee, 2018). The ability to build and leverage international linkages is a critical factor for firms seeking access to global markets, advanced technologies, and diversified talent pools (Rugman et al., 2012). Moreover, the composition of clusters, characterized by similar and complementary firms, significantly influences their innovation dynamics. Competitive firms often drive each other toward greater efficiency and innovation, while complementary firms foster collaborative advancements by integrating expertise and resources across the value chain (Hermundsdottir and Aspelund, 2021; Maciel and Fischer, 2020). These interactions underscore the essential role of well-structured inter-firm networks in cultivating an environment where knowledge sharing serves as a catalyst for sustained innovation.

The strength of inter-firm networks positively influences the dissemination of AI-driven manufacturing innovations among firms within the clustered innovation ecosystem. Robust and adventurous inter-firm networks facilitate knowledge exchange, collaborative learning, and joint problem-solving elements critical to the diffusion and adoption of AI-driven manufacturing innovations. Firms embedded in well-connected networks can leverage these relationships to access valuable information and resources, thereby enhancing their capacity to propagate innovation (Aviv et al., 2019). Consequently, it is imperative to strengthen the establishment of cluster networks to promote the advancement of manufacturing industrial clusters and facilitate their transition to a more sophisticated stage.

Cluster environment

The industrial cluster environment encompasses a variety of factors, including the economy, policy, industry standards and specifications, and market demand, all of which positively influence talent growth within the clusters (Weng, 2008). These factors within the cluster environment are graphically depicted in Fig. 4. The economic environment plays a crucial role in the development of industrial clusters by fostering talent growth, contributing to economic expansion, enhancing the business environment, and integrating clusters into global networks (Narayana, 2014). State-supported policies can promote the growth and development of industrial clusters, leading to economic benefits and poverty alleviation (Shakib, 2020). Industry standards enhance the development of industrial clusters by improving competitive advantage, overcoming bottlenecks, and influencing economic development, innovation, and digital transformation (Li and Wu, 2016). Market demand affects the development of industrial clusters by lowering average costs and impacting regional economic growth, which can subsequently lead to increases in GDP, labor value, land and property prices, and environmental consequences (Wang, 2018).

Fig. 4: Manufacturing cluster environment main composition diagram.
figure 4

This figure depicts the environmental factors influencing AI innovation in manufacturing clusters. The diagram includes economic conditions, policy support, industry standards and specifications, and market demand.

Full size image

A supportive cluster environment positively influences collective learning among firms within the cluster’s innovation ecosystem, thereby contributing to regional competitive advantage. Such an environment shapes incentive structures and fosters a knowledge-sharing atmosphere that facilitates collaborative learning and innovation among firms. Policies and institutions that encourage cooperation, protect intellectual property rights, and provide funding for research and development can motivate firms to engage in collective learning activities, ultimately enhancing the cluster’s competitive advantage (Tallman et al., 2004). The effective utilization of resources and networking within manufacturing industry clusters relies on a favorable political and economic environment, which acts as a catalyst for the integration of artificial intelligence (AI) into the innovative evolution process within these clusters.

Complex systems theory and cellular automata modeling

Complex systems theory

Complex systems encompass a variety of phenomena in nature, characterized by interactions among multiple factors (Corning, 1995).In the realm of economics, complex systems theory emphasizes analysis from a connectivity perspective, concentrating on value creation through new connections among elements (Foster, 2005).

Complexity theory, an emerging field since the 1980s, builds upon traditional theories such as control theory, information theory, and systems theory, while also integrating newer theories including dissipative structure theory, synergetics, disaster theory, chaos theory, and super cycle theory. Society itself is a complex system, and social simulation is increasingly recognized as a vital method for understanding the dynamics of social development in the age of artificial intelligence. At the heart of social simulation lies the construction of ‘emergence’ mechanisms, which accurately depict and illustrate the processes and outcomes of social operations. The term ‘emergence’ refers to properties or capabilities of a system that are not inherent in its individual components (Hegselmann and Flache, 1998).

By utilizing emergent mechanisms to interpret complex social systems, social simulation elucidates intricate relationships across various systems and layers within society, as well as the resultant social processes. In the 1990s, sociologist Luhmann employed the concept of ‘system complexity’ to analyze complex social behavior, demonstrating that strong correlations among elements in a social system can lead to phenomena such as self-organization and self-generation. Social systems exemplify typical complex systems, and their evolutionary patterns are neither entirely fixed nor completely random; rather, they cluster within specific boundaries and ‘emerge’ when surpassing a critical threshold. Consequently, emergence is a fundamental characteristic of complex social systems. Social simulation can be integrated with big data and extensive datasets to transcend mere mechanical data aggregation and analysis, thereby revealing the complex properties and evolutionary patterns inherent in these systems.

Cellular automata modeling

In this study, we employ Cellular Automata (CA) as a modeling tool to simulate the dynamic and emergent behaviors within AI-enabled manufacturing clusters. First introduced by Von Neumann and Ulam in the 1950s (Crutchfield, 2011). CA has been widely used across fields such as economics, sociology, and ecology due to its ability to capture the interactions between simple components that give rise to complex system behaviors. For instance, in the context of Industry 4.0, Cellular Automata-based models have been utilized to optimize Big Data processing while minimizing energy consumption, as evidenced by the cost-effective MapReduce model proposed by Mitra (2021). This work simplified the complexity of existing CA rules, facilitating efficient data shuffling and integration within industrial processes (MITRA, 2021). In both computational research and industrial applications, Cellular Automata have been effectively employed to exploit their inherent parallelism for high-performance computing on modern platforms, such as multiprocessors and GPUs, yielding significant modeling outcomes (Wa̧s & Sirakoulis, 2015; Was and Sirakoulis, 2015).

A manufacturing industry cluster represents a complex social network, characterized by the intricacies of its evolutionary processes and the challenges posed by the lack of dynamic data for quantitative analysis. This makes it particularly suited for studying innovation ecosystems like manufacturing clusters, where localized interactions between firms can lead to significant collective outcomes. By utilizing CA, we can model how AI technologies diffuse through a cluster and predict the evolution of innovation within these ecosystems. However, the capacity of Cellular Automata (CA) to simulate these evolutionary processes through image descriptions, combined with the powerful functions of MATLAB, renders it a suitable tool for this analysis.

The Cellular Automata simulation model developed in this study is specifically tailored to capture the unique dynamics of AI-enabled manufacturing clusters. These clusters are defined as geographically concentrated networks in which AI technologies are deeply integrated into manufacturing enterprises, research institutions, and collaborative ecosystems. The model incorporates three primary dimensions- resources, networks, and environments- into a computational framework that simulates the adoption and evolution of AI-driven innovations.

Cluster resources are represented by parameters such as human capital, AI infrastructure, and R&D capacity, which collectively influence the integration of AI technologies and firm-level innovation. Collaborative networks are modeled as interactions among manufacturing firms, represented by neighboring cells on the Cellular Automata grid, which emulate AI-driven knowledge sharing, strategic alliances, and competitive dynamics. Lastly, cluster environments are captured through external factors such as government policies, market demand, and economic conditions, all of which shape the adoption of AI technologies and influence the collective behavior of firms within the cluster.

This Cellular Automata framework is particularly well-suited for modeling AI-enabled manufacturing clusters, as it captures emergent behaviors driven by localized interactions. For instance, firms equipped with advanced AI tools and significant resource endowments can stimulate innovation in neighboring firms through spillover effects. Furthermore, supportive policies and a favorable economic environment create conditions that encourage collective learning and risk-taking across the cluster. By simulating the interplay between these dimensions, the model offers a detailed understanding of how AI technologies drive innovation, optimize resources, and transform traditional manufacturing clusters into hubs of technological advancement.

By aligning the model parameters and rules with the defining characteristics of AI-enabled manufacturing clusters, this study provides both theoretical and practical contributions. The results yield actionable insights for policymakers and industry leaders, highlighting strategies to enhance cluster resources, strengthen inter-firm networks, and create supportive environments for AI integration. This approach not only advances theoretical understanding of AI-enabled manufacturing ecosystems but also serves as a guide for fostering industrial competitiveness and innovation in the age of artificial intelligence. The findings underscore the transformative potential of AI technologies in driving systemic innovation and regional economic development.

The Cellular Automata (CA) model developed in this study is structured as follows:

$${\text{A}}=({\text{d}},\,{\text{L}}\,{\rm{d}},\,{\text{N}},\,{\text{S}},\,{\text{F}})$$
(1)

In this model:

  • Cell (d): Represents an individual enterprise within the manufacturing industry cluster.

  • Cell Space (Ld): Defines the grid system, typically structured as n × n square cells, encompassing all enterprises in the industrial cluster.

  • Neighborhood (N): Utilizes the von Neumann configuration, where each cell interacts with its four immediate neighbors (north, south, east, and west). This configuration was selected to mirror localized interactions within manufacturing clusters, where enterprises are primarily influenced by geographically proximate counterparts. The von Neumann neighborhood has been widely employed in Cellular Automata studies for its computational efficiency and its ability to simplify complex systems while preserving spatially-adjacent interaction dynamics. For example, von Neumann’s cellular model demonstrates that localized relationships can effectively simulate broader system behaviors through finite-state automata, supporting its suitability for modeling industrial clusters (Mukhopadhyay, 1968).

  • Cell State Space (S): Defines the state variable St(i, j), representing the evolution strategy of the enterprise located at (i, j) at time t. The state space is binary:

    $${\mathcal{S}}=\{0,1\}$$
    (2)

    where S = 0 means the enterprise remains in its current state without adopting AI-driven innovation, and S = 1 means the enterprise enters the AI innovation cluster.

The evolution rules are defined as:

  • The evolution strategies of neighboring enterprises are observable.

  • Evolution decisions occur sequentially.

Evolution Dynamics

The probability of state transition is governed by several parameters:

  • Resource Ownership Coefficient (p1): Reflects the enterprise’s access to cluster resources (e.g., capital, talent, infrastructure). The more resources a firm has access to (e.g., talent, capital, infrastructure), the more likely it is to adopt AI technologies and influence others in the cluster. It follows a normal distribution:

    $${p}_{1} \sim N(\mu ,{\sigma }^{2}),\quad 0 < \mu < 1$$
    (3)

    where μ indicates the mean level of resource ownership.

  • Knowledge Sharing Coefficient (p2): Represents the degree of inter-firm knowledge exchange. Firms that are more connected within the cluster are more likely to engage in knowledge sharing, facilitating the spread of AI adoption. It follows a uniform distribution:

    $${p}_{2} \sim {\mathcal{U}}(0,r),\quad 0 < r < 1$$
    (4)

    where r reflects the affinity for network contact.

  • Cluster Network Capability: Calculated as:

    $${\text{Network}}\,{\text{Capability}}\,={p}_{2}\times \frac{N(t)}{M}$$
    (5)

    where N(t) is the number of AI-adopting enterprises in the neighborhood at time t, and M = 4 (total neighbors in the von Neumann configuration).The number of neighboring firms that have already adopted AI influences a firm’s decision to adopt AI itself.

  • External Environment Factor (e): Accounts for political, economic, and regulatory influences, which can either support or hinder the adoption of AI technologies within a cluster:

    $$e\in (0,1)$$
    (6)

The evolution probability for a cell to transition to S = 1 is defined as:

$$p=e\times \left({p}_{1}+{p}_{2}\times \frac{N(t)}{M}\right)$$
(7)

The state transition follows these rules:

$${S}^{t-1}(i,j)=0,\quad \,{\text{if}}\,{p}\,>\,{p}_{0}\ \Rightarrow \ {S}^{t}(i,j)=1$$
(8)
$${S}^{t-1}(i,j)=1,\quad \ \Rightarrow \ {S}^{t}(i,j)=1$$
(9)

where p0 = 0.6 is the threshold value, optimized for causal emergence effects.

For a full breakdown of model parameters and their distributions (e.g., μ, r, and e), refer to Table 1, which summarizes the parameter settings aligned with theoretical foundations and prior literature.

Table 1 Model parameters and their settings.
Full size table

Complex systems theory, grounded in Cellular Automata modeling and simulation, provides a robust analytical framework for capturing the intricate dynamics and nonlinear interactions among the components of the innovation ecosystem(Chopard et al., 2002). By simulating the behaviors and adaptation patterns of complex systems, such as innovation clusters, this theory helps predict the emergence and diffusion of AI-driven manufacturing innovations. Cellular Automata modeling enables the examination of the collective behavior of firms within a cluster, taking into account the complex interplay between cluster resources, inter-firm networks, and the cluster environment.

Integration of evolutionary economic geography theory and complex systems theory

By synthesizing the principles of evolutionary economic geography theory with complex systems theory, informed by Cellular Automata modeling, the proposed theoretical framework elucidates the intricate interplay between cluster resources, inter-firm networks, and the cluster environment, as well as the emergence and diffusion of AI-driven manufacturing innovations within industry clusters. The unit of analysis is the firm, which is embedded within the broader innovation ecosystem and engages in collaborative learning and competitive interactions with other firms. The theory encompasses a feedback loop of exploration and exploitation, wherein firms utilize cluster resources and inter-firm networks to pursue new innovations. Simultaneously, the cluster environment shapes the development of incentive structures and knowledge-sharing mechanisms that either facilitate or hinder such exploration.

Dialog with existing literature

The proposed theoretical framework builds upon and extends existing theories in evolutionary economic geography by incorporating insights from complex systems theory and Cellular Automata modeling. This integration allows for a more nuanced understanding of the dynamics within clustered innovation ecosystems, capturing the nonlinear interactions and emergent properties of the system. Consequently, the framework offers a comprehensive explanation of how cluster resources, inter-firm networks, and the cluster environment collectively influence the emergence and diffusion of AI-driven manufacturing innovations.

In conclusion, this theoretical framework serves as a valuable lens for examining the dynamics of clustered innovation ecosystems and the theoretical foundations of collective learning and competitive advantage in the context of AI-driven manufacturing innovation. By integrating evolutionary economic geography theory with complex systems theory, informed by Cellular Automata modeling, the framework enhances our understanding of the intricate interplay among cluster resources, inter-firm networks, and the cluster environment in shaping the emergence and diffusion of innovations. This framework can also guide future empirical research and inform policy formulation aimed at fostering innovation. AI-driven innovation clusters play a crucial role in enhancing regional competitiveness. The proposed agent-based model integrates these elements to simulate the dynamics of clustered innovation ecosystems, as well as the theoretical foundations of collective learning and competitive advantage in the context of AI-driven manufacturing innovation. This model investigates the complex interplay between cluster resources, inter-firm networks, and the cluster environment, alongside the exploration-exploitation feedback loop that propels the emergence and diffusion of AI-driven manufacturing innovations within industry clusters.

Simulation analysis

Initial state simulation of cellular automata

Based on the Cellular Automata (CA) model, a MATLAB simulation was conducted. The simulation parameters were set as follows: Cell space size: n = 20, meaning the cellular space consisted of a 20 × 20 grid of cells. Evolution threshold: threshold = 0.6, which was used to determine whether an enterprise (represented by a cell) would transition to an AI innovation state. Number of evolution steps: steps = 30, indicating the number of iterative processes in the simulation. The initial state of the cells was randomly generated using the randi function in MATLAB, with values of either 0 or 1. A value of 0 represented a firm that had yet to engage in the AI innovation evolution within the manufacturing industry cluster, while a value of 1 indicated a firm actively participating in the evolution. The initial state of the cells was randomly generated (see Fig. 5). Green cells represent firms actively participating in the evolution of AI innovation within the manufacturing industry cluster, while blue cells indicate firms that have yet to engage. Figure 5 illustrates that several firms within the manufacturing industry cluster are already involved in the evolution of AI innovation.

Fig. 5: Initial state of the cellular space.
figure 5

This figure shows the initial configuration of the 20x20 cellular space used in the Cellular Automata (CA) simulation. Green cells represent firms actively participating in AI innovation (state S = 1), while blue cells indicate firms not yet engaged in AI innovation (state S = 0).

Full size image

Influence of simulation parameters on the evolution of AI in industrial clusters

In the subsequent analysis, we focus on the influence of the related parameters μ, r, and e on the evolution of AI innovation clusters within the manufacturing industry. Using Cellular Automata theory, a quantitative analysis was performed on the evolution process of AI innovation in the manufacturing cluster. Figures 6, 8, and 10 depict the state of firms in the cellular space when two parameters are held constant while varying the value of another parameter, after conducting 30 simulations. Figures 7, 9, and 11 provide the corresponding quantitative representations for Figs. 6, 8, and 10, respectively, with the horizontal axis denoting the number of simulations and the vertical axis representing the number of AI-innovative firms in the manufacturing industry cluster.

Fig. 6: Evolution of AI innovation with varying resource ownership (μ).
figure 6

This figure illustrates the AI innovation evolution process in a manufacturing cluster under different resource ownership coefficients (μ = 0.3, 0.5, 0.7) with fixed r = 0.5 and e = 0.5. The first, second, and third images show the cellular space after 30 simulation steps for μ = 0.3, 0.5, and 0.7, respectively. Green cells represent AI-adopting firms (S = 1), and blue cells indicate non-adopting firms (S = 0).

Full size image
Fig. 7: Number of AI-Innovative firms with varying resource ownership (μ).
figure 7

This figure quantifies the number of AI-innovative firms in the manufacturing cluster over 30 simulation iterations for different resource ownership coefficients (μ = 0.3, 0.5, 0.7) with fixed r = 0.5 and e = 0.5. The plot shows three lines, each corresponding to a μ value, with the x-axis representing simulation steps and the y-axis indicating the cumulative number of firms with state S = 1.

Full size image
Fig. 8: Evolution of AI innovation with varying knowledge-sharing (r).
figure 8

This figure depicts the AI innovation evolution process in a manufacturing cluster under different knowledge-sharing coefficients (r = 0.4, 0.6, 0.8) with fixed μ = 0.5 and ε = 0.5. The first, second, and third images display the cellular space after 30 simulation steps for r = 0.4, 0.6, and 0.8, respectively. Green cells represent AI-adopting firms (S = 1), and blue cells indicate non-adopting firms (S = 0).

Full size image

Employing Cellular Automata theory, a quantitative analysis was performed on the evolution process of AI innovation in the manufacturing cluster. Figures 6, 8, and 10 depict the state of firms in the cellular space when two parameters are held constant while varying the value of another parameter, after conducting 30 simulations. Figures 7, 9, and 11 provide the corresponding quantitative representations for Figs. 6, 8, and 10, respectively, with the horizontal axis denoting the number of simulations and the vertical axis representing the number of AI-innovative firms in the manufacturing industry cluster.

Fig. 9: Number of AI-innovative firms with varying knowledge-sharing (r).
figure 9

This figure plots the number of AI-innovative firms in the manufacturing cluster over 30 simulation iterations for different knowledge-sharing coefficients (r = 0.4, 0.6, 0.8) with fixed μ = 0.5 and e = 0.5. The plot includes three lines, each corresponding to an r value, with the x-axis representing simulation steps and the y-axis showing the cumulative number of firms with state S = 1.

Full size image

Impact of cluster resources on AI-innovative manufacturing industry cluster

To examine the effect of cluster resources on the emergence and diffusion of AI-driven manufacturing innovations, a simulation analysis was performed. In this analysis, the values of r and e were fixed at 0.5, while the μ value was iterated to obtain various states and assess the occurrence of causal emergence. This approach serves as an approximate method for investigating the impact of cluster resources on the evolution of AI innovation within manufacturing industry clusters. The parameterμ represents the coefficient of firm resource ownership in the manufacturing industry cluster, which follows a normal distribution with a mean of μ. Manufacturing industry clusters typically possess an abundance of financial resources, human capital, data and technology resources, as well as AI infrastructure, alongside the research and development (R&D) capabilities necessary to leverage AI technology. If a firm fosters a corporate culture that prioritizes AI innovation, a higher μ value is anticipated. Larger μ values indicate a greater average availability of resources to firms within the cluster.

The specific simulation process is as follows:

First, an evolution loop was defined, where the length of μ values is denoted as k. For each μ value, the simulation was run for a number of steps. In each step, a new matrix was created to store the next state. Each cell was traversed to calculate the number of neighboring cells entering the innovation cluster, using the Von Neumann neighborhood method, which considers 4 neighbors.

Next, the value of p1 was calculated, following a normal distribution N(μ, σ2), using the current μ value. A uniformly distributed random number in the range [0, r] was generated as p2. The value of P was then calculated using the formula \(\scriptstyle{p}={e}\times ({p}_{1}+{p}_{2}\times \frac{N(t)}{M})\).

The state was updated based on the value of P. If all P-values were greater than p0 and the current cell state was 0, it was updated to 1. After updating the cell states, the final state was saved.

Through this simulation process, the final states for different μ values were obtained. The simulation results indicate that higher μ values, representing a higher average amount of resources available to firms, facilitate the acceleration of AI-driven innovation evolution within manufacturing industry clusters. Figure 6 illustrates the evolution of AI innovation across firms in the manufacturing cluster at varying resource ownership levels (μ) of 0.3, 0.5, and 0.7. The green cells indicate the firms that are actively involved in AI innovation, while the blue cells represent those that have not yet participated in the process. The findings demonstrate a clear positive relationship between the availability of cluster resources and the number of firms engaged in AI innovation. As the resource ownership coefficient (μ) increases, the number of firms adopting AI technologies within the cluster also rises.

As the resource ownership coefficient (μ) increases from 0.3 to 0.7, the number of AI-innovative firms continues to grow, reflecting the accelerating impact of resource-rich environments on AI adoption. At μ = 0.3, only a few firms engaged in AI innovation, while at μ = 0.7, a large majority of firms in the cluster had adopted AI, highlighting that higher resource availability enables widespread technological diffusion.

These results underscore the importance of resource abundance, comprising funds, skilled labor, advanced digital infrastructure, and R&D capabilities, in accelerating AI-driven innovation. Firms in resource-rich environments are more likely to leverage their technological capabilities to adopt AI, enhancing their competitive advantage. This trend illustrates that manufacturing clusters with substantial resources provide an environment conducive to rapid AI innovation, as firms are better equipped to absorb and deploy advanced technologies.

Figure 7 provides a quantitative representation of the relationship between increasing resource availability (μ values) and the number of firms participating in AI innovation within the manufacturing industry cluster. The horizontal axis represents the number of simulation iterations, while the vertical axis shows the cumulative number of firms engaging in AI-driven innovation over time.

For example, at μ = 0.3, only a limited number of firms-three-joined the AI innovation evolution, reflecting a scarcity of resources in the cluster. At μ = 0.5, around 30 additional firms participated in the process, suggesting that a moderate availability of resources facilitates broader adoption of AI. At μ = 0.7, nearly all firms in the cluster had adopted AI, highlighting that higher resource availability enables widespread technological diffusion.

This data confirms that clusters with more substantial resources, such as access to financial capital, skilled talent, and digital infrastructure, are better positioned to drive the evolution of AI technologies within manufacturing firms. The results reinforce the view that resource availability is a vital determinant in the speed and scale of AI innovation diffusion, with well-resourced clusters acting as accelerators for technological adoption and innovation diffusion across firms.

The findings from both figures highlight that cluster resources, including human capital, digital infrastructure, financial assets, and R&D capabilities, are fundamental drivers of AI innovation evolution in manufacturing clusters. The increased resource availability fosters a favorable environment for firms to invest in and adopt AI technologies, leading to broader innovation within the cluster. This aligns with the evolutionary economic geography theory, which posits that resource concentration accelerates innovation through mechanisms such as knowledge spillovers and collaborative learning. As a result, clusters with rich resources are more likely to become hubs for AI-driven innovation, reinforcing the critical importance of resource-rich environments in the growth of technological ecosystems.

The Zhongguancun area in Beijing, often referred to as the “Silicon Valley of China," exemplifies the critical role of clustered resources in fostering the emergence and diffusion of AI-driven manufacturing innovations. As a hub for intelligent manufacturing, Zhongguancun demonstrates how financial resources, human capital, digital infrastructure, and R&D capabilities drive the evolution of AI innovation. For instance, Zhongguancun Smart Manufacturing Street spans 30,600 square meters and houses 93 enterprises across fields such as Internet of Things (IoT), AI, robotics, and 3D printing. These firms collectively generate an annual output exceeding CNY 3 billion, showcasing the transformative potential of resource-rich environments.

Zhongguancun’s success is underpinned by substantial investments in financial and human resources. By 2021, venture capital and government-supported funds had injected over CNY 150 billion into AI-related industries, significantly enhancing the R&D capacity of firms and enabling them to adopt cutting-edge AI technologies. The region’s proximity to premier academic institutions, including Tsinghua University and Peking University, further bolsters its talent pipeline. With approximately 60% of China’s AI workforce originating from these institutions, Zhongguancun ensures a steady flow of skilled professionals equipped to leverage AI technologies in manufacturing processes. The simulation analysis in this study highlights how higher μ values (representing the coefficient of firm resource ownership) accelerate the adoption and diffusion of AI innovation within manufacturing clusters. Zhongguancun exemplifies this relationship through its comprehensive digital infrastructure, which includes advanced broadband, high-speed data centers, and shared AI computing platforms, collectively enhancing firms’ capabilities to absorb and apply AI technologies. As firms with greater access to resources actively engage in innovation, they stimulate spillover effects, encouraging surrounding firms to participate in the evolution of AI innovation.

Zhongguancun’s resource richness encompasses not only financial and digital assets but also a robust culture of innovation. The region’s Smart Manufacturing Innovation Center and various smart factory initiatives, such as the ’Smart Manufacturing 100’ program, exemplify how a supportive innovation culture fosters the development of AI-driven solutions. These initiatives align with the simulation findings, where increasing μ values from 0.3 to 0.7 correspond to a significant increase in the number of firms actively engaging in AI innovation.

The real-world outcomes observed in Zhongguancun further validate the conclusions drawn from the simulation. The region’s strong cluster resources have facilitated the establishment of over 100 smart factories and intelligent manufacturing systems, significantly enhancing the adoption and diffusion of AI-driven manufacturing innovations. As a leading example of how financial capital, human talent, and digital infrastructure converge to accelerate AI innovation, Zhongguancun illustrates that resource-rich environments are crucial for promoting the evolution of AI-enabled manufacturing clusters.

Impact of cluster network on AI-innovative manufacturing industry cluster

To gain a deeper understanding of the role of knowledge sharing in the evolution of AI-driven innovations within manufacturing industry clusters, a simulation analysis was conducted. The parameters μ and e were fixed, while the range of r values was systematically explored. This approach facilitated the collection of data under varying conditions, enabling the determination of whether a causal emergence phenomenon occurred. As an approximate solution strategy, this method enhances comprehension of the intricate relationship between knowledge sharing and innovation evolution.

The results indicate that a higher degree of knowledge sharing within the cluster accelerates the evolution of AI-driven innovations. Specifically, an increased risk appetite among companies, along with collaborative knowledge sharing and strategic cooperation, correlates with a higher affinity coefficient of network contact. The parameter r represents the network contact affinity coefficient, which follows a uniform distribution. Given μ = e = 0.5, varying r values reveal that larger r values indicate a greater degree of knowledge sharing within the manufacturing industry cluster.

The specific simulation process is as follows:

First, an evolution loop was defined, where the length of r is denoted as k. For each r value, the simulation was run for a number of steps. In each step, a new matrix was created to store the next state. Each cell was traversed to calculate the number of neighboring cells entering the innovation cluster, using the Von Neumann neighborhood method, which considers 4 neighbors.

Next, the value of p1 was calculated, following a normal distribution N(μ, σ2), using the current μ value. A uniformly distributed random number in the range [0, r] was generated as p2. The value of P was then calculated using the formula \(\scriptstyle{p}={e}\times \left({p}_{1}+{p}_{2}\times \frac{N(t)}{M}\right)\).

The state was updated based on the value of P. If all P-values were greater than p0 and the current cell state was 0, it was updated to 1. After updating the cell states, the final state was saved.

Through this simulation process, the final states for different r values were obtained. The simulation results indicate that higher r values, representing a higher degree of knowledge sharing, facilitate the acceleration of AI-driven innovation evolution within manufacturing industry clusters.

Figure 8 illustrates the evolution of firms in the cellular space under varying levels of the knowledge-sharing coefficient (r), set at 0.4, 0.6, and 0.8. The green cells highlight firms actively engaged in AI innovation, while the blue cells represent those that have yet to take part in the process. The simulation results demonstrate a clear positive relationship between knowledge sharing and the acceleration of AI innovation within manufacturing industry clusters. At r = 0.4, only a small number of firms engaged in AI innovation, reflecting the limited diffusion potential in environments with weak inter-firm knowledge-sharing mechanisms. As the knowledge-sharing coefficient increases to 0.6, the diffusion process is moderately enhanced, with a greater number of firms adopting AI technologies. At r = 0.8, the majority of firms within the cluster actively engage in AI innovation, showing that strong knowledge-sharing networks significantly enhance the speed and scale of AI adoption. These results emphasize the pivotal role of collaborative knowledge-sharing ecosystems in driving AI innovation evolution within manufacturing clusters.

To further illustrate this relationship, Fig. 9 provides a quantitative representation of the cumulative number of firms involved in AI innovation as a function of knowledge-sharing intensity. The findings show a steady increase in AI adoption as r rises, with a notable surge in firm participation when r reaches 0.8. Specifically, at r = 0.4, approximately 20 additional firms participated in the AI innovation cluster, whereas at r = 0.8, this number exceeded 30, reflecting the exponential effect of enhanced inter-firm knowledge exchange.

This aligns with the evolutionary economic geography theory, which posits that well-connected clusters facilitate knowledge diffusion, reduce technological learning curves, and enable firms to improve their innovation capabilities collectively. The presence of strong collaborative networks accelerates AI technology diffusion and enhances the overall resilience and adaptability of manufacturing clusters in an ever-evolving technological landscape.

These findings have several important implications. First, manufacturing clusters should actively develop structured platforms for knowledge exchange, such as AI research alliances, joint R&D centers, and digital innovation hubs, to foster collaborative learning and maximize technological spillovers. Second, informal knowledge-sharing mechanisms, such as professional networking events, open-source AI collaborations, and mentorship initiatives, should be encouraged to facilitate the organic diffusion of AI innovation.

Finally, cluster networks act as a multiplier effect, meaning that firms embedded in highly interconnected clusters experience faster AI adoption compared to those in isolated environments. Therefore, fostering strong inter-firm connections and enhancing collaborative knowledge-sharing mechanisms is crucial for accelerating the evolution of AI-driven innovation within manufacturing clusters.

Shenzhen, a leading hub for AI-driven manufacturing in China, exemplifies the vital role of cluster networks in promoting knowledge sharing and expediting the evolution of AI-driven innovations. With over 2200 AI enterprises operating within its ecosystem, Shenzhen illustrates how collaborative networks enable the diffusion of innovation through strategic cooperation and knowledge exchange. Industry leaders such as Huawei and Tencent serve as pivotal hubs, connecting smaller firms and research institutions, thereby enhancing the overall connectivity of the cluster network.

The success of Shenzhen’s AI-enabled manufacturing cluster is evidenced by initiatives such as the Open AI Innovation Center, which provides shared resources, including computing power, datasets, and simulation tools. These resources enable firms to pool their expertise and collaborate on joint R&D projects, effectively increasing the affinity coefficient of network contact (r). For instance, partnerships between Huawei and Tencent, as well as Huawei’s collaboration with UBTech Robotics, illustrate how strategic alliances enhance innovation capabilities across the cluster. In particular, the joint development of AI-powered robotic assembly lines for automotive manufacturing by Huawei and UBTech exemplifies how knowledge sharing and collaborative efforts can lead to significant efficiency gains, validating simulation findings that higher r-values correspond to more rapid innovation diffusion.

Shenzhen’s cluster also highlights the importance of risk appetite and cooperative competition in driving innovation. Smaller firms, supported by shared infrastructure and mentorship from larger enterprises, actively engage in self-innovation, further enriching the cluster’s knowledge pool. For example, Orbbec Inc.’s advancements in AI vision technologies and the integration of voice-controlled robotic arms were achieved through collaboration with both startups and established players, demonstrating the self-reinforcing nature of the cluster network.

The impact of Shenzhen’s collaborative networks on innovation is further substantiated by measurable outcomes. Recent data indicates that Shenzhen’s AI industry achieved a revenue of CNY 248.8 billion in 2022, reflecting a year-on-year growth of 32.1%. A significant percentage of surveyed firms reported enhancements in production efficiency and product quality directly linked to shared AI solutions within the cluster. These findings are consistent with simulation results, where increasing the r-value (network contact affinity) from 0.4 to 0.8 resulted in a notable increase in the number of firms actively engaged in AI innovation processes.

Supporting the simulation’s conclusions, Shenzhen’s cluster illustrates that higher levels of knowledge sharing and collaborative networks expedite the evolution of AI-driven innovations. By fostering an environment conducive to strategic cooperation, resource sharing, and risk-taking, Shenzhen’s AI-enabled manufacturing cluster provides tangible evidence of the theoretical framework’s predictions. This ecosystem not only cultivates innovation within individual firms but also enhances the overall adaptability and competitiveness of the manufacturing cluster.

Impact of cluster environment on AI-innovative manufacturing industry cluster

To investigate the impact of the cluster environment on the evolution of AI-driven innovations within manufacturing industry clusters, a simulation analysis was conducted. The values of μ and r were fixed, while the range of e values was systematically traversed. This approach facilitated the collection of data under varying conditions, enabling the subsequent determination of causal occurrences. As an approximate solution strategy, this method enhances our understanding of the complex relationships between the cluster environment and innovation evolution.

Given that μ = r = 0.5, we varied e. When the environmental conditions of manufacturing industry clusters, such as economic factors, political influences, industry standards, and market demand, are all favorable, the value of e tends to increase. Larger e values indicate stronger policy support and improved economic conditions.

The specific simulation process is as follows: First, an evolution loop was defined, where the length of e values is denoted as k. For each e value, the simulation was run for a number of steps. In each step, a new matrix was created to store the next state. Each cell was traversed to calculate the number of neighboring cells entering the innovation cluster, using the Von Neumann neighborhood method, which considers 4 neighbors.

Next, the value of p1 was calculated, following a normal distribution N(μ, σ2), using the current μ value. A uniformly distributed random number in the range [0, r] was generated as p2. The value of P was then calculated using the formula \(p=e\times \left({p}_{1}+{p}_{2}\times \frac{N(t)}{M}\right)\).

The state was updated based on the value of P. If all P-values were greater than p0 and the current cell state was 0, it was updated to 1. After updating the cell states, the final state was saved.

Through this simulation process, the final states for different e values were obtained. The simulation results indicate that higher e values, representing stronger policy support and better economic conditions, facilitate the acceleration of AI-driven innovation evolution within manufacturing industry clusters.

Figure 10 illustrates the evolution of AI innovation across firms in the manufacturing cluster under varying environmental support coefficients (e) set at 0.4, 0.6, and 0.8, with resource ownership (μ) and knowledge sharing (r) held constant at 0.5. The green cells represent firms that are actively participating in the evolution of AI innovation, while the blue cells indicate firms that have yet to engage in the process. As e increases, the number of firms adopting AI innovation grows, showing a positive relationship between environmental support and innovation participation. At e = 0.4, there is limited engagement due to unfavorable environmental factors. At e = 0.8, the majority of firms participate, demonstrating how supportive economic conditions and policy environments encourage widespread AI adoption.

Fig. 10: Evolution of AI innovation with varying environmental support (e).
figure 10

This figure shows the AI innovation evolution process in a manufacturing cluster under different environmental support coefficients (e = 0.4, 0.6, 0.8) with fixed μ = 0.5 and r = 0.5. The first, second, and third images illustrate the cellular space after 30 simulation steps for e = 0.4, 0.6, and 0.8, respectively. Green cells denote AI-adopting firms (S = 1), and blue cells represent non-adopting firms (S = 0).

Full size image

Figure 11 presents a quantitative representation of the number of firms adopting AI innovation in the cluster as a function of varying environmental support coefficients (e) at values of 0.4, 0.6, and 0.8, while μ and r remain fixed at 0.5. The graph clearly shows that as e increases, so does the number of AI-innovative firms within the cluster. At e = 0.4, the low level of external support restricts the diffusion of AI technologies, and only a few firms participate. However, as e increases to 0.6 and 0.8, a marked increase in the number of AI-innovative firms is observed. This highlights the impact of a supportive economic and policy environment in fostering the adoption of AI technologies across a broader range of firms. The findings align with the understanding that government support, policy incentives, and favorable economic conditions are pivotal in driving technological innovation within industrial clusters.

Fig. 11: Number of AI-innovative firms with varying environment support (e).
figure 11

This figure quantifies the number of AI-innovative firms in the manufacturing cluster over 30 simulation iterations for different environmental support coefficients (e = 0.4, 0.6, 0.8) with fixed μ = 0.5 and r = 0.5. The plot displays three lines, each corresponding to an e value, with the x-axis representing simulation steps and the y-axis indicating the cumulative number of firms with state S = 1.

Full size image

The results indicate that a favorable cluster environment can facilitate the evolution of AI innovation within the cluster, whereas an unfavorable cluster environment can significantly impede it. The industrial cluster environment encompasses factors such as the economy, policy, industry standards, and market demand, all of which positively influence talent development within clusters. Consequently, it is crucial to establish a conducive cluster environment to foster the evolution of AI-driven innovations in manufacturing industry clusters.

Bangalore, often referred to as the ’Silicon Valley of India,’ serves as a compelling example of how a favorable cluster environment promotes the evolution of AI-enabled manufacturing clusters. As a global hub for intelligent manufacturing and semiconductor industries, the city exemplifies the transformative impact of policy support, economic strength, and market demand in driving AI-driven innovations. Leading companies such as Hindustan Aeronautics Limited, Bosch, IBM, and Intel have integrated AI technologies into their manufacturing processes, utilizing advanced techniques like selective laser sintering and fused deposition modeling to significantly reduce production times and enhance efficiency. These advancements underscore the importance of a resourceful and supportive environment in enabling firms to adopt and scale AI technologies.

Strong policy support is a cornerstone of Bangalore’s AI-enabled manufacturing ecosystem. Government initiatives such as the Karnataka Artificial Intelligence Policy (2019) and the Startup India program provide tax incentives, subsidies, and funding partnerships that promote the adoption of AI in manufacturing. Additionally, innovation hubs like the Center of Excellence for Artificial Intelligence have facilitated over 150 AI-driven manufacturing projects, focusing on Industry 4.0 applications such as predictive maintenance and process automation. These policies align with simulation findings indicating that higher e-values, which represent favorable cluster environments, significantly accelerate the adoption and diffusion of AI technologies. For example, simulation results show that at e = 0.8, all firms within a cluster engage in the AI innovation evolution, highlighting the transformative potential of a robust policy-driven cluster environment.

Bangalore’s economic foundation further strengthens its position as a leader in AI-enabled manufacturing. The city contributes over $77 billion annually to India’s GDP, equipping companies with the financial resources necessary to invest in cutting-edge AI technologies. These economic advantages create fertile ground for the integration of AI technologies into manufacturing clusters, enabling firms to maintain competitiveness in a rapidly evolving global landscape. Simulation analyses reveal that strong economic conditions (high e-values) significantly enhance firm participation in innovative evolution, as evidenced by Bangalore’s ability to attract major investments and foster large-scale AI-driven manufacturing initiatives.

Bangalore’s success also reflects the influence of market demand and adherence to global standards. Leading manufacturing firms, such as Bosch and Toyota, collaborate with local startups and research institutions to develop advanced AI solutions for robotics and quality control. For instance, the partnership between Toyota Kirloskar Motors and IISc to create AI-driven quality assurance algorithms illustrates how strategic cooperation can accelerate innovation. Furthermore, the city’s burgeoning semiconductor industry, exemplified by Intel’s Very Large-Scale Integration (VLSI) design hub, highlights how advanced R&D capabilities contribute to the ecosystem’s success. These real-world developments validate simulation findings that favorable cluster environments foster the widespread adoption and diffusion of AI technologies, with Bangalore serving as a model for AI-driven innovation in manufacturing clusters worldwide.

Discussion and conclusion

Discussion

This study employed a Cellular Automata (CA) model to investigate the complex dynamics of AI-driven innovation within manufacturing industry clusters. The simulation results reveal a strong positive correlation among key factors such as cluster resources, cluster networks, cluster environment, and the evolutionary process of AI-driven innovation within these clusters. These findings demonstrate that the dynamics of AI adoption within clusters are deeply interconnected and not simply additive.

Specifically, the richness of cluster resources significantly influences the likelihood of firms adopting AI technologies and engaging in innovative processes. Firms with access to abundant human capital, advanced AI infrastructure, and robust R&D capabilities are more likely to undertake transformative activities. Furthermore, as these resource-rich firms adopt AI technologies and interact with their neighbors, they generate spillover effects that enhance the innovation potential of surrounding firms. This finding underscores the critical role of resource concentration in AI-enabled manufacturing clusters, suggesting that resource-rich environments accelerate both the individual and collective innovation processes within these ecosystems.

Additionally, the degree of knowledge sharing within AI-enabled manufacturing clusters plays a pivotal role in accelerating the evolution of AI-driven innovation. Strong inter-firm networks, facilitated by AI-enabled collaboration platforms and real-time data exchange, significantly enhance the diffusion of innovation. Firms embedded in dense collaboration networks benefit from quicker access to cutting-edge AI tools and actionable market insights. These networks also enable firms to jointly tackle challenges and co-develop AI-driven solutions, leading to a more adaptive and resilient manufacturing cluster ecosystem. This suggests that fostering collaborative networks should be prioritized as a policy goal to expedite innovation diffusion.

The cluster environment, characterized by supportive government policies, robust economic foundations, and high market demand for AI-enabled manufacturing solutions, significantly influences the evolutionary trajectory of these clusters. Simulations indicate that clusters operating within environments that prioritize proactive AI-focused policies and exhibit strong demand for smart manufacturing technologies experience accelerated growth and heightened levels of innovation. In contrast, clusters situated in less favorable environments encounter stagnation and a slower diffusion of AI-driven innovations, highlighting the critical role of policy interventions and market incentives in nurturing the development of AI-enabled manufacturing clusters. This finding calls for targeted policy actions aimed at strengthening the external environment, such as implementing AI-specific regulations, tax incentives, and infrastructure investments.

From the simulation process, these findings suggest that the emergence and diffusion of AI-driven manufacturing innovations are profoundly influenced by the interplay among cluster resources, inter-firm networks, and the surrounding cluster environment. The three primary driving forces of AI innovation within manufacturing industry clusters are interconnected and mutually influential. The development of cluster networks relies heavily on the availability of resources, and a conducive external political and economic environment further enhances these networks. As the material and human foundation for cluster development, cluster resources underpin a favorable political and economic environment, which, in turn, ensures the realization of resource agglomeration advantages. To fully comprehend the impact of these three factors on the evolution of innovative clusters, it is essential to examine them collectively rather than in isolation. Consequently, this paper simulates and analyzes the evolutionary process of AI innovation in manufacturing clusters by constructing a driving force model. This approach provides a more holistic understanding of AI innovation in manufacturing clusters and quantifies the effects of the three driving forces on the development of manufacturing industry clusters.

Conclusion

The simulation results of the Cellular Automata (CA) model align with the theoretical perspectives of evolutionary economic geography and complex systems theory, which are grounded in CA modeling and simulation. This alignment reinforces the proposed model's robustness as a tool for understanding the dynamics of AI innovation in manufacturing clusters.

This study integrates both theoretical and simulation-based approaches to offer a nuanced understanding of the dynamics driving the evolution of AI-enabled manufacturing clusters. The key findings emphasize the interplay among cluster resources, AI-driven collaborative networks, and supportive environments in shaping the innovation potential of these clusters. By focusing on these interrelations, the study provides a more comprehensive view of how AI innovation evolves within a clustered ecosystem.

By employing these theoretical frameworks, this research provides valuable insights into the intricate relationships between cluster resources, inter-firm networks, the cluster environment, and the emergence of AI-driven manufacturing innovations. Furthermore, the findings contribute to a broader understanding of clustered innovation ecosystems and the theoretical foundations of collective learning and competitive advantage within the AI context. These insights could be leveraged to design more effective AI innovation strategies for policymakers and business leaders alike.

The implications of this study extend to systems theory in practice, encompassing areas such as digital systems and complex social systems. By elucidating the micro drivers of macro changes, the proposed model serves as a useful tool for researchers to analyze motivation at the micro level while observing overall emergence at the macro level. This approach facilitates the visual simulation of the development and evolution of complex systems under varying scenarios, based on empirical data and utilizing computational tools. The real-world applicability of these insights is enhanced by the ability to adjust agent rules and attributes, ensuring that the model is adaptable and relevant to policymakers and practitioners in the field.

Policy implications

To promote the evolution of artificial intelligence (AI) innovations in manufacturing enterprises, targeted and actionable policy recommendations are proposed to enhance cluster resources, networks, and environments.

Firstly, to improve cluster resources, including human capital, digital infrastructure, research and development (R&D), and corporate culture, it is recommended that policies focus on actively fostering AI talent development and providing incentives to companies for AI-driven initiatives. Specifically, governments should allocate funding to AI-focused education and training programs, directly addressing skills gaps in critical areas such as data science, machine learning, and AI applications in manufacturing. In addition to fostering a skilled workforce, policies should encourage companies to invest in AI R&D by offering tax breaks or subsidies for firms developing AI technologies. Furthermore, policies should aim to build and maintain cutting-edge digital infrastructure, such as high-speed networks and advanced computing resources, which are essential for integrating AI into manufacturing processes. Government-backed AI innovation hubs could also play a pivotal role in nurturing startups and facilitating collaboration between academic institutions and industrial players. Lastly, policies should promote a corporate culture that embraces innovation and risk-taking, which could be achieved through recognition programs, performance-based grants, and awards for companies making significant strides in AI-driven innovation.

Secondly, to enhance cluster networks, including corporate risk appetite, collaborative knowledge sharing, and strategic cooperation, policies should focus on fostering a stronger collaborative ecosystem within manufacturing clusters. This could involve the creation of dedicated platforms for knowledge exchange, such as AI-focused forums, workshops, and networking events, where companies, researchers, and policymakers can collaborate and share insights. Encouraging public-private partnerships in AI development is crucial, and policies should promote the establishment of AI consortia and industry alliances to tackle challenges together and drive innovation forward. Additionally, policies should promote a culture of experimentation and risk-taking within manufacturing clusters, which could be supported by risk capital programs and innovation grants. These programs would help firms de-risk investments in AI technologies, encouraging them to push the boundaries of innovation while ensuring they can sustain early-stage developments.

Finally, to enhance the cluster environment, encompassing economics, policy, industry standards, and market demand, policies should focus on creating a conducive and dynamic economic and regulatory framework for AI innovation. This should include offering tax incentives, grants, and funding for AI research and development projects to reduce the financial burden on firms investing in AI. Moreover, simplifying regulatory frameworks is essential to encourage AI adoption while ensuring compliance with ethical and legal standards. Policies should also prioritize the establishment of industry standards and certifications for AI technologies to ensure interoperability and trust in AI solutions. Establishing AI standards bodies at the national or regional level could help guide the development of universally accepted norms. Furthermore, to stimulate demand for AI innovations, policies should encourage public procurement of AI solutions, especially in sectors such as healthcare, manufacturing, and transportation. Government-backed campaigns and awareness initiatives would help raise awareness about the transformative potential of AI and its applications across different industries.

Research limitations and future prospects

This study, while contributing valuable insights into AI-driven innovation within manufacturing clusters, has several limitations that must be considered. First, the sample size used in the MATLAB simulation was limited to n = 20, representing 400 firms within a single sector of the cellular industry. While this approach provided initial insights into the dynamics of AI adoption in these clusters, the small sample size limits the generalizability of the findings. A larger and more diverse sample would likely provide a more comprehensive understanding of AI innovation processes across various industrial sectors and regions. Future studies would benefit from expanding the sample size to encompass a broader range of industries, thus enhancing the robustness and applicability of the findings.

Second, the study focused exclusively on manufacturing industry clusters. This narrow scope may limit the applicability of the results to other types of clusters, particularly those in sectors such as services, high-tech, or creative industries, where the dynamics of innovation and AI adoption might differ. By extending the model to include other industry clusters, future research could explore how AI-driven innovation evolves in different economic contexts and identify sector-specific drivers or barriers to adoption. This expansion would also enable comparisons across sectors and provide a deeper understanding of the nuances involved in AI innovation within various clusters.

The study also relied on a single simulation method—Cellular Automata (CA)—to model the evolution of AI innovation within manufacturing clusters. While the CA model is a powerful tool for simulating localized interactions and emergent behaviors, it has limitations in capturing the broader and more intricate interaction patterns that occur in real-world industrial ecosystems. The von Neumann neighborhood configuration used in this study may not fully represent the complexity of interactions between firms, especially those that occur at a regional or global scale. Future research could overcome this limitation by incorporating alternative simulation approaches, such as agent-based models or higher-dimensional CA configurations. These alternative methods would allow for a more detailed exploration of the interactions within clusters, capturing both localized and global dynamics, and would further enrich the understanding of AI adoption processes.

Another limitation of this study is the potential for biases in the data. The data used in this study were concentrated in specific regions and industrial contexts, which could introduce regional or temporal biases that affect the accuracy and validity of the findings. To mitigate these biases, future research could incorporate more diverse data sources, including longitudinal data that captures the evolving dynamics of AI innovation over time. By broadening the scope of data collection, future studies would be able to offer a more comprehensive and unbiased analysis, providing more reliable insights into the diffusion of AI technologies within clusters.