Winning by Intelligence: Leveraging the Innovative Advantages of Intelligent Transformation in Market Competition

Abstract

Intelligent transformation plays an important role in enhancing innovation capabilities. This study utilizes data from 5000 A-share firms in China to investigate the effects of intelligent transformation on innovation capability and to elucidate the underlying mechanisms. We find that (1) there is a positive relationship between intelligent transformation and innovation capability. Furthermore, pursuing profit maximization and sustainable competitive advantage drives firms to capitalize on the technological dividends of intelligent transformation to enhance substantive innovation capability. (2) Intelligent transformation affects the innovation capability and structure through the mechanism of total factor productivity and market competition. (3) Within the intelligent transformation, non-state-owned enterprises exert a stronger influence on enhancing innovation capability, but state-owned enterprises remain the primary drivers of substantive innovation. The synergistic effect of intelligent transformation, combined with the scale advantages of large enterprises, create substantial opportunities for firms to enhance their innovation capability. Additionally, intelligent transformation notably enhances both the innovation capability and the innovation quality of labor-intensive enterprises. While capital-intensive enterprises experience significant improvements in their overall innovation capability, their substantive innovation capability has not shown comparable advancements. These findings contribute to a more accurate evaluation of the effects of intelligent transformation on innovation capability. The findings also help cultivate differentiated competitive advantages and drive high-quality development through intelligent empowerment.

1. Introduction

Since the advancement of the new era, Chinese firms have encountered an increasingly complex macroeconomic environment and heightened risks stemming from market uncertainty. Strengthening innovation capability has become essential for enterprises to cope with these risks and enhance their competitiveness [1]. However, under the constraint of limited R&D investment, enterprise managers often face dilemmas in making innovation-related decisions [2]. The prevailing innovation model of Chinese enterprises, which primarily relies on cheap labor and substantial resource inputs, significantly constrains their potential for innovation and development [3]. This limitation becomes more pronounced as resource advantages gradually diminish. Additionally, Chinese enterprises face a shortage of technological innovation capability, along with insufficient technical resources and talent. Moreover, the overall level of industrial technology requires significant improvement [4]. In this context, exploring new pathways for firm innovation has become a critical focus of contemporary research.

Intelligent transformation refers to the integration of artificial intelligence technologies into the production processes, which optimizes the production process, reduces costs, improves the production and product quality, and enhances the overall competitive advantage of the organization [5]. It markets a strategic transition toward leveraging advanced technologies to redefine value creation [6]. This transformation fundamentally impacts market structures and drives firms to adjust their innovation capability in response to evolving competitive pressures. Prior research widely acknowledges that changes in market dynamics and productivity significantly influence firms’ motivation and ability to innovate [4]. However, there has always been a divergence in the positive and negative of this effect. Some scholars argue that increased market competitiveness drives firms to them improve innovation capability [7]. The underlying mechanism is driven by firms’ pursuit of material benefits, which leads them to focus more on improving their products, services, and brands through innovation [4]. Others suggest that intense market competition may impede R&D investment by lowering potential returns and restricting productivity growth [8]. Scholars point out that, by increasing efficiency, intelligent transformation can intensify market competition [9]. The following questions arise: Can smart transformation create a new path for enterprise innovation? What is the role of production efficiency and market competition in this process?

The existing literature has examined intelligent transformation and firm innovation separately. Firstly, the research on intelligent transformation. The focus of the existing literature is on identifying the socio-economic benefits of smart transformation. Its role in promoting green development [10], enterprise information transparency [11], government subsidy efficiency [12], and value chain upgrading [13] has been widely demonstrated. Advancing the digitalization level, driving R&D investment, promoting information sharing, and optimizing resource allocation are the possible paths of action. Some scholars have paid attention to the innovation effect of intelligent transformation, but their studies focus on the discussion of the innovation effect on a single industry [14]. Another strand of the literature pays more attention to the heterogeneity of the socio-economic effects of intelligent transformation. Some scholars have found that smart transformation will have stronger effects on eastern region firms, state-owned enterprises (SOEs), and capital-intensive firms [15]. Cao et al. (2024) further emphasized that non-state-owned firms with strong market competitiveness can better utilize smart transformation to generate environmental effects [16].

Second, the research on firm innovation. This literature can be divided into the following two parts: research on the influencing factors and discussion of the innovation duality. Many factors influence firms’ innovation capability. These factors include internal elements such as the organizational structure, organizational culture [13], organizational strategy [17], and manager characteristics [18]. Furthermore, external factors have a crucial impact. These include market competition, capital market reforms, and institutional protection [19]. In recent years, the influencing factors of firm innovation has been connected to the widespread use of new technology [20]. In terms of the innovation duality discussion, scholars have emphasized that it is reflected in pioneering and exploratory innovations [21]. Exploratory innovation entails radical redesign and fundamental rethinking, whereas exploitative innovation strives for incremental and continuous improvements [22]. In recent years, China has experienced a surge in innovations. While innovations have increased, there is growing concern about the proportion of exploratory innovation.

In summary, after answering the question of whether intelligent transformation affects firm innovation, there are still some unresolved issues. Firstly, whether the intelligent transformation has brought substantial innovation needs further exploration. Secondly, production efficiency and market competition are incentive mechanisms that drive firm innovation. However, the impact of intelligent transformation on production efficiency and market competition, and its influence on innovation capability through this transmission mechanism, has not yet been revealed. In addition, in emerging market economies such as China, the market structure is not only influenced by institutional monopolies of state-owned capital, but also by natural monopolies generated by market competition. Understanding the impact of intelligent transformation on the innovation capability and structure of different ownership types and resource-intensive industries is also an urgent research need.

In response, we provide an empirical analysis aimed at determining how intelligent transformation has affected firm innovation in China. This includes measuring the intelligent transformation level and the exploratory innovation level of Chinese firms, using a panel fixed-effects model to quantify the effect of intelligent transformation on firms’ innovation, discussing the path mechanism of the role of intelligent transformation on firms’ innovation, and analyzing the heterogeneity of the innovation effect of intelligent transformation.

The significance of this study is presented as follows: (1) It provides theoretical significance by expanding the frontiers of firm innovation research. We have integrated intelligent transformation and firm innovation into a unified analytical framework. Through the investigation of the responsiveness of firm innovation to intelligent transformation, the research scope of the factors influencing enterprise innovation has been broadened, and a deeper comprehension of firm innovation activities has been achieved. (2) It holds theoretical significance for deepening the understanding of resource-based and market competition theories. We have emphasized the mediating roles of total factor productivity and market competition within the frameworks of the resource-based theory and market competition theory. This has enhanced our understanding and supplemented these theories. (3) It offers guiding significance for the innovative decision making of firm managers. We have examined the heterogeneous impacts of intelligent transformation on different types of firms, thereby providing a practical foundation for managers of diverse firm types to make differential innovation decisions. (4) It has a referential value for other developing countries in formulating policies. Our findings reveal the relationship between China’s intelligent transformation and firm innovation. Given China’s representatives among developing countries, our conclusions can also function as a practical reference for other developing countries in devising relevant policies.

The paper is organized as follows: Section 2 presents the theoretical framework; Section 3 presents the study design; Section 4 and Section 5 analyze the results; Section 6 presents the conclusions and discussion.

2. Theoretical Framework

2.1. The Impact of Intelligent Transformation on Innovation Capability

Intelligent transformation emphasizes the integration of intelligent technology into the existing production processes to rapidly collect, analyze, plan, and reorganize products, processes, and resource information so to facilitate product design, function simulation, and prototype manufacturing, thus quickly producing products to meet the needs of customers [17]. Innovation entails the combination of knowledge to develop new and beneficial products, processes, or services [23]. This means that knowledge is the foundation of innovation [24], and effective knowledge management is the basis of innovation [25]. Corresponding to knowledge management, intelligent transformation enhances the efficiency of knowledge management, including knowledge acquisition, integration, and production. Specifically, intelligent transformation significantly accelerates data acquisition while enhancing accuracy and reliability. For example, in agriculture, researchers use sensors to monitor soil humidity, temperature, and crop growth in real time, ensuring continuous data collection [26]. This supports the discovery of new agricultural production laws, optimizes planting strategies, and improves the crop yield and quality. Furthermore, intelligent technologies include machine learning, deep learning, and natural language processing [27]. They can complete numerous repetitive and complex tasks quickly, enabling researchers to explore various methods of knowledge integration, verify assumptions, improve processes, and accelerate innovation. In addition to improving the process efficiency, intelligent transformation also extracts relevant knowledge from multiple domains and disciplines. This facilitates the rapid creation of new knowledge and computational schemes, thus speeding up the process of knowledge integration. In general, intelligent transformation hastens the process from capturing and structuring vast data to transforming it into information and then into knowledge [28]. As shown in H1 of Figure 1, we propose the following hypothesis:

Hypothesis 1:

Intelligent transformation can enhance the innovation capability.

2.2. The Impact of Intelligent Transformation on Innovation Quality

Evidence from successful firms suggests that technology is the main starting point, as it allows for the efficient detection of technology and market trends (i.e., the outside–in orientation). On this basis, firms can gain sustainable competitive advantages by improving their innovation capability. As an advanced technology, intelligent technology can improve production efficiency and help firms discover new applications for existing knowledge, thereby enhancing their innovation capability. However, innovation requires a substantial investment of R&D resources and time. Firms must effectively configure their resources to support these efforts [29]. According to organizational dualism theory, innovation predominantly arises from either independent or imitative innovation, known as exploratory innovation and exploitative innovation, respectively [30]. Exploratory innovation is more radical. It develops new products and services, frequently based on cutting-edge technologies, to meet the needs of growing markets [31]. Due to the uniqueness and novelty of exploratory innovation, firms can gain more competitive advantages and occupy a greater market share [32]. In contrast, exploitative innovation addresses the needs of existing customers and markets. It centers on manifest requirements, which are defined as improved and updated iterations of current goods and brands [33]. Although the risk is low, the benefits to the firms are also low. As illustrated in H2, given the nature of economics, firms are likely to leverage intelligent transformation to strengthen their exploratory innovation capability to obtain more profit, namely, substantive innovation. Based on the analysis, we propose the following hypothesis:

Hypothesis 2:

Intelligent transformation is positively associated with the substantive innovation capability.

2.3. The Moderate Total Factor Productivity

Resource-based theory maintains that innovation is the ongoing search for knowledge that yields discoveries and products [34]. The effectiveness of this effort largely depends on acquiring resources and sufficient financial and administrative support [35]. According to Liu (2020), the provision of resources can enhance innovation by attracting talent, covering laboratory costs, and providing necessary equipment [36]. Intelligent transformation, as a typical representative of production technology progress, has transformed traditional resource allocation [37]. This precise matching of factor resources improves total factor productivity and strengthens the support needed for firms to improve their innovation capability [38]. Specifically, intelligent technology penetrates all aspects of production, management, research and development, and social consumption. This integration enhances coordination among production factors, enabling firms to allocate resources more efficiently, thereby reducing waste and maximizing output. Firms can reallocate saved resources—including time, money, and labor—toward fostering innovative activities. On the other hand, intelligent equipment serves as a cost-effective capital element with higher production efficiency in repetitive tasks [39]. Compared to manual labor, intelligent technology is more efficient and precise, leading to increased productivity. The reduction in manpower demand and productivity improvement allows firms to allocate more financial and human resources to R&D [40], thereby enhancing their capacity to develop new products, processes, and technologies. Additionally, intelligent equipment requires programming, maintenance, and troubleshooting. Integrating intelligent technologies into the production processes also requires skilled professionals. Consequently, intelligent transformation drives the need for a highly skilled workforce. Such a workforce not only ensures the effective use of intelligent technology in production but also helps develop new applications and enhance firms’ innovation capability. As shown in hypothesis 3 of Figure 1, intelligent transformation improves the production efficiency by increasing the total factor productivity. This process provides important resources for enterprises to enhance their innovation capabilities. Based on this, we propose the following hypothesis:

Hypothesis 3:

Intelligent transformation affects the innovation capability through the total factor productivity mechanism.

2.4. The Moderate Market Competition

Modern theories of economic growth assert that technological advancement boosts firms’ productivity and competitiveness, creates new growth opportunities, and significantly influences the overall market structure [41]. As illustrated by H4, the impact of intelligent transformation on market competition, specifically, intelligent transformation, can reduce marginal costs by using the comparative advantages of programmed and processed jobs to replace low-skilled labor [42]. Cost reduction boosts product price markups and profit margins, providing firms with a competitive advantage and expanding their market shares. Additionally, intelligent transformation improves the production efficiency and attracts new entrants to the market by lowering market entry costs and increasing the profits of existing ones [43]. This dynamic cycle reduces the market concentration and intensifies market competition in the markets.

Transitioning to the effect of market competition, it is commonly acknowledged that a firm’s inclination for innovation and its strategic orientation are influenced by the level of market competition. Specifically, a highly competitive market environment is defined by an abundance of substitute goods and services that customers can select from, an ongoing stream of goods and services, and permanent price erosion [44]. Firms have to employ a range of innovative strategies to set themselves apart from the competition and increase the market share by lowering prices [45]. However, competitive advantage is hard to acquire and simple to lose. Resources are scarce and flexibility is constrained in highly competitive markets, which exposes firms to both inherent and external risks. Intelligent technologies can quickly assess large volumes of information and identify market trends and client needs. Firms can also leverage the resource allocation capability of intelligent technologies to modify current products and launch new ones in response to changing customer needs. Conversely, in markets with low levels of competition, intelligent transformation may not be as effective [18]. This market is characterized by limited customer choices, fewer innovations, and less price competition [44]. Furthermore, the market is often monopolized by a small number of firms that lack the incentive to innovate. Therefore, the ability to innovate—that is, to modify current products in response to shifting customer needs or to introduce whole new ones—does not confer the same advantages on new ventures as it does in markets with intense competition. Considering these observations, we propose the following hypothesis:

Hypothesis 4:

Intelligent transformation affects the innovation capability through the transmission mechanism of market competition.

3. Study Design

The section constructs an econometric model to investigate the relationship between intelligent transformation and innovation capability. It also explains the related variables.

3.1. Model Specification

To explore the effects of intelligent transformation on firms’ innovation capability, we refer to the approach of Zhang et al. (2024) and constructed a linear regression model with two-way fixed effects [4], as shown in Equation (1):

$${Innov}_{jt}={\alpha }_0+{\alpha }_1{Intelligent}_{jt}+{\alpha }_2\sum {Control}_{jt}+{\mu }_k+v_t+{\epsilon }_{jt}$$

(1)

where $\mathrm{j}$, $\mathrm{t}$, and $k$ represent the firm, year, and industry, respectively; ${Innov}_{jt}=\left\{\mathrm{i}\mathrm{n}\mathrm{n}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{c}\mathrm{a}\mathrm{p}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}, \mathrm{i}\mathrm{n}\mathrm{n}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{s}\mathrm{t}\mathrm{r}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}\right\}$ represents the set of explained variables; ${Intelligent}_{jt}$ represents the logarithm of the intelligent transformation as the explanatory variable; ${\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{j}\mathrm{t}}$ is a series of controls, including firms’ sustainable growth rate (Sgrow), leverage (Lev), size (Size), capital intensity (Density), shareholder concentration (Stock), age (Age), and market capitalization (Bm); ${\mathsf{\mu }}_{\mathrm{k}}$ and $v_t$ represent time-fixed effects and industry-fixed effects, respectively; ${\epsilon }_{jt}$ is the random error term.

To test whether intelligent transformation affects enterprises’ innovation capabilities through the transmission mechanism of total factor productivity and market competition, we constructed a mediation effect model based on a two-step approach, reference to Jiang et al. (2022) [46], as follows:

$${Mediator}_{jt}={\beta }_0+{\beta }_1{Intelligent}_{jt}+{\beta }_2{\sum Control}_{jt}+{\mu }_k+v_t+{\epsilon }_{jt}$$

(2)

where ${Mediator}_{jt}$ represents the mediating variables of “total factor productivity” and “market competition”; ${\beta }_0$ is the constant term; ${\mathsf{\beta }}_1$ denotes the effect of the independent variable on the mediating variable. The remaining variables are defined in the same way as in Equation (1).

3.2. Variable Declaration

3.2.1. Dependent Variable: Innovation Capacity

Given its intangible nature, innovation capability is often assessed using indirect measures such as R&D intensity, patents, and licenses. Our focus on patents as an indicator of innovation capability follows the idea that, “without inventions, there are no innovations” [47]. To account for the inherently high-risk nature of innovation endeavors, we used the number of applied patents as an indicator of the innovation capability (Innov). A higher number of applied patents indicates greater ongoing innovation by a firm. This paper considers the possibility that a firm may receive no patents applied each year, as well as the potential of fat tails. When processing the data, we estimated a firm’s innovation capability by taking the logarithm and adding 1 to the applied patents.

The State Intellectual Property Office of China (SIPO) grants the following three types of patents: invention patents, utility model patents, and design patents [48]. In comparison to utility model patents and design patents, invention patents are characterized by a greater investment of resources, higher R&D risks, and increased technical complexity. To further investigate the impact of intelligent transformation on the innovation quality, we defined invention patents as indicative of exploratory innovation, while utility model patents and design patents are classified as forms of exploitative innovation. We utilized the ratio of the number of invention patent applications to the total number of patent applications as a proxy for innovation quality (Innov_quality).

3.2.2. Independent Variable: Intelligent Transformation

As discussed in Section 2.1, intelligent transformation enhances the innovation capability by increasing the effectiveness of knowledge management. To empirically test this theory, we utilized Python’s ‘Jieba’ tool to analyze 29 keywords associated with intelligent transformation in the annual reports of listed companies. These keywords include terms such as “intelligence” “digitization”, and “informatization”1 Additionally, we refer to the methods of Wu et al. (2021) [49] to derive the proxy variable for firm intelligent transformation by calculating the proportion of the total number of relevant keywords to the total number of similar keywords for enterprises within the same industry during the same year, as shown in Equation (3):

$${DigIntTransition}_{jt}=q_{jt}/{\sum }_j^t{q}_{jt}$$

(3)

where $q_{jt}$ represents the total number of keywords for firm $j$ in period $t$; ${\sum }_{\mathrm{j}}^{\mathrm{t}}{\mathrm{q}}_{\mathrm{j}\mathrm{t}}$ represents the total number of similar keywords.

3.2.3. Mediating Variable: Total Factor Productivity

The total factor productivity (TFP) is often used to measure a firm’s productivity [50]. We assume that the production function of each firm is in the form of a Cobb–Douglas (CD) function, as shown in Equation (4):

$$Y_{jt}=A_{jt}{L}_{jt}^{\alpha }{K}_{jt}^{\beta }$$

(4)

where ${\mathrm{Y}}_{\mathrm{j}\mathrm{t}}$ represents the firm’s total output; ${\mathrm{L}}_{\mathrm{j}\mathrm{t}}$ and ${\mathrm{K}}_{\mathrm{j}\mathrm{t}}$ represent the firm’s labor and capital inputs, respectively; $A_{jt}$ represents the TFP. The logarithmic processing of Equation (4) yields a production function of the following form:

$$y_{jt}=\alpha l_{jt}+\beta k_{jt}+{\mu }_{jt}$$

(5)

where ${\mathrm{y}}_{\mathrm{j}\mathrm{t}}$, ${\mathrm{l}}_{\mathrm{j}\mathrm{t}}$ and ${\mathrm{k}}_{\mathrm{j}\mathrm{t}}$ represent the logarithms of $Y_{jt}$, ${\mathrm{L}}_{\mathrm{j}\mathrm{t}}$ and ${\mathrm{K}}_{\mathrm{j}\mathrm{t}}$, respectively. The residual ${\mu }_{jt}$ is the logarithm of ${\mathrm{A}}_{\mathrm{j}\mathrm{t}}$ (TFP). We estimate the above production function based on the semiparametric estimator (the OP method) developed by Olley and Pakes (1996) [51], which calculates the TFP for each firm. We further consider the LP method and OLS method based on Levinsohn and Petrin (2003) [52], and Lu and Lian (2012) [53], respectively, to calculate the TFP to ensure the robustness of the results.

3.2.4. Mediating Variable: Market Competition

Market concentration and profit margins are two widely used proxy variables for measuring market competition. Considering that profit margins may fluctuate due to temporary cost changes or demand shocks, which may obscure the true level of competition, the market concentration provides a clearer representation of the distribution of the market share among firms, and directly reflects the market structure and competition intensity. In highly concentrated markets, a small number of dominant firms reduce the competitive pressure, while, in more dispersed markets, the presence of numerous smaller firms typically fosters more intense competition. Thus, we selected market concentration as the proxy variable for measuring the market competition [54]. Market concentration is calculated as shown in Equation (6):

$${Cr5}_{kt}={\displaystyle \frac{{\sum }_{j=1}^5{Rev}_{kjt}}{{Rev}_{kt}}}$$

(6)

where ${Cr5}_{kt}$ represents the market concentration of industry $\mathrm{k}$ in period $t$; ${\sum }_{\mathrm{j}=1}^5{\mathrm{R}\mathrm{e}\mathrm{v}}_{\mathrm{k}\mathrm{j}\mathrm{t}}$ represents the sum of the main business revenues of the five firms with the highest in industry $k$; ${\mathrm{R}\mathrm{e}\mathrm{v}}_{\mathrm{k}\mathrm{t}}$ represents the sum of the main business revenues of industry $k$.

To ensure the robustness of the results, we also performed an additional analysis using the Lerner Index as an alternative measure [55]. The Lerner Index is calculated as shown in Equation (5):

$${Lerner}_{jt}={\displaystyle \frac{S_{jt}-{OC}_{jt}-{SE}_{jt}-{AE}_{jt}}{S_{jt}}}$$

(7)

where ${\mathrm{L}\mathrm{e}\mathrm{r}\mathrm{n}\mathrm{e}\mathrm{r}}_{\mathrm{j}\mathrm{t}}$ represents the Lerner Index; $S$ represents the sales; $OC$ represents the operating costs; $SE$ represents the selling expenses; $AE$ represents the administrative expenses; $j$ and $t$ represent the meanings consistent with the preceding.

3.2.5. Control Variables

To ensure the objectivity of the conclusions, we incorporated several control variables based on an in-depth review of the existing innovation research literature. Specifically, these variables include firms’ sustainable growth rate (Sgrow), leverage (Lev), size (Size), capital intensity (Density), shareholder concentration (Stock), age (Age), and market capitalization (Bm). For clarity and precision, the specific definitions of these variables are provided in

Table 1. Measurement of the control variables.

Variable Name	Variable Code	Meaning and Calculation Description	Reference
Sustainable growth rate	Sgrow	Return on net assets × retention of earnings/(1 − return on net assets × retention of earnings)	(Zhou et al., 2022) [56]
Firm leverage	Lev	Total debt/total assets × 100	(Nemlioglu and Mallick, 2021) [57]
Firm size	Size	The natural logarithm of total assets	(Stock et al., 2002) [58]
Capital intensity	Density	Total assets/operating income	(Lartey, 2020) [59]
Shareholder concentration	Stock	The proportion of shares held by the first largest shareholder	(Minetti et al., 2015) [60]
Firm age	Age	The natural logarithm of the year since the firm’s establishment plus 1	(Coad et al., 2021) [61]
Market capitalization	Bm	The natural logarithm of market values A	(Fedorova et al., 2022) [62]

.

3.3. Data Sources and Preprocessing

The A-share listed firms refer to firms registered in China whose shares are traded on mainland China’s stock exchanges. On the one hand, these firms are required to disclose financial data and major events within stipulated timeframes, enabling researchers to promptly access the latest information; on the other hand, the A-share market covers nearly all economic sectors, including manufacturing, high-tech industries, and the service industry, making it highly representative. Previous firm-level studies were mostly conducted based on A-share listed firms [63,64].

Our study focuses on A-share listed firms. Specifically, we constructed a firm-level dataset of A-share listed companies spanning the period from 2008 to 2022. The stock codes of these firms were obtained from the Wind database (https://www.wind.com.cn (accessed on 22 June 2024)). After excluding delisted firms, the dataset included 5000 companies. Our data integration process is as follows: First, we extracted keywords from each firm’s annual report and measured the intelligent transformation level. Second, we collected data on firms’ innovation capabilities, including applied invention patents, utility patents, and design patents, from the CSMAR database (https://data.csmar.com/ (accessed on 22 June 2024)) and measured firms’ innovation capabilities and quality of innovation. Third, we extracted data on firms’ innovation capabilities and quality of innovation from the Wind database and the CSMAR database to extract data on firms’ addresses, number of employees, and business and capital status. Finally, we summarized the above data and synthesized an unbalanced panel dataset based on the stock codes of the listed firms.

To avoid the influence of data magnitude on the analysis results, we logarithmically processed the data before constructing the econometric model. Figure 2 and

Table 2. Variable descriptive statistics.

Variable	N	Mean	SD	Min	p50	Max
Innov	41,773	1.529	1.616	0.000	1.099	9.358
Innov_quality	41,773	0.471	0.346	0.000	0.429	1.000
Intelligent	41,773	2.250	1.420	0.000	2.200	7.270
TFP_OP	41,773	6.617	0.916	2.049	6.535	11.447
Cr5	41,773	0.402	0.680	0.010	0.100	2.510
Sgrow	41,773	0.0500	0.0700	−0.130	0.0500	0.200
Lev	41,773	0.420	0.200	0.100	0.410	0.770
Size	41,773	3.090	0.0500	3.010	3.090	3.200
Density	41,773	2.380	1.510	0.710	1.930	6.510
Stock	41,773	0.340	0.140	0.130	0.320	0.620
Age	41,773	1.990	0.950	0.000	2.200	3.220
Bm	41,134	6.560	1.050	4.970	6.420	8.790

show violin plots and descriptive statistics for each variable after logarithmization, respectively. We also performed variance-inflated factor tests on the variables to avoid the problem of multiple contributions. The VIF values of the variables are below 10, which means that there is no multicollinearity problem and the econometric model can be constructed.

4. Empirical Results

4.1. Regression Analysis

This paper utilizes a double-fixed-effect model for the regression analysis to address the potential bias caused by omitted variables.

Table 3. Benchmark regression results.

	(1)	(2)	(3)
Variables	Innov	Innov	Innov_Quality
Intelligent	0.246 ***	0.180 ***	0.010 ***
	(16.951)	(12.891)	(2.693)
Sgrow		0.104	−0.017
		(0.643)	(−0.385)
Lev		−0.346 ***	−0.005
		(−3.753)	(−0.203)
Size		2.994 ***	−0.524 ***
		(4.001)	(−2.612)
Density		−0.102 ***	0.010 ***
		(−9.451)	(2.731)
Stock		−0.027	−0.064 **
		(−0.206)	(−2.155)
Age		−0.219 ***	−0.007
		(−11.216)	(−1.452)
Bm		0.194 ***	0.051 ***
		(5.020)	(5.280)
Constant	0.976 ***	−8.578 ***	1.754 ***
	(27.989)	(−4.058)	(3.098)
Observations	41,771	41,132	24,162
Adjusted R-squared	0.290	0.328	0.208
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; ** and *** represent significance at the 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

presents the benchmark regression results concerning intelligent transformation and firms’ innovation capability. In Column (1), the model includes the industry and year fixed effects without control variables. The coefficient indicating the impact of intelligent transformation on firms’ innovation is significantly positive at 1% of statistical significance. In Column (2), after adding the control variables, the regression results show a positive estimated coefficient for Intelligent. Additionally, the adjusted R² gradually increased. The results confirm that intelligent transformation has a positive impact on firms’ innovation capability in China, thus verifying hypothesis 1. The increasing adjusted R² value indicates the rationality of the model construction and variable selection. To sum up, as the level of intelligent transformation increases, firms can allocate production factors more efficiently, freeing up human, material, and other resources. This resource optimization supports innovative activities and enhances firms’ innovation capability.

This paper further examines the effect of intelligent transformation on firms’ innovation quality. The results are presented in Column (3). We observe that intelligent transformation significantly enhances firms’ substantial innovation capability. Hypothesis 2 is therefore confirmed. This phenomenon can be attributed to various factors. Firstly, firms may introduce intelligent technology primarily to meet market demands and address competitive pressures by enhancing the production efficiency and product quality. As a result, intelligent transformation is predominantly employed in existing processes, fostering an environment conducive to exploitative-oriented innovations. Secondly, exploratory innovation entails higher technological and market risks, requiring intricate technological breakthroughs and greater innovation capability. In contrast, exploitative innovation is comparatively easier to achieve and enables faster commercialization. Thus, firms choose to use intelligent technology to improve the production efficiency and achieve incremental product improvements. This approach helps to mitigate innovation risks, enhance the market position, and foster product differentiation.

To further explore the dynamic effects of intelligent transformation on firms’ innovation capability across different quantiles, we constructed a dynamic trend chart depicting the impact coefficients between the 10th and 90th quantiles. As shown in the quantile regression results in

Table 4. Quantile regression results.

Variables	Low		High
	Q10	Q25	Q75	Q90
Intelligent	0.050 ***	0.056 ***	0.078 ***	0.085 ***
	(0.017)	(0.014)	(0.010)	(0.013)
Sgrow	−0.017	−0.015	−0.006	−0.004
	(0.189)	(0.152)	(0.109)	(0.145)
Lev	−0.076	−0.073	−0.062	−0.059
	(0.116)	(0.094)	(0.067)	(0.089)
Size	0.878	0.878	0.881	0.881
	(0.983)	(0.791)	(0.566)	(0.754)
Density	−0.030	−0.030	−0.029	−0.028
	(0.013)	(0.011)	(0.008)	(0.010)
Stock	0.045	0.021	−0.077	−0.107
	(0.210)	(0.169)	(0.121)	(0.161)
Age	0.120	0.078	−0.089	−0.141
	(0.037)	(0.030)	(0.022)	(0.029)
Bm	−0.025	−0.024	−0.021	−0.019
	(0.044)	(0.035)	(0.025)	(0.034)
Observations	41,134	41,134	41,134	41,134
Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes

Note: This study selects the 10th, 25th, 75th, and 90th quantiles, where the lower quantiles (10th and 25th) represent lower levels and the higher quantiles (75th and 90th) represent higher levels. The values in parentheses indicate robust standard errors. *** represent significance at the 1% significance levels.

, when the level of intelligent transformation is low, the coefficient of Intelligent is positively statistically significant. As the level of intelligent transformation increases, the regression coefficients remain significantly positive, rising from 0.050 at the 10th quantile to 0.085 at the 90th quantile, indicating a notable intensification of the effect. This suggests that intelligent transformation plays a crucial role in driving technological progress and innovation. Particularly in environments characterized by advanced intelligent transformation, firms gain greater access to innovation resources and tools, leading to enhanced innovation capability. Moreover, the marginal effects of intelligent transformation show a clear upward trend, reflecting its increasing impact on firms’ innovative outcomes.

4.2. Endogenous Test

To address potential endogeneity issues caused by sample selection bias, this paper utilizes the Heckman two-step method and instrumental variables to modify the model. In the first stage, it conducts a probit regression using the explanatory variable as a dummy variable. If the value of the intelligent transformation exceeds the industry median for a given year, it is assigned a value of one, indicating a high degree of intelligent transformation; otherwise, it is set to zero. In addition, in the first stage, considering that the characteristic variables of the samples are not directly affected by the behavior of individual companies, but are closely related to the explanatory variables, this paper uses the mean values of other firms within the same industry, excluding the firm itself, as the instrumental variable. The results from this stage are used to calculate the inverse Mills ratio (IMR), which is subsequently incorporated into the second-stage model. As shown in Column (2) of

Table 5. Endogeneity and robustness test results.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	Innov	Innov	Innov2	Innov	Innov	Innov
Intelligent	0.180 ***	0.179 ***	0.161 ***	0.182 ***
	(12.891)	(12.706)	(12.339)	(12.867)
imr		0.334 ***
		(3.617)
LIntelligent					0.170 ***
					(11.404)
LIntelligent2						0.161 ***
						(10.171)
Sgrow	0.104	0.138	−0.202	0.139	0.221	0.282
	(0.643)	(0.852)	(−1.334)	(0.853)	(1.270)	(1.544)
Lev	−0.346 ***	−0.391 ***	−0.321 ***	−0.344 ***	−0.377 ***	−0.385 ***
	(−3.753)	(−4.181)	(−3.733)	(−3.722)	(−3.821)	(−3.614)
Size	2.994 ***	4.165 ***	2.872 ***	2.736 ***	3.945 ***	4.200 ***
	(4.001)	(4.985)	(4.141)	(3.578)	(5.022)	(5.082)
Density	−0.102 ***	−0.121 ***	−0.092 ***	−0.104 ***	−0.102 ***	−0.105 ***
	(−9.451)	(−9.904)	(−9.081)	(−9.523)	(−8.804)	(−8.524)
Stock	−0.027	−0.039	−0.053	−0.025	−0.010	0.013
	(−0.206)	(−0.303)	(−0.443)	(−0.194)	(−0.072)	(0.090)
Age	−0.219 ***	−0.246 ***	−0.188 ***	−0.225 ***	−0.311 ***	−0.339 ***
	(−11.216)	(−11.326)	(−10.357)	(−11.596)	(−12.199)	(−10.540)
Bm	0.194 ***	0.191 ***	0.186 ***	0.209 ***	0.158 ***	0.153 ***
	(5.020)	(4.941)	(5.177)	(5.320)	(3.955)	(3.595)
Constant	−8.578 ***	−12.349 ***	−8.274 ***	−7.867 ***	−11.019 ***	−11.669 ***
	(−4.058)	(−5.093)	(−4.223)	(−3.647)	(−4.968)	(−5.006)
Observations	41,132	41,132	41,132	36,661	35,750	31,233
Adjusted R-squared	0.328	0.328	0.340	0.328	0.337	0.334
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; *** represent significance at the 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

, the regression coefficient for Intelligent remains significantly positive at the 1% level, confirming the robustness of the positive relationship between intelligent transformation and firms’ innovation capability.

4.3. Robustness Test

To ensure the robustness of the regression results, this paper further verifies the impact of intelligent transformation on firms’ innovation capability by replacing the core variables, changing the samples, and performing lag tests. For the replacement of the core variables, the previous section measured firms’ innovation capability by the number of patents applied. In this section, the number of patents applied is replaced with the number of patents granted to represent the innovation ability of firms. This paper replaces samples by randomly removing data from the years 2011 and 2015. The estimation results, presented in Column (4) of Table 5, indicate that intelligent transformation continues to have a significant positive impact on firms’ innovation capability. Additionally, this paper examines the influence of intelligent transformation on the innovation capability by considering one-period and two-period lags, respectively. The results in Columns (5) and (6) of Table 5 indicate a significant enhancement in firms’ innovation capability due to intelligent transformation. These findings confirm the robustness of the benchmark regression results.

5. Further Analysis

5.1. The Mediating Role of Total Factor Productivity

Previous studies have widely demonstrated that improvements in total factor productivity will trigger innovative activities [65]. On the one hand, as the total factor productivity increases, firms can produce more products with the same resource input. This leads to lower production costs and mitigates capital misallocation [66], enabling firms to allocate more funds to R&D investment, thus fostering innovative activities. On the other hand, firms accumulate substantial knowledge and experience during the process of improving the production efficiency, which provides a foundation for future innovative activities. In summary, verifying the effect of intelligent transformation on firms’ productivity is sufficient to support hypothesis 3. When the explanatory variable is the total factor productivity measured by the LP method (Column (1) of

Table 6. The impact of intelligent transformation on total factor productivity.

	(1)	(2)	(3)
Variables	TFP_LP	TFP_OP	TFP_OLS
Intelligent	0.009 ***	0.009 ***	0.006 ***
	(4.365)	(3.683)	(3.865)
Sgrow	0.535 ***	0.778 ***	0.290 ***
	(18.400)	(23.229)	(12.267)
Lev	0.114 ***	0.128 ***	0.134 ***
	(8.839)	(8.638)	(12.829)
Size	13.951 ***	10.100 ***	18.825 ***
	(123.411)	(77.859)	(205.092)
Density	−0.355 ***	−0.264 ***	−0.371 ***
	(−218.417)	(−140.733)	(−281.398)
Stock	0.054 ***	0.060 ***	0.091 ***
	(3.597)	(3.487)	(7.491)
Age	−0.026 ***	−0.008 **	−0.025 ***
	(−8.402)	(−2.209)	(−10.241)
Bm	0.072 ***	0.034 ***	0.074 ***
	(12.723)	(5.208)	(16.260)
Constant	−34.586 ***	−24.352 ***	−47.401 ***
	(−109.747)	(−67.346)	(−185.237)
Observations	36,680	35,480	36,560
Adjusted R-squared	0.881	0.779	0.943
Year FE	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; ** and *** represent significance at the 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

), it can be found that the estimated coefficient of intelligence is significantly positive at the 1% level, which means that the intelligent transformation of firms will significantly improve their productivity. This is consistent with the previous analysis. TFP is further measured using the OP method and the OLS method, and the model is thus re-estimated, with the results shown in Columns (2) and (3). It can be found that the estimated coefficient of Intelligent is still significantly positive at the 1% level, indicating that the estimation results in the first column are robust. Based on the above analysis, hypothesis 3 is supported.

5.2. The Mediating Role of Market Competition

Market competition is widely regarded as a key factor in improving a firm’s performance and behavior. It is generally believed that competition pressures firms to lower costs, minimize managerial and operational inefficiencies, incentivize efficiency optimization, and foster innovation. Some scholars have developed two proxies to measure the market competition, namely, the market concentration and profit margin. Usually, a highly concentrated market is regarded as having a low level of competition. A substantial body of literature has verified the impact of market competition on innovation capability. Therefore, this paper only reports on the impact of intelligent transformation on market competition. The results are shown in

Table 7. The impact of intelligent transformation on market competition.

	(1)	(2)	(3)	(4)
Variables	Innov	Cr5	Cr10	Lerner
Intelligent	0.180 ***	−0.005 ***	−0.010 ***	−0.035 **
	(26.719)	(−2.795)	(−2.723)	(−2.474)
Sgrow	0.104	0.034	0.093 *	4.101 ***
	(1.037)	(1.388)	(1.793)	(19.474)
Lev	−0.346 ***	0.010	0.034	−0.001
	(−7.836)	(0.966)	(1.505)	(−0.015)
Size	2.994 ***	−0.057	−0.239	−3.787 ***
	(7.686)	(−0.611)	(−1.184)	(−4.650)
Density	−0.102 ***	−0.003 **	−0.005	0.047 ***
	(−18.746)	(−2.209)	(−1.603)	(4.126)
Stock	−0.027	−0.010	−0.032	−0.529 ***
	(−0.524)	(−0.847)	(−1.213)	(−4.974)
Age	−0.219 ***	0.005 **	0.012 ***	−0.034 *
	(−24.894)	(2.223)	(2.709)	(−1.869)
Bm	0.194 ***	0.003	0.010	0.170 ***
	(9.934)	(0.693)	(0.994)	(4.162)
Constant	−8.578 ***	0.563 **	1.518 ***	22.041 ***
	(−7.887)	(2.145)	(2.698)	(9.693)
Observations	41,132	41,059	40,923	41,132
Adjusted R-squared	0.328	0.779	0.779	0.709
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; *, **, and *** represent significance at the 10%, 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

. When the explained variable is the market concentration (Cr5), intelligent transformation is significantly negative at the 1% level. The new market concentration (Cr10) is further calculated by replacing the top five firms in Cr5 with the top ten firms, and the results remain unchanged. The results of Column (4) show that the regression results using the industry Lerner Index also reflect the same law of change, that is, with the improvement of the level of intelligent transformation, the market concentration decreases and the market competition becomes fierce. Hypothesis 4 is therefore verified.

5.3. Heterogeneity Analysis

5.3.1. Heterogeneity in Ownership

The previous literature has shown that there are differences in the innovation patterns of firms with different ownership [67]. This may cause intelligent transformation to have different effects in different ownership firms. To verify this difference, we divided the samples into SOEs and non-SOEs according to their equity nature [68]. The results presented in Column (1) of

Table 8. Results of the ownership heterogeneity test.

	SOEs	SOEs	non-SOEs	non-SOEs
Variables	Innov	Innov_Quality	Innov	Innov_Quality
Intelligent	0.095 ***	0.017 **	0.220 ***	0.009 **
	(3.637)	(2.202)	(13.935)	(2.408)
Sgrow	−0.069	−0.037	0.189	0.009
	(−0.235)	(−0.448)	(1.048)	(0.183)
Lev	−0.475 ***	−0.048	−0.295 ***	0.008
	(−2.815)	(−1.009)	(−2.791)	(0.271)
Size	4.564 ***	−0.641	2.108 **	−0.648 ***
	(3.162)	(−1.564)	(2.501)	(−2.805)
Density	−0.077 ***	0.007	−0.106 ***	0.012 ***
	(−4.148)	(0.905)	(−8.504)	(3.205)
Stock	−0.453 *	−0.001	0.159	−0.125 ***
	(−1.902)	(−0.009)	(1.049)	(−3.601)
Age	−0.193 ***	0.001	−0.257 ***	−0.024 ***
	(−4.746)	(0.111)	(−11.522)	(−4.230)
Bm	0.126 *	0.058 ***	0.244 ***	0.051 ***
	(1.657)	(2.744)	(5.602)	(4.838)
Constant	−12.919 ***	2.063 *	−6.172 ***	2.159 ***
	(−3.208)	(1.794)	(−2.579)	(3.292)
Observations	14,499	6806	26,632	17,353
Adjusted R-squared	0.377	0.152	0.307	0.249
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; *, **, and *** represent significance at the 10%, 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

illustrate the innovation effects in SOEs, while Column (3) shows the changes in the innovation capacity in non-SOEs. The regression coefficients indicate that intelligent transformation significantly enhances the innovation capacity of both SOEs and non-SOEs. Moreover, it is evident that the intelligent transformation coefficient for non-SOEs is greater than that for SOEs. This indicates that the effect of intelligent transformation on enhancing the innovation capabilities of non-SOEs is significantly stronger than that on SOEs. This finding is similar to that of Han et al. (2023) [14]. We further explored the impact of intelligent transformation on the innovation quality in different forms of equity. The results are shown in Columns (2) and (4). From the results, we can see the effect of intelligent transformation on the exploratory innovation capability of SOEs is significantly greater than that of non-SOEs.

The reason for this phenomenon may be that the government has substantial influence over resources, including land, financial capital, bank loans, and subsidies, in developing nations like China [69]. The government grants these firms significant market leverage and preferential access to scarce resources because it is the biggest stakeholder in SOEs. Such support alleviates resource constraints for innovation investment and mitigates uncertainty risks. In addition, in order to preserve their political legitimacy as extensions of governments and their agencies, SOEs place a high priority on advancing national goals [70]. Recognizing that innovation drives sustainable economic growth, the Chinese government has prioritized indigenous innovation in its national development plan since 2006, implementing supportive policies [71]. Such regulatory pressures make SOEs the primary entities for undertaking social responsibilities, including engaging in innovative activities [72]. This is especially true for exploratory innovations that are risky and have a long payback period. Moreover, since SOEs are owned and managed by the government, their senior managers often act like quasi-officials, prioritizing political promotion linked to firm performance. Innovation serves as a crucial assessment criterion for SOE managers, termed “responsible innovation” [69]. In contrast, as commercial organizations, non-SOEs have stronger incentives for exploratory innovation, but fewer resources to mobilize. At the same time, the maximization of self-interest primarily guides firms’ behaviors [73]. Given these differing motivations and operational principles, SOEs are more likely to use the dividends of intelligent technology to carry out exploratory innovation.

5.3.2. Heterogeneity in Firm Sizes

Firm size is closely related to R&D and invention expenditures [47]. Therefore, this section explores the heterogeneities among different firm sizes. We used a sample operating income of USD 400 million to distinguish between large and small firms [74]. The results in

Table 9. Results of the firm size heterogeneity test.

	Large	Large	Small	Small
Variables	Innov	Innov_Quality	Innov	Innov_Quality
Intelligent	0.184 ***	0.010 **	0.165 ***	0.010
	(12.031)	(2.558)	(7.395)	(1.252)
Sgrow	0.108	−0.051	−0.454 *	0.200
	(0.609)	(−1.096)	(−1.806)	(1.379)
Lev	−0.359 ***	−0.000	−0.365 ***	−0.044
	(−3.483)	(−0.012)	(−2.745)	(−0.597)
Size	2.766 ***	−0.709 ***	3.999 ***	0.916
	(3.108)	(−3.267)	(3.118)	(1.088)
Density	−0.114 ***	0.007	−0.049 ***	0.017 *
	(−7.922)	(1.622)	(−2.979)	(1.767)
Stock	−0.001	−0.069 **	−0.115	−0.049
	(−0.010)	(−2.176)	(−0.649)	(−0.642)
Age	−0.228 ***	−0.004	−0.250 ***	−0.028 **
	(−10.632)	(−0.735)	(−8.296)	(−2.146)
Bm	0.209 ***	0.059 ***	0.021	−0.009
	(4.967)	(5.885)	(0.362)	(−0.298)
Constant	−7.949 ***	2.268 ***	−10.715 ***	−2.274
	(−3.145)	(3.694)	(−2.911)	(−0.923)
Observations	36,399	21,579	4731	2575
Adjusted R-squared	0.328	0.208	0.360	0.239
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; *, **, and *** represent significance at the 10%, 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

indicate that intelligent transformation significantly enhances the innovation capability of both large and small firms, and it has a stronger effect on enhancing the large firms’ innovation capability (see Column (1) and Column (3)) [75]. Furthermore, as shown in Columns (2) and (4), intelligent transformation significantly enhances the exploratory innovation capability of large enterprises and improves the quality of innovation. However, this transformation does not appear to have a significant effect on the exploratory innovation capability of small enterprises. Possible reasons include that large firms can bear higher risks and undertake bolder innovation initiatives due to their relatively stable financial positions and substantial resource advantages. They are more likely to invest in basic and applied research, as well as technological development [61]. In contrast, small enterprises often opt for low-risk innovation strategies when faced with resource constraints. Additionally, the competitive advantages that large firms hold in the market allow them to capitalize on the technological benefits of intelligent transformation, enabling more effective entry into new markets and the development of innovative products. Conversely, small firms typically concentrate on specific market segments; while they may achieve innovation within certain niches, their overall exploratory innovation capability remains relatively limited, resulting in a lack of significant improvement in this area.

5.3.3. Heterogeneity in Factor-Intensive Types

Labor-intensive industries are those that rely mainly on a large number of labor inputs in the production process, and which rely relatively less on technology and equipment [76]. Capital-intensive industries are those that require large capital inputs for the purchase of advanced production equipment, and the construction of large-scale plants and infrastructure, while the labor inputs are relatively small in the production process. It is not difficult to find that there are significant differences between the two in terms of the technological foundation, labor quality, capital investment, and innovation focus. This difference will influence the difficulty of promoting intelligent transformation, the support of talents, and the strength of investment, which, in turn, will affect the release of its innovation effect. Therefore, we divided the study sample into two categories of labor-intensive and capital-intensive firms based on the logarithm of the ratio of net fixed assets to the number of employees. Enterprises with factor intensity above the median were classified as capital-intensive, while those below the median were classified as labor-intensive [77]. We then further analyzed the impact of intelligent transformation on the innovation capability.

Columns (1) and (3) of

Table 10. Results of the factor-intensive industries heterogeneity test.

	Labor-Intensive	Labor-Intensive	Capital-Intensive	Capital-Intensive
Variables	Innov	Innov_Quality	Innov	Innov_Quality
Intelligent	0.165 ***	0.021 ***	0.176 ***	−0.001
	(8.610)	(4.704)	(10.057)	(−0.115)
Sgrow	0.443 *	−0.082	−0.259	0.052
	(1.846)	(−1.337)	(−1.257)	(0.860)
Lev	−0.393 ***	−0.021	−0.266 **	0.018
	(−2.988)	(−0.641)	(−2.284)	(0.546)
Size	3.696 ***	−0.392	2.604 ***	−0.811 ***
	(3.386)	(−1.471)	(2.860)	(−2.920)
Density	−0.095 ***	0.021 ***	−0.099 ***	−0.001
	(−6.357)	(4.285)	(−7.115)	(−0.290)
Stock	−0.079	−0.079 **	−0.032	−0.019
	(−0.444)	(−2.004)	(−0.207)	(−0.488)
Age	−0.223 ***	−0.004	−0.219 ***	−0.010
	(−8.124)	(−0.641)	(−8.994)	(−1.462)
Bm	0.214 ***	0.048 ***	0.190 ***	0.056 ***
	(3.855)	(3.733)	(4.023)	(4.361)
Constant	−10.718 ***	1.297 *	−7.448 ***	2.653 ***
	(−3.462)	(1.726)	(−2.903)	(3.382)
Observations	19,721	12,080	21,353	12,063
Adjusted R-squared	0.342	0.257	0.317	0.170
Year FE	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes

Note: Cluster robust standard errors are in parentheses; *,**, and *** represent significance at the 10%, 5%, and 1% significance levels, respectively, and the regression models control both industry and year fixed effects.

indicate that intelligent transformation significantly enhances the innovation capability, particularly in capital-intensive industries, where the innovation effect is the strongest, with an influence coefficient of 0.176. However, intelligent transformation does not significantly improve the substantial innovation capability of capital-intensive enterprises. This phenomenon can be explained as follows: Labor-intensive enterprises typically depend on substantial human resources to execute production tasks. Intelligent transformation enhances the production efficiency by streamlining workflows through the integration of automation and intelligent systems. This transition enables employees to shift their focus toward creative problem solving and innovation. In contrast, capital-intensive enterprises predominantly rely on significant capital investments to fuel their operations. However, the nature of such investments often compels these enterprises to prioritize immediate enhancements and upgrades to existing products. As a result, intelligent transformation tends to foster utilization-based innovations that leverage current technologies and markets rather than encouraging high-risk exploratory innovations.

6. Conclusions and Discussion

6.1. Conclusions

Based on the above analysis, this paper reveals the relationship between intelligent transformation and both firms’ innovation capability and innovation quality. The findings are supported by panel data from 5000 Chinese A-share listed firms between 2008 and 2022. In comparison to the literature, this paper highlights several of the following key findings: (1) Intelligent transformation significantly enhances the innovation capability. As the level of intelligent transformation increases, the driving effect of intelligent technologies on improving the innovation capability becomes more pronounced. (2) Intelligent transformation is beneficial for improving firms’ exploratory innovation capabilities. This finding remains robust after a series of robustness checks and endogeneity tests. (3) The increase in the TFP and the intensified market competition are the crucial paths for intelligent transformation to affect the firm’s innovation capacity and innovation quality. (4) The marginal effect of intelligent transformation on innovation capacity improvement is significantly large for non-SOEs compared to SOEs, and SOEs are more likely to benefit from exploratory innovation. Larger firms and labor-intensive firms are better able to benefit from the innovation effects of intelligent transformation.

6.2. Research Implications

6.2.1. Theoretical Implications

This research has theoretical implications in two aspects. Firstly, we confirm that intelligent transformation promotes innovation capacity and quality, expanding the studies on intelligent transformation and innovation. Previous studies have extensively discussed the impact of Investment in new technologies on innovation [78]. The Fourth Industrial Revolution is advancing rapidly. In recent years, the focus of China’s new technology development has gradually shifted from infrastructure investment to the intelligent transformation represented by industrial robots. By examining the effects of intelligent transformation, this study broadens the research perspective on innovation. The important role of intelligent transformation in the fourth wave of the Industrial Revolution is emphasized.

Secondly, the mechanism analysis part of this research reveals how intelligent transformation affects firms’ innovation, identifying the TFP and market competition as important pathways. On the one hand, this is an important complement to resource-based theory, which emphasizes acquiring resources, as well as sufficient financial and administrative support, for innovation [35]. Our findings also point out that an increase in the level of market competition achieves the same effect. On the other hand, the mechanism analysis part also proves the objective fact that intelligent transformation forces firms to participate in market competition and ultimately improve their level of innovation, which has not been mentioned in previous discussions on the theory of market competition.

6.2.2. Practical Implications

The practical implications of this research are as follows: Firstly, the heterogeneity analysis reveals that the magnitude of the innovation effect of intelligent transformation is influenced by firm ownership, firm size, and industry characteristics. This finding provides a new perspective for fully understanding the impact of intelligent transformation on corporate innovation, emphasizing the interference of the above factors on the innovation effect of intelligent transformation, and suggesting that there is no uniform pattern for achieving leapfrogging in corporate innovation levels through intelligent transformation. This provides a practical basis for managers to formulate differentiated innovation-level enhancement strategies.

Secondly, this study provides a complete empirical analysis framework, including research hypotheses, data integration, and econometric analysis, which can serve as a reference for subsequent research on intelligent transformation and corporate innovation. The main conclusions of this paper can reflect the relationship between intelligent transformation and enterprise innovation in China, and can also provide practical references for other developing countries to formulate relevant policies.

6.3. Policy Implications

In view of these findings, we propose the following policy recommendations: First, the government should actively promote intelligent technology across various industries and increase the level of intelligent transformation. This strategy will leverage the benefits of enhanced production efficiency and technology spillover effects. By providing sufficient resources and knowledge support, firms will be better positioned to enhance their innovation capability.

Second, it is crucial to capitalize on the innovative effects of intelligent transformation in large firms, particularly by leveraging the leading role of SOEs. Large firms possess greater resources and stronger capabilities, enabling them to maximize the transformation of innovative outcomes. Policymakers should focus on providing incentives and fostering collaboration to concentrate innovation resources on high-quality firms and products. Encouraging large firms to lead major national research projects will fully unleash their innovative potential in intelligent transformation.

Third, it is essential to strengthen innovation incentives for small- and medium-sized enterprises (SMEs). Financing constraints and inadequate channels for the commercialization of scientific and technological achievements significantly undermine the innovation drive of SMEs. To address this, the government could implement initiatives such as innovation subsidies, tax breaks, and targeted talent subsidies, encouraging SMEs to actively participate in innovation-related activities.

Fourth, it is crucial to cultivate a conducive environment for market competition. The advancement of intelligent technology can intensify market competition, but excessive competition can distort market mechanism, disrupt order, promote short-termism, and hinder innovation. To address this, the government should rigorously implement the fair competition review system, ensuring equal treatment for all market participants, and create a level playing field. Furthermore, strengthening the enforcement of anti-monopoly and anti-unfair-competition regulations will help prevent dominant players from engaging in unfair practices that stifle competition, thereby cultivating an environment that prompts both innovation and competition.

6.4. Limitations and Future Research

Despite the contributions of this paper, further work is needed. First, our study was conducted based on Chinese firms, and there may be limitations in the representation of the sample. Future research will give more consideration to the differences in the impact of intelligent transformation on the innovation capability between the world’s emerging markets and mature markets, as well as in different political and cultural contexts. Second, due to limited data availability, we discussed the innovation effects of intelligent transformation only in the context of patents. Future research will focus on considering the complexity of firms’ innovation and will design a new indicator to reflect this complexity.