Latest / 1 More Paper / Factors influencing generative artificial intelligence adoption in Vietnam’s banking sector: an empirical study — Youssef et al. (2025)
Transcript
- You’re listening to “Factors influencing generative artificial intelligence adoption in Vietnam’s banking sector: an empirical study,” by W. Ajili Ben Youssef, N. Bouebdallah, and H. Long. Published in 2025. Abstract. This study aims to identify the key factors influencing the adoption of generative AI (GenAI) by Vietnamese banks and highlight the challenges and opportunities in digital transformation. It extends the technology-organization-environment (TOE) framework to incorporate GenAI-specific factors in the Vietnamese banking sector, characterized by rapid digitization and stringent regulations. A survey yielded 236 valid responses. The data were analyzed via partial least squares structural equation modeling (PLS- SEM). The key factors identified include organizational readiness (OR), compatibility (CPT), competitive pressure (CP), complexity (CPL), relative advantage (RA), firm size (FS), and government support (GS). OR emerged as the most influential factor because of a robust IT infrastructure and skilled personnel. CPT and CP were also significant, driving banks to adopt GenAI for a competitive edge. However, CPL presents challenges, requiring simpler AI solutions and clear risk mitigation policies. This study enhances the understanding of GenAI adoption within the Vietnamese banking sector, emphasizing the importance of tailored strategies for different bank sizes and the critical role of technology readiness for effective integration. The findings provide actionable insights into banks navigating their digital transformation journeys. Introduction. Generative artificial intelligence (GenAI) is an emerging technology that gener-ates content such as text, images, and audio. It has gained popularity through advancements in generative pretrained transformer (GPT) models, particularly GPT-3 and GPT-4, which were developed by OpenAI. GenAI uses deep learning methods such as generative adversarial networks (GANs) and variational autoencoders (VAEs) to create human-like content. These systems, built on sophisticated neural networks and large-scale models, have transformative applications across various fields, including content creation and decision support. However, they also present risks related to misinformation, bias, privacy, and environmental impact risks, necessitat-ing careful policy guidance. The adoption of GenAI is crucial for its ability to revolutionize industries by auto-mating content generation and enhancing decision-making processes. Previous research has highlighted its potential in various sectors, including banking, where it can be used for robot advisers, chatbots, and advanced AI applications in fraud detec-tion and antimoney laundering (AML) measures. GenAI enhances virtual assistance services by guiding customers through various banking processes and providing 24/7 support. The ability of GenAI to process vast amounts of data in near real time in the back office supports bank management in making informed decisions. This technology increases the competitiveness of banks that embrace and invest in it. Vietnam’s GenAI market is projected to grow annually by 46.47% from 2024 to 2030 (Statista Market Insights). A 2023 Finastra survey revealed that 91% of Vietnamese bank executives are interested in GenAI. High-profile events such as “The Future of Generative AI 2023” further accelerated GenAI adoption in Vietnam. However, challenges remain, particularly regarding the regulatory framework and regional talent. Data governance and the integration of domain knowledge influence GenAI adoption in Vietnam’s banking sector. Liu et al. (2023) highlighted the importance of man-aging data quantity to enhance machine learning performance, which is crucial for GenAI applications. Xu et al. (2023) discuss advancements in generative models that improve GenAI’s capabilities in complex banking tasks. These insights emphasize the need for robust data management and advanced generative techniques for effective GenAI adoption in Vietnam’s banking sector. In 2023, Vietnam’s banking sector had 49 active commercial banks with total assets exceeding $565 billion (State Bank of Vietnam, 2023). Despite significant digital transformation efforts, the application of GenAI in Vietnamese banks remains limited due to a lack of empirical research on critical adoption factors. This study aims to fill that gap by identifying key determinants of GenAI adoption in Vietnamese banks and using empirical analysis to guide these institutions in making informed decisions about investing in GenAI projects. This study is crucial because it explores the factors influencing the adoption of GenAI in Vietnam’s banking sector, a rapidly growing market with significant potential for technological advancement. By identifying key drivers and barriers, this research provides valuable insights for banks aiming to implement AI-driven technologies effectively, thereby enhancing operational efficiency and promoting financial inclusion. This study addresses several research gaps by focusing on GenAI adoption. First, it expands the technology-organization-environment (TOE) framework to include factors specific to GenAI adoption in the banking sector. Second, while there has been significant research on AI adoption, few studies specifically address the unique context of Vietnam’s banking industry, which is characterized by increasing digitalization and stringent government regulations. This research breaks new ground by examining the factors influencing GenAI adoption in this region. Finally, the research offers actionable insights and strategic recommendations for Vietnamese banks, helping them navigate the complexities of GenAI implementation and leverage the technology to improve their services and competitiveness. Insights from Vietnam’s experience can offer valuable lessons for other emerging mar-kets, expanding the appeal of this research to a broader audience interested in global banking trends. This study highlights Vietnam’s distinct challenges and opportunities and provides a framework for understanding GenAI adoption in similar contexts worldwide. The remainder of the paper is structured as follows: Sect. "Literature review" provides a comprehensive literature review. Sect. "Conceptual framework and hypotheses" outlines theconceptual framework and discusses the hypotheses. Sect. "Methodology" describes the research methodology. Sect. "Discussion"presents the discussion and offers recommendations. Literature review. Theoretical framework The theoretical literature on new technology adoption is based on innovation diffusion theory (IDT), the technology acceptance model (TAM), and the technology–organization–environment (TOE) framework. Innovation diffusion theory (IDT) Introduced in 1962 and refined by Rogers in 1995, IDT explains how new ideas or inno-vations spread across a population over time. The diffusion process relies on five fun-damental attributes: relative advantage, compatibility with existing values, complexity, trialability, and observability. Furthermore, adopters of inno-vations can be categorized into five groups on the basis of the speed of their acceptance: innovators, early adopters, early majority, late majority, and laggards. IDT has evolved to include a multilayered approach that considers macro, meso, and micro levels of societal penetration. This layered framework enhances the acceleration of innovation adoption in both individual and organizational contexts. Technology acceptance model (TAM) The TAM focuses on individual users’ acceptance of technology and highlights two primary factors: perceived usefulness (PU) and perceived ease of use (PEOU). These criteria significantly affect how users adopt new technology. The TAM model has been further updated into the unified theory of acceptance and use of technology (UTAUT) by adding factors such as social influence and performance expectancy. While TAM and UTAUT effectively explain individual-level adoption, they may not sufficiently address the organizational complexities of a heavily regulated banking industry. The TAM effectively outlines the technology adoption factors from individual perspectives, providing further explanations and acceptances through the UATUT model to provide a broader view of the elements influencing individual innovation adoption. When these two theories are applied in the Vietnamese banking sector, the model still fails to account for the nature of the top-down and regulatory strict aspects of the banking industry, which are, on the other hand, well catered to by the TOE framework . Although comprehensive, IDT’s coverage of the adoption process may oversimplify the complex interplay of factors in highly regulated environments. Technology–organization–environment framework (TOE) The TOE framework outlines three key contexts that influence innovation adoption: technological, organizational, and environmental contexts. The technological context encompasses a firm’s internal and external technological resources and innovations. The organizational context refers to a firm’s characteristics and resources, including its size, scope, and management structure. The environmental context involves external surroundings, which include sector characteristics, the regulatory landscape, and competition. Despite its inception decades ago, the TOE framework remains robust and adaptable to evolving information systems. Recent empirical studies have validated their effective-ness in explaining technology adoption across diverse organizational settings. Empirical studies Business sectors, particularly banking, are receiving increasing amounts of academic attention with respect to the adoption of GenAI technologies. Kruse et al. (2019) reported that regulatory obligations and competitive pressures drive AI adoption in Germany’s financial services sector, highlighting how external factors can shape technological transformation within established institutions. This regulatory-driven approach contrasts with Gupta’s (2024) examination of GenAI in entrepreneurship, emphasizing the importance of entrepreneurs recognizing the usefulness and ease of use of AI technology. These differing perspectives reflect the tension between institutional constraints and individual perception factors that collectively shape GenAI adoption patterns across diverse business environments. Factors influencing AI acceptance Various factors influence the acceptance of AI technologies, including organizational readiness, leadership support, customer desires, and regulatory demands. According to Almashawreh (2024), SMEs require technological readiness and strong support from top management for successful AI implementation. Although the study focuses on Jordanian SMEs, its findings are particularly relevant for banks in Vietnam, as these institutions also need a solid technological foundation and leadership back to implement AI solutions effectively. Al-Khatib (2023) estimated that the most important elements behind the integration of GenAI in Jordan’s retail market are organizational readiness, relative advantage, and management support. Kruse et al. (2019) further reinforce this conceptual framework, arguing that empowering a culture of organizational readiness and strong management support is critical for GenAI implementation in the financial sector. To address the challenges of adopting generative AI, Gupta (2024) suggested that companies prioritize ease of use, shifting focus from organizational to user-level concerns. This means that employees can easily integrate GenAI technologies into their daily tasks. Achieving this requires ongoing training programs and a user-centered approach to adoption within the banking sector—an important consideration that bridges the gap between institutional readiness and practical implementation. GenAI in the Vietnamese banking context As a new technology, GenAI is gaining attention in the banking industry for its potential and various use cases as a game-changing innovation. However, the literature has limitations in outlining the factors influencing GenAI adoption in the Vietnamese banking sector, with most studies focusing either on conventional AI or broader regional trends without the specificity needed for this unique market. Nguyen et al. (2022a, b) reported that adopting AI in Vietnamese enterprises, including banks, is motivated by technological preparedness and organizational support. Competitive pressures and the regulatory environment also play significant roles in AI implementation. While not exclusive to the banking sector, this study emphasizes the importance of clear regulations and a robust technological foundation for adopting GenAI in Vietnam. However, Nguyen’s (2022a) study leaves unexplored the specific mechanisms by which these factors operate within Vietnamese banking institutions, revealing a crucial theoretical and practical gap in the current understanding. Vietnamese‐specific findings Anh (2024) examined the financial services sector in Vietnam and reported that perceived usefulness and ease of use are critical mediators. These findings, relevant to Vietnamese banks, align with international data from Gupta (2024), suggesting that some universal principles of technology adoption transcend cultural and institutional bound-aries. However, the prevailing focus on conventional AI, rather than GenAI, limits its broader applicability and raises questions about whether traditional technology acceptance models adequately capture the unique characteristics of generative technologies. Technological readiness has consistently emerged as a recurring theme in various studies on AI adoption in Vietnam. According to Nguyen et al. (2022a, b), an organization’s preparedness to embrace new technologies and solid organizational support are essential for successfully integrating AI into business operations, highlighting the importance of preadoption organizational conditions. Anh (2024) extends this framework by demon-strating that perceived usefulness and ease of use influence adoption decisions in financial services. These elements are essential for banks, as employees need to find GenAI applications that are valuable and easy to use to ensure a smooth workflow. According to Nguyen et al. (2022a, b), competitive pressures and the regulatory environment are key external motivators for adopting AI. Banks in Vietnam operate within a competitive and regulated market where advancements in AI can offer a significant advantage. This highlights the necessity for Vietnamese banks to consider internal and external factors when developing strategies for adopting GenAI, suggesting a more comprehensive adoption framework than is typically addressed in the literature. The literature suggests several evidence-based approaches for promoting GenAI adoption in Vietnamese banks. According to Nguyen et al. (2022a, b) and Anh (2024), banks should improve their technological readiness by enhancing their existing IT infrastructure and ensuring that their staff are trained to work with GenAI technologies. The findings of both studies show that success in AI adoption largely depends on readiness, suggesting a sequential approach to implementation where foundational capabilities precede advanced applications. Kruse et al. (2019) and Gupta (2024) emphasized the importance of top management support in fostering an environment that promotes AI innovation within organizations. Banks should prioritize building a leadership team that has a solid understanding of AI and is committed to its integration across various business functions. Finally, Vietnamese banks should work closely with regulators to ensure that their adoption strategies conform to current regulations. As Nguyen et al. (2022a, b) reported, the regulatory framework strongly affects AI adoption, particularly in the highly regulated banking sector, where compliance concerns can accelerate or inhibit innovation. This interplay between technological capability, leadership vision, and regulatory navigation represents a comprehensive framework for understanding GenAI adoption that extends beyond simple technology acceptance models. Research gap Our literature review demonstrates that GenAI adoption in banking is influenced by a complex interplay of technological factors (readiness, perceived usefulness, ease of use), organizational elements (leadership support, organizational culture), and environmental considerations (regulatory framework, competitive pressures). While studies from diverse contexts, including Jordan, Germany, and Vietnam, have examined various aspects of AI adoption, a significant research gap exists regarding the specific dynamics of GenAI implementation in Vietnamese banking. Existing research has focused primarily on conventional AI rather than GenAI’s distinct characteristics and implementation challenges, representing a fundamentally different technological paradigm with unique adoption considerations. Furthermore, there is a limited understanding of how Vietnam’s banking sector’s cultural and institutional specificities might moderate the established adoption factors identified in international contexts. This gap is particularly consequential given the transformative potential of GenAI for banking operations and the strategic importance of effective adoption strategies in this rapidly evolving technological landscape. Our research aims to address these limitations by developing a contextually sensitive framework tailored explicitly to GenAI adoption in Vietnamese banking institutions. Conceptual framework and hypotheses Conventional AI involves systems designed for specific tasks, such as rule-based systems and traditional machine learning models, which excel in predefined tasks but lack flexibility. In contrast, with advancements in generative pretrained transformers (GPTs) and other deep learning models, GenAI can autonomously generate new content such as text, images, and audio by learning patterns from large datasets. Our study focuses on GenAI adoption in Vietnam’s banking sector, highlighting its unique capabilities and potential applications. By examining factors specific to GenAI adoption, we provide insights not covered by previous studies on conventional AI, emphasizing the transformative potential of GenAI and the need for tailored strategies to facilitate its adoption in the banking industry. The TOE framework can measure technological, organizational, and environmental factors, making it a better framework for the banking sector. Moreover, the banking industry is also part of a complex ecosystem where regulatory and technological advancements often overlap, and the TOE framework can incorporate these diverse factors to understand the adoption process in many contexts. The TOE framework clearly outlines the key factors influencing the adoption of GenAI and serves as a foundation for developing strategies to incorporate these technologies into banking systems. Technology context Technological advancements and readiness within the banking sector, as well as innova-tive technology itself, play critical roles in the adoption of GenAI. Relative advantage Relative advantage (RA) is crucial in the adoption of new technologies. RA refers to the perceived benefits of an innovation over existing alternatives. Organizations are more likely to adopt technologies that offer significant advantages. For example, Nguyen et al. (2022a, b) emphasized that relative advantage is essential in technology adoption. Al-Khatib (2023) noted that the perceived benefits of GenAI, such as increased operational efficiency, drive its adoption across various sectors. Additionally, the advantages of technology are closely linked to support from a firm’s management, which can accelerate its adoption. The following hypothesis is derived: H1: Relative advantage (RA) has a positive effect on GenAI adoption in Vietnamese banks. Compatibility Compatibility (CPT) is a crucial factor in the adoption of technology. It refers to how well a technological innovation aligns with a firm’s existing values, experiences, and needs of its adopters. Nguyen et al. (2022a, b) emphasized the importance of technical compatibility in Vietnamese enterprises adopting AI, noting that higher compatibility can lead to reduced implementation costs and shorter timeframes. Awa et al. (2016) noted that compatibility is crucial in adopting ERP systems within the TOE framework. The alignment of new technologies with existing infrastructure and processes plays a significant role in their uptake. Additionally, if new technology does not align with the current tech stack, traction is unlikely to be achieved. On this basis, we propose the following hypothesis: H2: Compatibility (CPT) has a positive effect on GenAI adoption in Vietnamese banks. Complexity Complexity (CPL) refers to the challenges associated with implementing technology. More complex and new technologies often lead to lower adoption rates. Additionally, complex technologies require more time, resources, and expertise, which can prevent organizations from adopting them. Owing to its novelty and the short-age of skilled technical expertise, the technical challenges associated with AI hinder its adoption in Vietnamese firms. Additionally, high maintenance costs and the potential for downtime make computer programming languages (CPLs) a barrier to technology adoption. Consequently, we propose the following hypothesis: H3: Complexity (CPL) has a negative effect on GenAI adoption in Vietnamese banks. Organization context The organizational structure, available resources, and level of management support play crucial roles in the successful implementation of GenAI. Organizational readiness Organizational readiness (OR) is critical in successfully implementing new technologies within companies. OR refers to the extent to which an organization possesses the financial resources and human skills to effectively integrate innovation. In the context of Vietnamese enterprises, OR can significantly influence the adoption of AI technologies. Companies with robust technological infrastructure and a skilled workforce are better positioned to adopt AI tools. Therefore, the following hypothesis can be proposed: H4: Organizational readiness (OR) has a positive effect on GenAI adoption in Vietnamese banks. Firm size Firm size (FS) is a significant factor influencing the adoption of new technologies within organizations. Larger firms typically have more resources, which can facilitate the implementation of complex technologies. Awa (2016) highlighted that firm size plays a crucial role in technology adoption, suggesting that more prominent firms are more likely to embrace new technology because of their ability to manage integration challenges. Nguyen et al. (2022a, b) also emphasized the importance of firm size in AI integration among Vietnamese companies. Larger firms are better equipped to navigate the difficul-ties of adopting new technologies. Therefore, we can formulate the following hypothesis: H5: Firm size (FS) has a positive effect on GenAI adoption in Vietnamese banks. Environment context External factors, including regulatory frameworks and competitive pressures, are pivotal in shaping the adoption of generative AI in the banking industry. Competitive pressure Competitive pressure (CP) is a significant factor that drives organizations to adopt inno-vations, including GenAI. CP refers to maintaining or improving a market position, enhancing operational efficiency, or meeting customer expectations. It motivates firms to embrace innovation to remain competitive. Additionally, Nguyen et al. (2022a, b) demonstrated that CP notably influences Vietnamese companies’ adoption of AI. To remain competitive, companies must integrate with GenAI. This leads to the development of the following hypothesis: H6: Competitive pressure (CP) has a positive effect on GenAI adoption in Vietnamese banks. Government support According to Nguyen et al. (2022a, b), the decision of organizations to adopt innovation is significantly influenced by government support (GS). Vietnamese companies are more inclined to implement AI solutions when they receive assistance from the government. The literature indicates that government support can take various forms, including infrastructure development, regulatory frameworks, and financial subsidies. Therefore, the following hypothesis is proposed: H7: Government support (GS) has a positive effect on GenAI adoption in Vietnamese banks. Conceptual model The proposed research model presented in Fig. 1 identifies the key determinants affect-ing GenAI adoption by Vietnamese banks. It categorizes the factors into three primary contexts: technology context, including relative advantage (RA), compatibility (CPT), and complexity (CPL); organizational context, including organizational readiness (OR) and firm size (FS); and environmental context, including competitive pressure (CP) and government support (GS) factors. Each of these factors is hypothesized to impact GenAI adoption. These hypotheses provide a structured framework for understanding how technological, organizational, and environmental elements can influence GenAI implementation in Vietnam’s banking sector. Methodology. Sample and data collection In our study, we used a survey to collect primary data. This quantitative approach ena-bles us to measure latent variables and analyze their relationships effectively. Given the need to examine adoption factors across various banks, this method guarantees a comprehensive understanding of the dynamics involved. The data collection period lasted from September 1 to September 19, 2024. Dur-ing this period, we gathered 248 survey responses. Following an initial review, we Fig. 1 The conceptual model excluded 12 incomplete responses, resulting in a final sample size of 236 valid survey results. We administered the questionnaires via Google Forms, facilitating easy access and data collection. This robust dataset is the foundation for further exploration of GenAI adoption in the Vietnamese banking sector. Partial least squares structural equation modeling (PLS-SEM) was employed to ana-lyze the data via Smart PLS-4 software. PLS-SEM is particularly appropriate for our study, as it fulfills several criteria Hair et al. (2014) outlined in their meta-analysis of management science methods. This method is resilient to the multivariate normal-ity assumption and is effective with relatively small sample sizes. Our sample of 236 observations meets the minimum requirement set by the tenfold rule method, further validating our analytical approach. Measurement items The development of our measurement scales was grounded in a comprehensive review of the literature to ensure content validity. We adapted items from established scales used in previous studies and tailored them to the context of GenAI adoption in the Vietnamese banking sector. Expert validation was conducted by consulting with two industry professionals and an academic expert in AI and banking. Their feedback was instru-mental in refining the items for clarity and relevance. Additionally, we conducted a pilot test with a sample of 20 respondents from the target population to assess the reliability and validity of the scales. The pilot test results were analyzed via Cronbach’s alpha and exploratory factor analysis (EFA), leading to further adjustments to ensure the robustness of the measurement instruments. This iterative process ensured that our scales were reliable and valid for the context of the study. To ensure the robustness of our measurement items, we anchored them in the TOE framework and relevant literature. The items were organized into coherent questions to capture the constructs related to generative AI adoption accurately. We developed the questionnaire via a 5-point Likert scale, which gauges the respondent’s agreement with each statement. To enhance the clarity and relevance of the questionnaire, we sought input from banking experts to guarantee that the language and content aligned with our research objectives. The measurement items are shown in Table 1. Sample analysis The demographic characteristics of the 236 respondents are summarized in Table 2. Frequency analysis reveals a diverse respondent group in terms of experience, job title, bank type, and company size. Most respondents have substantial professional experience, with more than 75% having more than six years in the field. Additionally, the sample predominantly comprises nonexecutives (80.5%), which suggests focusing on operational rather than strategic viewpoints. The representation of various types of banks—state-owned, private, and foreign—provides a comprehensive understanding of Vietnam’s banking landscape. Methodology limitations A limitation of our methodology is the reliance on a quantitative questionnaire with a Likert scale to assess factors influencing GenAI adoption in banking. While this approach allows for structured data collection, it can oversimplify complex attitudes and perceptions. Respondents may select neutral or extreme options without fully reflecting their views, leading to potential bias. Furthermore, the Likert scale fails to capture the nuances of individual experiences or contextual factors influencing decision-making. This limitation could restrict our understanding of the multifaceted nature of GenAI adoption, as it overlooks qualitative insights that might provide a deeper understanding of the motivations and barriers faced by financial institutions. Results and discussion Reliability and collinearity analysis On the basis of the survey results, we conducted a reliability analysis to examine the alignment of measurement items with factors influencing GenAI adoption. The assess-ment of the quality of the measurement model focused on indicators of reliability and validity. The outcomes of the reliability analysis are presented in Table 3. Loading factors and Cronbach’s alpha were used to assess reliability. The loading factor, which should be greater than 0.70, is the correlation between the manifest variables and the latent variable of interest. The questionnaire’s Cronbach’s alpha coefficient must exceed 0.70 for all parts; therefore, it is considered reliable. The results indicate that the measurement items demonstrate good internal consist-ency across each factor, with Cronbach’s alpha values generally exceeding 0.7. Government support (GS) exhibited the highest consistency, with a value of α = 0.825. These findings validate that the measurement items accurately rep-resent the key factors influencing GenAI adoption in Vietnamese banks. Table 3 shows that the indicators for GenAI adoption met this criterion (CR = 0.849). Convergent validity was assessed on the basis of the AVE indicator and had to be greater than 0.5. The AVE of GenAI adoption was 0.585, thus exceeding the minimum and suggesting good convergent validity. Before testing the hypotheses, we applied variance inflation factor (VIF) values to investigate potential collinearity problems. Table 4 shows that all the VIF values are less than 3, indicating the absence of collinearity issues. Discriminant validity a practical alternative test of discriminant validity. The HTMT contrasts indicator correlations between constructs with correlations within indicators of the same constructs. It provides a comprehensive, low-constrained approach to assessing discriminant validity for PLS-SEM studies. Franke and Sarstedt (2019) provide further evidence of the robustness of the HTMT as a fully reliable estimator of disattenuated correlations between concepts. The degree to which a construct is distinctive from other constructs is determined by discriminant validity. Thus, the HTMT ratio was used to establish discriminant validity and had to be less than 0.900. Each set of reflective constructs had an HTMT value less than 0.900, as presented in Table 5, indicating that discriminant validity was satisfactorily proven. In conclusion, the concepts and their measures demonstrate good reliability and validity. These results allow us to estimate the structural part of the model. Hypothesis testing shows the results from using the standard error threshold as the criterion for hypothesis validation. The findings indicate that relative advantage (RA), compatibility (CPT), complexity (CPL), organizational readiness (OR), firm size (FS), competitive pressure (CP), and government support (GS) had significant impacts on GAIA. Accordingly, H1, H2, H3, H4, H5 and H6 were supported. Overall, the factors selected for the analysis had an impact on GAIA. The summarized hypothesis results are shown in Table 6. The regression model provides strong explanatory rationality. The R-squared value of 0.622 implies that the independent variables explain 62.2% of the variance associated with GenAI adoption (GAIA). With an adjusted R-squared = 0.609, which slightly underperforms R-squared, the model is still a good fit. Figure 2 presents the results of structural equation modeling. Discussion. Our research analyzes the factors that impact the adoption of GenAI in Vietnam’s banking sector by applying the TOE framework proposed by Baker (2012). The findings reveal a complex interplay of driving and challenging factors unique to Vietnam’s banking context, with significant implications for theory and practice. Driving factors of GenAI adoption Our results highlight seven driving factors determining the adoption of GenAI: relative advantage (RA), compatibility (CPT), complexity (CPL), organizational readiness (OR), firm size (FS), competitive pressure (CP), and government support (GS). These factors are closely tied to the unique context of the banking industry in Vietnam, where both technological and environmental forces play crucial roles in shaping GenAI adoption. Technology context Relative advantage (RA) emerges as a fundamental driver, confirming Al-Khatib’s (2023) findings on the positive relationship between perceived benefits and technology adoption. New technology offers significant advantages to companies in terms of speed and accuracy, making it superior to outdated technology. Vietnamese banks recognize the benefits of adopting GenAI, including improved operational efficiency, reduced costs, Significance levels: p < 0.001, p < 0.05 and enhanced customer engagement. GenAI applications, such as AI-powered chatbots, fraud detection algorithms, and personalized banking services, provide significant value that drives adoption. Successful examples such as Techcombank and VPBank illustrate how leveraging GenAI leads to tailored, data-driven services that meet customer needs. Compatibility (CPT) is a critical factor for banks, as they prefer to implement GenAI when it aligns with their existing processes and technological ecosystems. Nguyen et al. (2022a, b) provide evidence of the positive relationship between technical compatibility and AI adoption. Kruse et al. (2019) noted that the IT architecture can act as a leg-acy ballast, making modifications difficult and hindering digital transformation and AI utilization. This underscores the importance of compatibility in the adoption process. Our study reveals that the seamless integration of GenAI with existing IT infrastructure facilitates adoption. Banks with robust technological systems are more likely to embrace AI. Our findings indicate that banks with advanced IT ecosystems are more successful in adopting GenAI, underscoring the importance of strategic alignment with existing technologies. Organization context Organizational readiness (OR) represents the most significant organizational driver, encompassing technological infrastructure and human capital preparedness. This aligns with Kruse et al.’s (2019) and Al-Khatib’s (2023) findings, emphasizing that banks with strong IT frameworks and human capital are better positioned to integrate AI into their operations. Vietnam’s central state-owned and private commercial banks, including Viet-combank, VietinBank, and BIDV, are poised to embrace GenAI owing to their significant financial and technological resources. These institutions leverage their strong IT frameworks and skilled human capital to integrate GenAI into their operations, enhancing efficiency and innovation. Firm size (FS) significantly impacts GenAI adoption, as larger banks possess the resources for extensive experimentation and implementation. Awa et al. (2016) noted that larger institutions can often invest in training and innovative projects, positioning them advantageously over smaller firms. For example, banks with over 15,000 employees typically can train their staff and invest in extensive GenAI projects, unlike smaller firms with limited human resources. Environment context Competitive pressure (CP) has emerged as a powerful external driver, with heightened competition in Vietnam’s increasingly digital banking landscape compelling institutions to adopt GenAI solutions. Our findings build upon Nguyen et al.’s (2022a, b) suggestion that technological innovation often arises from competitive pressure. Firms may strategi-cally adopt new technology to strengthen their market position. This highlights the need to remain competitive, driving companies to embrace AI technology. The increasing digital shift in the Vietnamese banking sector compels institutions to implement AI solutions to meet the heightened expectations of tech-savvy consumers. Government support (GS) has a significant positive relationship with GenAI adoption. This finding aligns with Nguyen et al. (2022a, b), Almashawreh (2024), and Al-Khatib (2023). Nguyen et al. (2022a, b) noted that government involvement is essential for promoting IT innovation. Al-Khatib (2023) mentioned that government support can establish new rules and strategies for developing AI technologies and facilitating com-mercialization. He highlighted the importance of government backing in encouraging the adoption of technology. In addition, Almashawreh (2024) suggested that government initiatives, such as training programs and incentives, positively affect AI adoption. He demonstrated that government involvement is positively and significantly associated with the adoption of AI applications. Policy measures, such as financial incentives or more transparent regulations, could further promote AI adoption. Recently, the State Bank of Vietnam (SBV) has introduced several initiatives to promote the digital transformation of the banking sector, such as the National Strategy on Industry 4.0 and the development of a legal framework for FinTech and AI-driven financial services, which reflects strong support for GenAI adoption. Challenging factors of GenAI adoption Complexity (CPL) represents the most significant barrier to GenAI adoption. This result is consistent with the findings of Nguyen et al. (2022a, b) and Almashawreh (2024). Nguyen et al. (2022a, b) confirmed a negative relationship between complexity and AI adoption. Almashawreh (2024) explained that the immaturity of AI, coupled with a shortage of experts in technical skills and IT specialists, contributes to the technical complexity of AI. Banks that perceive GenAI as challenging to implement or require substantial time and resources may hesitate to adopt it. The perceived complexity of integrating GenAI systems highlights the need for more precise guidelines and support mechanisms. Simplifying the adoption process and providing comprehensive training can alleviate concerns and foster a more conducive environment for GenAI integration. Our study validates the TOE framework’s application in the context of GenAI adoption in Vietnam’s banking sector, enriching technology adoption theory with context-specific insights. The findings reveal four critical factors that institutions must master: demonstrating tangible GenAI benefits while ensuring system compatibility with existing infrastructure; strengthening organizational readiness through targeted infrastructure investments and developing specialized AI talent pools as adoption pre-requisites; advocating for balanced regulatory frameworks that protect consumers while enabling innovation; and standardizing implementation approaches to over-come complexity barriers through phased adoption strategies. These observations underscore the critical importance of our research at a time when Vietnam’s banking sector stands at a technological inflection point—institutions that effectively navigate these adoption factors and barriers will gain significant competitive advantages beyond mere cost reduction, potentially transforming customer experiences, optimizing risk management, and creating new revenue streams. The findings suggest that early adopters who address these challenges systematically will not only increase their operational efficiency and service quality but also fundamentally reshape the competitive landscape of financial services in Vietnam, potentially accelerating the sector’s digital transformation by 3–5 years compared with its regional counterparts. Figure 3 summarizes the main factors influencing GenAI adoption in the banking industry in Vietnam. Fig. 3 Factors influencing GenAI adoption on the basis of PLS-SEM, R.2 = 0.622 Conclusion, implications, and future research directions Conclusion. The adoption of GenAI in the banking sector has been significantly influenced by the need for digital transformation, increased operational efficiency, and enhanced customer experience. GenAI solutions enable traditional banks to innovate their product offerings effectively and gain a competitive advantage. Our study identifies several key factors driving the adoption of GenAI by Vietnamese banks. These factors include organizational readiness (OR), compatibility (CPT), competitive pressure (CP), complexity (CPL), relative advantage (RA), firm size (FS), and government support (GS). Among these, organizational readiness, characterized by strong IT infrastructure and skilled human capital, emerged as the most influential factor in the adoption of GenAI. Compatibility with existing systems (CPT) and competitive pressure (CP) are also crucial, as they compel banks to adopt GenAI to stay relevant in a rapidly digitalizing market. On the other hand, complexity (CPL) poses a significant challenge, highlighting the need for more straightforward AI solutions and precise implementation guidelines to reduce the risk of project failure. Our research indicates that Vietnamese banks have reached a crucial point in their digital transformation journey, with GenAI significantly influencing the digital banking landscape. Owing to their robust financial and technological infrastructure, large state-owned and private banks, such as Vietcombank, BIDV, and MBBank, leverage GenAI. These institutions highlight the importance of investing in internal capacity and technology readiness for effective GenAI integration. Conversely, medium-sized banks should seek more straightforward GenAI solutions to address complexity and cost challenges. Implications Vietnamese banks should prioritize improving their organizational readiness to facilitate GenAI adoption. This requires increased investment in technological infrastructure and human resource development. Upskilling and retraining employees, particularly those with AI competencies, are the first steps in this process. For example, banks could implement targeted training programs focusing on machine learning fundamentals and prompt engineering. Additionally, establishing dedicated teams to oversee GenAI initiatives is essential for a more organized approach to integrating GenAI. Vietnamese banks must also align their GenAI strategies with long-term business goals to ensure seam-less and scalable AI implementation. In addition, ensuring GenAI’s compatibility with existing technological stacks is essential to ease integration. GenAI solutions should be designed to integrate seamlessly with current IT systems and operational workflows. Furthermore, clear evidence of GenAI’s operational and cost benefits should be provided to increase the perceived relative advantage, making GenAI adoption more compelling to the bank. The following recommendation streamlined the adoption process. GenAI vendors should collaborate closely with banks to develop user-friendly GenAI solutions. They also need to offer interactive training and ongoing technical support. This approach helps alleviate the perception that new technologies are overly complex, which can hinder adoption. Banks can minimize resistance by focusing on pilot projects that are less complex before moving on to more sophisticated applications. For example, internal document processing systems or simple customer FAQ chatbots should be used before more complex credit analysis tools can be implemented. This incremental method allows banks to build confidence and gradually adapt to their internal processes, paving a smoother path for integrating GenAI. Vietnamese banks should strategically leverage competitive pressure to support GenAI initiatives. Private banks must focus on differentiating themselves through inno-vations in GenAI technology to attract a growing, tech-savvy customer base in Vietnam. For example, personalized financial advisory services powered by GenAI can be imple-mented, or real-time translation services for international banking can be offered. Promoting GenAI applications in larger banks first is advisable, as they are generally more capable and prepared for these projects. By successfully implementing GenAI, larger banks can provide valuable insights and best practices from their experiences, which will help smaller banks adopt technology more knowledgeably, affordably, and smoothly. This approach will ultimately benefit the banking industry in Vietnam. Additionally, improved government support for adopting GenAI is essential for achieving broader acceptance. Vietnamese banks should actively engage in discussions with policymakers to ensure that regulations keep pace with technological advancements. The Vietnamese government can increase incentives, such as tax deductions or grants, to promote the research and development of GenAI within the banking sector. For example, implementing regulatory sandboxes for testing GenAI applications or offering tax incentives for banks investing in AI R&D would help accelerate the adoption of GenAI technologies. In the long term, Vietnamese banks that successfully integrate GenAI into their operations will enhance their competitiveness in an increasingly digital economy. These banks can effectively implement GenAI applications by focusing on organizational readiness, simplifying processes, and utilizing government support. This leads to improved operational efficiency and increased customer satisfaction, all while maintaining a significant cost advantage over their existing technology stacks. Practical examples include auto-mated loan processing systems that reduce approval times from days to hours or customer service bots that handle 80% of routine inquiries, allowing staff to focus on more complex customer needs. Future research directions Our study highlighted the factors influencing the adoption of GenAI in Vietnam’s banking sector. However, future research offers ample opportunities to enhance understanding and address some of the limitations identified in this study. Future research should build on current knowledge to address emerging trends. Expanding the research scope by incorporating qualitative questions could lead to a deeper understanding. A qualitative approach can provide insights into organizational culture, leadership attitudes, and employee perceptions—factors often overlooked in quantitative analyses—but can significantly impact adoption success. Additionally, including cross-sector comparisons in the research could lead to more insightful observations. Analyzing other industries, such as insurance and retail finance, may offer a broader perspective on GenAI adoption within Vietnam’s financial services landscape. Furthermore, future research should examine the ethical considerations of adopting GenAI, especially in banking sectors where data security and privacy are critical. Given the sensitive nature of financial data and personal information, this issue is especially relevant to the adoption of GenAI by Vietnamese banks. Additionally, while this study highlights the importance of government support, more in-depth analyses of specific policy interventions, such as tax incentives or grants for GenAI, could provide valuable insights into this crucial aspect. Finally, future research using subgroup analysis among different bank types (state-owned, private, and foreign banks) could enrich the discussion by identifying how ownership structures influence technological adoption decisions. This nuanced understanding would enhance theoretical models and practical implementation strategies across diverse banking ecosystems. Wissem Ajili Ben Youssef Supervision, project administration, and final approval of the manuscript. Najla Bouebdallah Literature review, and critical revision of the manuscript. Long HA Conceptualization of the study, methodology design, and data analysis. Funding No funding. Availability of data and materials Available if requested. Declarations Ethics approval and consent to participate No data was collected that could personally identify participants, ensuring the confidentiality and privacy of all respondents. Consent for publication All co-authors hereby give their explicit consent for the publication. Competing interests No interest conflict. Received: 29 November 2024 Accepted: 5 September 2025