Driving Sustainable Entrepreneurship Through AI and Knowledge Management: Evidence from SMEs in Emerging Economies

Abstract

This study investigates how artificial intelligence (AI) capabilities shape sustainable entrepreneurship (SE) among small and medium-sized enterprises (SMEs) in emerging economies. Focusing on knowledge management (KM) as a mediator, entrepreneurial orientation (EO) as a moderator, and government policy support (GPS) as an enabler, the research draws upon the Knowledge-Based View, Dynamic Capabilities Theory, and Institutional Theory. Using data from Saudi Arabian SMEs operating within the Vision 2030 agenda, the structural model demonstrates that AI primarily influences sustainability when firms possess robust KM systems capable of translating digital insights into actionable practices. Both EO and GPS strengthen the conversion of knowledge into sustainable outcomes, where EO fosters innovation and proactivity, and GPS provides essential resources and legitimacy. Nevertheless, excessive reliance on policy incentives may divert firms toward compliance rather than substantive transformation. Conceptually, this paper situates KM at the core of sustainability transformation, with policy support shaping the institutional context. The findings offer actionable guidance for SME managers and policymakers seeking to advance the United Nations Sustainable Development Goals (SDGs) through strategic engagement with AI and KM.

1. Introduction

Organizations face the dual imperative of achieving economic growth while fulfilling environmental and social responsibilities. Small and medium-sized enterprises (SMEs) are pivotal contributors to economic development, yet they often encounter significant challenges in implementing sustainability initiatives at scale. The 2030 Sustainable Development Agenda underscores the urgency of integrating the United Nations Sustainable Development Goals (SDGs) within national and organizational strategies. In Saudi Arabia, Vision 2030 has placed renewed emphasis on digital innovation, economic diversification, and sustainability, particularly among SMEs [1,2]. Vision 2030 is Saudi Arabia’s comprehensive national development strategy, launched in 2016 by the Saudi government to address post-oil economic sustainability [3]. The initiative establishes an integrated framework encompassing three pillars: (1) a thriving economy that diversifies revenue sources away from petroleum dependence through innovation-driven sectors; (2) an empowered society with enhanced quality of life for citizens and workers; and (3) environmental stewardship and resource conservation. Within Vision 2030, SMEs are strategically positioned as engines of job creation and GDP growth, particularly in the manufacturing, services, and technology sectors. To facilitate SME participation, the Saudi government provides institutional support mechanisms including regulatory frameworks that encourage digital adoption, financial incentives for technology investment, capacity-building programs in AI and data analytics, and sustainability certification schemes aligned with international standards. For SMEs seeking a competitive advantage, Vision 2030 creates both a policy imperative and a practical framework for integrating artificial intelligence and knowledge management systems into core business operations—the focus of this research. Despite these policy priorities, SMEs continue to experience tensions in balancing growth and sustainability, prompting questions regarding both their capabilities and the types of support required [4,5,6].

Adopting AI [7] presents opportunities for improving eco-efficiency and innovation, but also entails risks, such as rebound effects, algorithmic bias, and elevated energy consumption [8]. The effective use of AI for sustainability, therefore, depends on the presence of strong knowledge management (KM) systems capable of converting technological insights into actionable strategies [9,10,11]. This aligns with the Knowledge-Based View (KBV), which foregrounds information as a strategic resource, and the Dynamic Capabilities Theory (DCT), which highlights the capacity for resource reconfiguration in dynamic environments [12,13]. While these frameworks illuminate internal organizational processes, external forces—particularly entrepreneurial orientation (EO) and government policy support (GPS)—also shape the pursuit of sustainable business models [14,15]. EO encompasses innovativeness, proactiveness, and a willingness to experiment, while GPS introduces an institutional dimension through motivational, planning, and regulatory mechanisms [16,17,18].

Despite growing interest, few studies systematically examine how AI, KM, EO, and GPS interact to drive sustainable entrepreneurship. Addressing this gap, the present research proposes and empirically tests an integrated model in which KM mediates the relationship between AI capabilities and SE, with EO and GPS moderating key pathways. By extending KBV and DCT to include institutional considerations, the study demonstrates how digital and organizational resources can be mobilized to advance sustainability under supportive environmental conditions. Drawing on data from Saudi SMEs, this work (1) clarifies that AI advances sustainability primarily through KM; (2) specifies dual roles for EO and underscores the institutional importance of GPS; and (3) delivers practical insights for advancing Vision 2030 and the SDGs.

2. Literature Review

2.1. AI Capabilities

Artificial intelligence (AI) increasingly underpins strategic decision-making in SMEs by automating data analysis, forecasting market trends, and optimizing operations [19,20,21]. Recent research demonstrates that AI can enhance organizational flexibility, responsiveness, and risk management [20,22,23]. However, these benefits are often contingent on the firm’s capacity to interpret and integrate AI-derived insights within core business processes [24]. While AI adoption can drive eco-efficiency and innovation, adverse outcomes such as algorithmic bias, increased energy consumption, and rebound effects must also be considered [25]. Therefore, effective AI utilization requires robust knowledge management systems to translate technological capabilities into sustainable entrepreneurial outcomes.

2.2. Knowledge Management

Knowledge management (KM) is essential for transforming data-driven insights into actionable strategies within SMEs [12,26]. Effective KM systems facilitate the acquisition, codification, sharing, and application of both internal and external knowledge, thereby enabling firms to harness AI for sustainable innovation [27,28]. Empirical evidence indicates that integrating KM with digital capabilities generates continuous improvements in sustainability practices and competitive advantages [29]. Without robust KM processes, however, the benefits of AI technologies and sustainability initiatives are unlikely to be fully realized [19,30,31].

2.3. Sustainable Entrepreneurship

Sustainable entrepreneurship integrates economic performance with environmental and social impact, aligning business objectives with the Sustainable Development Goals (SDGs) and the triple bottom line (TBL) framework [32,33]. SMEs pursuing sustainable entrepreneurship aim to generate value across profit, planet, and people dimensions, but often face tensions and trade-offs when addressing diverse stakeholder goals [34,35,36]. Empirical evidence suggests that while integrating sustainability promotes long-term competitiveness, firms may prioritize economic returns over social or environmental benefits, especially under resource constraints or policy pressures [37,38,39]. Overcoming these challenges requires not only digital transformation and knowledge management but also organizational cultures attuned to innovation and ethical practice.

2.4. Entrepreneurial Orientation

Entrepreneurial orientation (EO) reflects a firm’s propensity for innovativeness, proactiveness, and risk-taking, shaping its strategic responses to environmental challenges [40,41]. Evidence indicates that EO promotes opportunity recognition, experimentation, and resilience, thereby enhancing both competitive performance and the implementation of sustainability initiatives [42,43]. However, EO’s effectiveness depends on its integration with organizational knowledge systems and contextual factors such as policy support [44]. Within this study, EO is posited to play a dual role—serving as both an antecedent of knowledge management and as a moderator amplifying the relationship between KM and sustainable entrepreneurship.

2.5. Government Policy Support

Government policy support (GPS) plays a pivotal role in enabling or constraining sustainable entrepreneurship among SMEs, functioning through regulatory mandates, normative pressures, and cognitive framing, as articulated in Institutional Theory [45]. Effective policies provide strategic incentives, regulatory clarity, and legitimacy, reducing uncertainty and encouraging investment in sustainability-oriented innovation [46,47]. Empirical research demonstrates that well-designed policy support stimulates resource mobilization and organizational commitment to sustainable practices [27]. However, excessive dependence on government incentives may foster a compliance mentality or subsidy dependence, potentially undermining long-term transformation [48]. This study conceptualizes GPS as both a direct institutional enabler of sustainable entrepreneurship and a moderator amplifying the effectiveness of internal capabilities.

2.6. Model Constructs, Theoretical Underpinnings, and Research Hypotheses

The conceptual model integrates Dynamic Capabilities Theory, the Knowledge-Based View, and Institutional Theory to explain how AI capabilities, knowledge management, entrepreneurial orientation, and government policy support interact to drive sustainable entrepreneurship in SMEs [12,45,49].

Table 1. Summary of constructs, theoretical origins, critical debates, the literature gaps, and their link to the research hypotheses.

Construct	Theoretical Base	Major Debate/Contradiction	Gaps in the Literature	Link to Hypotheses
AI	DCT, KBV	Efficiency paradox, energy cost, rebound/justice risk	Pathways from AI to SE still poorly understood	H1, H2, H7
KM	KBV	Org. culture and absorptive capacity, underutilization	Green KM in emerging markets	H3, H4, H5, H6
EO	DCT, KBV	Innovation or risk spiraling?	Moderating effect on KM → SE	H5
GPS	Institutional Theory	Compliance vs. transformation, subsidy dependency	Boundary/spillover effects	H6, H8
SE	All	TBL trade-off, authenticity	Measurement, contextual drivers	H4–H8

summarizes the constructs, their theoretical grounding, key academic debates, and the associated research hypotheses. Figure 1 visualizes this framework, positioning Knowledge Management as the central mediating construct through which AI translates to sustainable entrepreneurship, with Entrepreneurial Orientation and Government Policy Support as moderators (shown as dashed lines).

3. Formulation of Hypotheses

To support the proposed model, hypotheses were developed, starting with the relationship between artificial intelligence abilities and knowledge management. AI helps companies collect, analyze, and interpret large volumes of data. The value of AI for sustainability is only realized when these insights are used effectively in practice. The KBV defines knowledge as the most vital strategic resource for a firm. The DCT explains how AI increases sensing, seizing, and reconfiguring. However, AI must be supported by intentional KM to yield results. Past studies support that AI helps firms acquire, share, and implement knowledge [19,50].

H1. 

AI Capabilities have a positive effect on Knowledge Management.

Extending the discussion to sustainability, the next section takes into consideration the direct effect of artificial intelligence and integrity on sustainable businesses. In addition to the knowledge processes, AI promotes the idea of sustainability directly through increases in efficiency, prediction, and eco-innovation [51,52,53]. In addition, AI can ensure the most efficient use of resources and generate new business sustainability opportunities, particularly in SMEs [54,55].

H2. 

AI Capabilities positively affect Sustainable Entrepreneurship.

Our aim reveals the way in which knowledge management processes are intertwined with the orientation of both enterprises and their entrepreneurial approach. KM processes assist businesses in codifying, sharing, and systematically using sustainability-related knowledge. In turn, this enhances EO, which encourages active discovery, innovation, and risk-taking at a calculated cost [56,57]. SMEs that have a good KM system will be more inclined to pursue entrepreneurial opportunities associated with sustainability [58].

H3. 

Knowledge Management positively influences Entrepreneurial Orientation.

However, knowledge management is crucial to sustainable entrepreneurship, as discussed below. The evidence that KM is a direct way of enhancing SE is increasing. It assists in transforming knowledge into eco-innovations, a circular economy approach, and engagement with stakeholders [59,60]. KM is an engine that transforms AI-driven insights into economic, environmental, and social outcomes.

H4. 

There is a positive impact of Knowledge Management on Sustainable Entrepreneurship.

The connection between knowledge management and sustainable entrepreneurship is also slightly complicated by the moderating influence of entrepreneurial orientation. Even though KM can offer knowledge, the entrepreneurial spirit of the firm will be critical in converting this into SE. EO enhances the presence of a connection between KM and SE due to the encouragement of experimentation, rapid commercialization, and market pioneering [61,62]. Firms that have low EO are unable to embrace sustainability knowledge to its full potential.

H5. 

There is a positive mediating effect of Knowledge Management on Sustainable Entrepreneurship via EO.

Also, the moderating variable of the supporting government policy is evaluated. Government policies provide institutional backing to entrepreneurial initiatives. These include subsidies, carbon reporting, and national sustainability programs. Support of a policy mitigates risks and costs, which makes KM more efficient in generating sustainability outcomes [63,64]. Examples include Vision 2030 policies, which increase the advantages of knowledge-based sustainability in Saudi Arabia.

H6. 

Government Policy Support positively mediates the knowledge management–sustainable entrepreneurship relationship.

The other dimension is the mediating factor of knowledge management, especially between AI and sustainable entrepreneurship. AI may boost sustainability results through direct means, but the best value occurs in cases where knowledge processes exploit its insights. KBV considers knowledge to be the primary intermediary of technology benefits [65,66]. KM, therefore, offers a robust channel: it facilitates a relationship between AI capabilities and SE.

H7. 

Knowledge Management is a mediating variable between AI Capabilities and Sustainable Entrepreneurship.

GPS exerts influence not only as a contextual moderator but also as a direct catalyst for sustainable entrepreneurship in SMEs. Institutional Theory states that the regulative mechanisms, incentives, and legitimacy provided by government bodies can drive sustainability independently of firms’ internal capabilities [45,46]. Recent research in emerging economies demonstrates that strong policy support can lower barriers, promote resource mobilization, and accelerate the uptake of sustainable practices, even among firms at nascent stages of digital or knowledge development [27,48]. The present study therefore proposes that government policy support has a direct, positive impact on sustainable entrepreneurship, beyond its moderating effects on internal capabilities.

H8. 

Government policy support directly enhances sustainable entrepreneurship in SMEs.

4. Research Methodology

4.1. Research Design and Approach

This cross-sectional quantitative study validates the interconnections between AI, KM, EO, and GPS and their contributions to sustainable entrepreneurship. In the cross-sectional design, several constructs can be tested within a single framework. This provides a diagnosis of SME practices. The complex relationships were analyzed with the help of Partial Least Squares Structural Equation Modeling (PLS-SEM). PLS-SEM is used to explain and predict variations, rather than model fit. It applies to emerging areas such as digital technologies in the transformation of sustainability. SEM using covariance requires big samples and normal data; PLS-SEM can be applied to small samples and complicated models. It is suitable for applied research and predictive research, which is the case in sustainability research [67].

4.2. Research Context

The study targets small and medium-sized businesses in Saudi Arabia. The economic transformation of Vision 2030, focusing on digitalization, innovativeness, and sustainability, plays a significant role in the economic transformation of SMEs. Nonetheless, the resources of SMEs are minimal in terms of money, human resources, and technology. This is why they struggle to achieve sustainability and digital innovation. Another issue troubling these companies is the need to minimize their environmental effects and to increase their social contribution while remaining competitive. This twin-pronged aspect renders SMEs an excellent solution to investigate the potential of AI, organizational learning, and policy support in developing sustainable entrepreneurship.

4.3. Sampling and Data Collection

The study targeted small and medium-sized enterprises (SMEs) in Saudi Arabia, following the national definition of firms with between 11 and 500 employees as per the Ministry of Commerce. A stratified random sampling method was utilized to ensure adequate representation from the manufacturing, services, and trade sectors, which are the dominant SME classifications in the country.

The initial sampling frame was developed in partnership with industry associations and business registries, resulting in a list of 1200 eligible firms. Invitation emails outlining the study’s purpose were sent to organizational leaders including CEOs, operational managers, sustainability officers, and IT managers. The survey was conducted electronically from March to August 2024, allowing for flexibility in response and maximizing accessibility.

A total of 523 valid survey responses were received, yielding a response rate of 43.6%. Following data cleaning procedures involving checks for completeness, consistency, and outlier analysis, the final analytic sample comprised 500 SMEs. Respondent demographics were diverse, with firms ranging in age from 1 to 36 years (average 8.3, SD = 6.2), with a median size of 94 employees (range: 11–498). Sector representation included manufacturing (38%), services (42%), and trade (20%), while regional distribution covered Riyadh (40%), Jeddah (30%), Dammam (20%), and other urban centers (10%).

Survey participants held key decision-making roles—35% served as CEO or founder, 30% as operations managers, 20% as sustainability officers, and 15% as IT managers. The majority (85%) possessed at least a bachelor’s degree. Notably, 72% of firms reported some level of AI adoption, and 68% had a formal sustainability program in place.

The sampling protocol, stratification approach, and procedural details were designed to enhance the generalizability and credibility of study findings, while also ensuring coverage of digitally active and sustainability-engaged organizations most relevant to the research objectives.

4.4. Measurement of Constructs

The survey used in the study is based on tried and tested questions from past research, updated to match the situation of small businesses focusing on sustainability in Saudi Arabia. All questions asked people to rate their agreement on a five-point scale from 1 (strongly disagree) to 5 (strongly agree).

The measurement scales used in the study were well established and were the same that had been used in previous studies, ensuring their reliability and fit with the context. AI capabilities were measured using items that reflected decision support, recognition of opportunity, minimization of resources, operational efficiency, and integration in sustainability strategies. KM talks about five dimensions, and these dimensions include acquisition, sharing, storing, application, and creating knowledge pertaining to sustainability. The EO addressed the level of innovativeness, proactiveness, and risk taking and GPS addressed aspects such as regulatory encouragement, policy incentives, country programs, and institutional support. Lastly, SE was evaluated based on triple bottom line measures: economic, environmental, and social performance. A pilot study was conducted on a sample of SME managers to make sure that the questionnaire was clear and relevant, and some slight modifications were made to correct it. The questionnaire is provided in detail in Annexure I.

4.5. Data Analysis Procedure

Data analysis was performed using SmartPLS 4. To verify that the measurement model was reliable and valid, we initially used tests such as Cronbach’s alpha, AVE, HTMT, and VIF to determine that the items used were consistent and different. Then, we evaluated the structural model, estimating the path relationships and their significance using bootstrapping, and looked at R², f², and Q² values in order to evaluate their explanatory power and predictive relevance. We then mediated, moderated, and considered whether knowledge management mediated or moderated relationships between AI and sustainability outcomes, and whether entrepreneurial orientation and government policy reinforced these associations. Other indices, like SRMR and NFI, were also used to confirm model fit. Lastly, sophisticated methods were used: IPMA to reveal the most problematic and underperforming aspects of sustainable entrepreneurship, and PLS-POS to reveal the hidden subgroups of the SME sample to learn more about them.

4.6. Ethical Considerations

Strict ethical standards were upheld in this research. Respondents were allowed to join and participate voluntarily, without revealing the academic purpose of the study. Participation was informed, and anonymity was guaranteed to ensure the confidentiality of the organizational information. There was no collection of any personally identifiable data. Institutional guidelines were adhered to by obtaining ethical approval at the College of Business at Jazan University.

4.7. Methodological Rigor

Potential common method bias was addressed through (a) the use of Harman’s single-factor test, which revealed that no single factor explained more than 31% of the variance, and (b) temporal separation within the survey design, where independent and dependent variables were measured in separate sections. Reverse-coded items and assurances of anonymity further minimized bias. Non-response bias was assessed by comparing early and late respondents in key variables, finding no significant differences.

Given the cross-sectional survey design, causal inferences are limited. This study addresses potential endogeneity and reverse causality, particularly in the KM–SE relationship, by grounding hypotheses in established theory (Dynamic Capabilities Theory, Knowledge-Based View, Institutional Theory) and using statistical controls such as mean-centering for moderation effects and robust variance inflation factor (VIF) diagnostics (all VIF < 5). Nonetheless, the potential for omitted variable bias and simultaneity remains; this is acknowledged as a limitation, and longitudinal follow-up is recommended.

4.8. Reporting Standards

This study adheres to the leading standards for academic transparency and reporting. Full details on model specification, estimation procedures, indicator loadings, and validity statistics are provided in Appendix A and the Supplementary Materials.

5. Data Analysis and Interpretation

5.1. Assessment of Model

Figure 2 displays the complete PLS-SEM model, along with both measurement and structural components. The left side shows how each construct is operationalized through indicators; the right side illustrates the pathways between constructs with standardized coefficients and t-statistics. Solid lines represent direct effects and mediation relationships, while dashed lines show the moderating influences of EO and GPS on the KM → SE link. AI Capabilities and Knowledge Management emerge as the primary drivers of sustainable entrepreneurship, with KM serving as the transformation mechanism converting digital insights into practice. The moderating pathways reveal that entrepreneurial orientation and government policy support amplify knowledge management’s effectiveness in generating sustainability outcomes.

5.1.1. Measurement Model Assessment

The first step in PLS-SEM analysis involves evaluating the reliability and validity of the measurement model to ensure that constructs are appropriately operationalized through their respective items. This assessment confirms that each measured indicator accurately captures its intended construct before proceeding to structural model testing.

Table 2. Convergent and discriminant validity.

Constructs	Cronbach’s α	rho_A	Composite Reliability (ρc)	AVE	AI	EO	GPS	KM	SE	EO × KM	GPS × KM
AI Capabilities (AI)	0.905	0.907	0.930	0.726	–	0.348	0.267	0.684	0.659	0.088	0.020
Entrepreneurial Orientation (EO)	0.910	0.914	0.933	0.736	0.348	–	0.047	0.245	0.255	0.008	0.034
Government Policy Support (GPS)	0.910	0.935	0.932	0.734	0.267	0.047	–	0.141	0.129	0.029	0.062
Knowledge Management (KM)	0.894	0.895	0.922	0.702	0.684	0.245	0.141	–	0.692	0.060	0.042
Sustainable Entrepreneurship (SE)	0.909	0.909	0.932	0.733	0.659	0.255	0.129	0.692	–	0.282	0.210
EO × KM	–	–	–	–	0.088	0.008	0.029	0.060	0.282	–	0.081
GPS × KM	–	–	–	–	0.020	0.034	0.062	0.042	0.210	0.081	–

presents the key reliability and validity statistics for all six constructs, including Cronbach’s alpha, composite reliability, average variance extracted (AVE), and heterotrait–monotrait (HTMT) ratios for discriminant validity assessment.

The measurement model results decisively validate the reliability and validity of each construct. All Cronbach’s alpha coefficients exceeded 0.89 and composite reliability values exceeded 0.92, substantially exceeding the 0.70 threshold, indicating strong internal consistency across all constructs. Average variance extracted (AVE) ranged from 0.70 to 0.73, all exceeding the 0.50 threshold for convergent validity, confirming that measurement items adequately represent their constructs. Discriminant validity was confirmed through Fornell–Larcker criterion (square root of AVE exceeds inter-construct correlations) and HTMT ratios (all < 0.85, well below the 0.85 conservative threshold). These results provide strong confidence in measurement model quality and allow for the study to proceed to structural model testing. Notably, interaction terms (EO × KM and GPS × KM) showed acceptable correlations with no multicollinearity concerns, supporting their use as moderators in subsequent analysis.

5.1.2. Overall Model Fit and Explanatory Power

After confirming measurement model adequacy, a comprehensive assessment of overall model fit and predictive power is conducted. This evaluation examines whether the proposed structural model appropriately represents the data and whether the model explains meaningful variance in endogenous constructs.

Table 3. Comprehensive model assessment results.

Category	Indicator	Value(s)	Interpretation
Post hoc Power Analysis	Required Effect Size (α = 1%, 80% Power)	0.107	Minimum detectable effect; indicates sufficient sensitivity for small effects
	Required Effect Size (α = 5%, 80% Power)	0.084	Demonstrates strong statistical power under standard significance level
	Required Effect Size (α = 1%, 90% Power)	0.121	Indicates ability to detect medium effects with high confidence
	Required Effect Size (α = 5%, 90% Power)	0.098	Confirms robust statistical adequacy for hypothesis testing
Model Fit Indices	SRMR	Saturated = 0.033; Estimated = 0.050	Values < 0.08 demonstrate good model fit
	d_ULS	0.356 (Sat.); 0.824 (Est.)	Lower values indicate better approximation of empirical data
	d_G	0.158 (Sat.); 0.165 (Est.)	Acceptable consistency between empirical and model-implied matrices
	Chi-square	820.692 (Sat.); 844.915 (Est.)	Indicates acceptable model–data discrepancy
	NFI	0.944 (Sat.); 0.943 (Est.)	Values > 0.90 indicate strong model fit
Explained Variance (R2)	Entrepreneurial Orientation	0.050 (Adj. 0.049)	Weak explanatory power—suggests influence from additional external factors
	Knowledge Management	0.380 (Adj. 0.380)	Moderate explanatory power—indicates meaningful variance explained
	Sustainable Entrepreneurship	0.540 (Adj. 0.536)	Substantial explanatory power—confirms strong predictive relevance
Multicollinearity (VIF)	AI Indicators	2.228–2.484	All < 5, indicating no multicollinearity concerns
	EO Indicators	2.352–2.661	Acceptable range, supporting construct independence
	GP Indicators	2.340–2.768	Below threshold; confirms model stability
	KM Indicators	2.049–2.323	Reflects acceptable variance inflation
	SE Indicators	2.385–2.559	Within recommended limits
	Interaction Terms	EO × KM = 1.000; GPS × KM = 1.000	Perfect centering eliminates collinearity risk
Predictive Relevance (Q2 predict)	Entrepreneurial Orientation	Q2 = 0.067; RMSE = 0.968; MAE = 0.807	Indicates weak predictive accuracy, suggesting exploratory nature
	Knowledge Management	Q2 = 0.378; RMSE = 0.790; MAE = 0.636	Demonstrates strong predictive relevance
	Sustainable Entrepreneurship	Q2 = 0.368; RMSE = 0.797; MAE = 0.649	Indicates strong predictive relevance for sustainability outcomes

presents multiple fit indices, power analysis results, multicollinearity diagnostics, and predictive relevance (Q²) values across all constructs.

The model demonstrates strong fit and adequate explanatory power. Standardized root mean square residual (SRMR = 0.050) and normed fit index (NFI = 0.943) both indicate good overall model fit, with values well within acceptable thresholds. Post hoc power analysis confirms that the study design can detect even small effect sizes (α = 5%, 80% power = 0.084 minimum effect size), ensuring robust hypothesis testing. Variance inflation factors across all indicators and interaction terms remain below 5.0, confirming the absence of multicollinearity problems. Most importantly, the model explains 54% of the variance in Sustainable Entrepreneurship (R² = 0.54), a substantial effect size indicating that the proposed relationships meaningfully predict sustainability outcomes. Knowledge Management shows 38% explained variance, while Entrepreneurial Orientation shows 5% variance, suggesting that EO is influenced by external factors beyond KM alone. Predictive relevance (Q² = 0.368 for SE; 0.378 for KM) demonstrates strong out-of-sample predictive power, confirming the model’s generalizability beyond the current sample. These comprehensive fit indices collectively support the structural model as an appropriate representation of the dynamics between AI, knowledge management, and sustainable entrepreneurship.

5.1.3. Hypothesis Testing and Path Relationships

With measurement model validity and overall model fit confirmed, the hypothesized path relationships are now examined.

Table 4. Hypothesis testing.

Hypothesis	Path Relationship	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	t-Statistics	p-Value	Effect Size (f2)	Decision
H1	AI Capabilities → Knowledge Management	0.617	0.617	0.022	28.424	0.000	0.614 (Large)	Supported
H2	AI Capabilities → Sustainable Entrepreneurship	0.327	0.326	0.030	10.727	0.000	0.129 (Medium)	Supported
H3	Knowledge Management → Entrepreneurial Orientation	0.223	0.225	0.032	6.947	0.000	0.053 (Small–Medium)	Supported
H4	Knowledge Management → Sustainable Entrepreneurship	0.399	0.400	0.030	13.485	0.000	0.214 (Medium–Large)	Supported
H5 (Moderation)	Entrepreneurial Orientation × Knowledge Management → Sustainable Entrepreneurship	0.208	0.208	0.022	9.259	0.000	0.092 (Small–Medium)	Supported
H6 (Moderation)	Government Policy Support × Knowledge Management → Sustainable Entrepreneurship	0.160	0.160	0.021	7.555	0.000	0.056 (Small–Medium)	Supported
H7	Entrepreneurial Orientation → Sustainable Entrepreneurship	0.039	0.040	0.025	1.548	0.122	0.003 (Negligible)	Not Supported
H8	Government Policy Support → Sustainable Entrepreneurship	0.285	0.282	0.027	10.556	0.000	0.210 (Medium–Large)	Supported

presents the standardized path coefficients (β), t-statistics, p-values, effect sizes (f²), and hypothesis decisions for all eight hypotheses. Significance is assessed using bootstrapping with 5000 resamples to ensure robust statistical inference. Effect sizes are interpreted as small (f² = 0.02), medium (f² = 0.15), or large (f² = 0.35), following Cohen’s guidelines for PLS-SEM.

5.1.4. Summary of Hypothesis Testing Results

Hypotheses testing revealed strong support for seven of the eight proposed relationships, with notable variations in effect magnitudes that illuminate the relative importance of different pathways to sustainable entrepreneurship. H1 demonstrated the strongest relationship: AI Capabilities → Knowledge Management (β = 0.617, t = 28.424, p < 0.001, f² = 0.614, large effect). This result confirms that AI is a primary driver of knowledge management capability, with firms leveraging AI technologies to sense, acquire, and systematize knowledge about sustainability opportunities. H2 (AI → SE direct effect) was supported (β = 0.327, p < 0.001, medium effect), indicating that AI also has direct positive effects on sustainable outcomes through efficiency improvements and innovation. However, the medium effect size relative to H1 suggests that AI’s sustainability impact is substantially mediated through knowledge management systems rather than being a purely technological effect. H3 (KM → EO, β = 0.223, p < 0.001, small–medium effect) and H4 (KM → SE, β = 0.399, p < 0.001, medium–large effect) both supported the central role of knowledge management. Notably, KM’s direct effect on sustainable entrepreneurship (H4) is substantially stronger than its effect on entrepreneurial orientation (H3), confirming that KM serves as the primary transformation mechanism converting technological and organizational inputs into sustainability outcomes. These findings are also in line with those of the Journal of Cleaner Production, which indicates that green KM facilitates eco-innovation and competitive advantages [68]. The moderating effects proved significant: H5 (EO × KM → SE, β = 0.208, p < 0.001, small–medium effect) and H6 (GPS × KM → SE, β = 0.160, p < 0.001, small–medium effect) confirm that both entrepreneurial orientation and government policy support amplify KM’s effectiveness. The larger moderation coefficient for EO versus GPS suggests that firm-level organizational culture may be slightly more influential than external policy in determining whether knowledge management translates to sustainability outcomes. Notably, H7 (EO → SE direct effect) was NOT supported (β = 0.039, t = 1.548, p = 0.122, negligible effect). This non-significant direct effect indicates that entrepreneurial orientation does not automatically guarantee sustainable outcomes; rather, EO requires integration with robust knowledge management systems to translate entrepreneurial energy into substantive sustainability practices. This finding protects against greenwashing by high-EO firms that pursue aggressive strategies without a knowledge-based assessment of true sustainability impacts. H8 (GPS → SE direct effect) was strongly supported (β = 0.285, p < 0.001, medium–large effect), confirming that government policy support provides both direct institutional catalysts for sustainability adoption and moderating effects that amplify internal capabilities. In combination, these results demonstrate that sustainable entrepreneurship emerges from the integrated effect of technological capacity (AI), organizational knowledge systems (KM), internal organizational culture (EO), and institutional support (GPS), with KM serving as the critical transformation hub.

IPMA was used in the study to show the factors that have the greatest impact on sustainable entrepreneurship. It demonstrates the importance of each of them in relation to their current performance. IPMA assists in determining the constructs (AI, KM, EO, or GPS) that are most important to the sustainability results and the strengths and weaknesses of the company. This allows managers and researchers to determine which enhancement to focus on, depending on the extent to which it will assist the environment. According to the results in

Table 5. Importance Performance Map Analysis (IPMA).

Construct	Item Code	Questionnaire Statement	Importance (IPMA)	Performance (Revised)
AI Capabilities (AI)	AI1	AI supports decision-making in sustainability projects.	0.139	68
	AI2	AI identifies new sustainability opportunities.	0.135	70
	AI3	AI optimizes resource use and environmental impact.	0.141	65
	AI4	AI improves operational sustainability efficiency.	0.125	67
	AI5	AI is integrated into long-term sustainability strategies.	0.138	66
Entrepreneurial Orientation (EO)	EO1	We seek innovative sustainability solutions.	0.008	60
	EO2	We exploit sustainability-driven opportunities.	0.009	62
	EO3	We take calculated risks in sustainability ventures.	0.009	59
	EO4	We lead in sustainability-oriented offerings.	0.010	61
	EO5	Sustainability innovation is part of our strategy.	0.010	63
Government Policy Support (GP)	GP1	Regulations encourage sustainability entrepreneurship.	0.007	57
	GP2	Policy incentives support our initiatives.	0.004	55
	GP3	National programs (Vision 2030, SDGs) influence practices.	0.007	58
	GP4	Policies make sustainability implementation easier.	0.006	56
	GP5	We receive adequate institutional support.	0.005	54
Knowledge Management (KM)	KM1	We acquire sustainability-related knowledge.	0.101	64
	KM2	Knowledge sharing supports sustainability.	0.095	63
	KM3	We maintain sustainability knowledge repositories.	0.092	61
	KM4	Knowledge application informs sustainability decisions.	0.098	65
	KM5	We update our sustainability knowledge continuously.	0.102	66

of IPMA ranking, the most significant elements of sustainable entrepreneurship are the AI Capabilities and KM. In the AI, AI3 (to use AI to make sure resources are used efficiently and cause less harm to the environment; MV = 0.141) and AI1 (to use AI to assist in making decisions related to sustainability projects; MV = 0.139) make the biggest contribution. This implies that companies can enhance sustainability through the incorporation of AI to enhance environmental efficiency and decision-making. AI5 (MV = 0.138) stresses that the usage of AI in long-term sustainability plans is important, while its contribution to short-term operational activities is still less significant. It is implied that it is strategic rather than operational uses of AI that are the drivers of sustainability. Knowledge management comes up as one of the key enablers. KM5 (process of continuous creation and updating of knowledge to enhance sustainability; MV = 0.102) emphasizes the fact that sustained learning is a more important determinant of sustainability performance as compared to the storage of knowledge in place. The role of KM1 and KM4 is also important, which means that the process of learning and implementing sustainability is critical to change resources into entrepreneurial behavior. This highlights the central role of KM: it directs AI knowledge into real, sustainable operations, serving as a driver of organizational change. The implication is that organizations should maintain proactive KM processes to increase sustainability. There is no significant value addition (MVs 0.008–0.010) of EO items. This implies that EO can reinforce the effect of KM by innovating and taking responsibility, but this is not the sole way of promoting sustainability. Items under GPS are moderate in importance (MVs 0.004–0.007): they are not as effective as AI and KM, yet institutional frameworks (e.g., regulations and national programs) provide a facilitating environment for sustainable practices. The implication is that although government support does not automatically assure a sustainable outcome, there are policy implications and institutional credibility that can orient films towards sustainable entrepreneurship, particularly when policies are consistent with long-term national goals.

5.2. Mechanism and Analytical Rigor

This study advances analytical rigor by clarifying the organizational pathways through which AI capabilities drive sustainable entrepreneurship. Specifically, the results show that artificial intelligence enhances knowledge management by enabling efficient data interpretation, proactive information sharing, and knowledge codification. In turn, robust knowledge management systems facilitate the translation of digital insights into sustainable practices and innovative business models. The effect is further shaped by entrepreneurial orientation and government policy support, which strengthen the link between knowledge management and entrepreneurial outcomes, respectively.

6. Discussion

To illustrate practical significance, case studies from Saudi Arabia’s digital transformation initiatives, including the Knowledge-based Advanced Manufacturing Innovation Network (KAMIN), demonstrate improved operational efficiency [69]. The KAMIN program trained 800 participants from 120 SMEs and manufactured 1400 prototypes, showcasing how structured knowledge transfer enhances productivity. Quantitative research on 605 Saudi service sector organizations confirms that knowledge management processes significantly improve operational performance [70]. Conversely, firms reliant solely on policy incentives without internal capability development primarily achieved compliance, not transformation. These cases highlight that successful transitions require both robust internal systems and external support, affirming the quantitative results.

The study results allow us to empirically justify the idea that AI skills and knowledge management can be used to create sustainable entrepreneurship, characterized by the effects of entrepreneurial orientation and government policy advocacy. The variance that is explained by the fact that the SE model is high (54) [71,72,73]. This framework covers institutional and organizational aspects that will help SMEs to be sustainable within emerging economies, herein the Saudi Arabia Vision 2030 setting. The research adds to the existing body of literature by illuminating the correlation between AI capabilities, KM, EO, and GPS in developing SE in SMEs. According to the DCT, KBV, and Institutional Theory, digital technologies can only help to achieve sustainability when there are good internal knowledge practices and positive institutional conditions. The implication is that AI is a facilitator, KM is the process of turning knowledge into action, and EO and GPS play an important role in the environment as far as effective sustainability results are concerned. For details, refer to

Table A1. Total Effect.

Constructs	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	p Values
AI Capabilities → Entrepreneurial Orientation	0.138	0.139	0.022	6.352	0.000
AI Capabilities → Knowledge Management	0.617	0.617	0.022	28.424	0.000
AI Capabilities → Sustainable Entrepreneurship	0.578	0.578	0.022	26.391	0.000
Entrepreneurial Orientation → Sustainable Entrepreneurship	0.039	0.040	0.025	1.548	0.122
Entrepreneurial Orientation x Knowledge Management → Sustainable Entrepreneurship	0.208	0.208	0.022	9.259	0.000
Government Policy Support → Sustainable Entrepreneurship	−0.025	−0.023	0.024	1.082	0.279
Government Policy Support x Knowledge Management → Sustainable Entrepreneurship	0.160	0.160	0.021	7.555	0.000
Knowledge Management → Entrepreneurial Orientation	0.223	0.225	0.032	6.947	0.000
Knowledge Management → Sustainable Entrepreneurship	0.408	0.409	0.030	13.541	0.000

and

Table A2. Specific Indirect Effect.

Constructs	Original Sample (O)	Sample Mean (M)	Standard Deviation (STDEV)	T Statistics (|O/STDEV|)	p Values
AI Capabilities → Knowledge Management → Sustainable Entrepreneurship	0.246	0.247	0.020	12.196	0.000
Knowledge Management → Entrepreneurial Orientation → Sustainable Entrepreneurship	0.009	0.009	0.006	1.500	0.134
AI Capabilities → Knowledge Management → Entrepreneurial Orientation → Sustainable Entrepreneurship	0.005	0.005	0.004	1.490	0.136
AI Capabilities → Knowledge Management → Entrepreneurial Orientation	0.138	0.139	0.022	6.352	0.000

.

6.1. AI Is an Enabling Force, Rather than a Panacea to Sustainability

The idea is that AI serves as an enabler, rather than the only driver of sustainability. Although AI helps in predictive analytics, resource optimization, and eco-efficiency, when used with KM processes to codify and act upon digital insights, a major impact is generated. This reinforces the significance of the integration of AI with KM to achieve significant organizational change, which is in line with recent findings that show AI has been utilized to improve decision-making processes and not to automate sustainability, which aligns with previous research [74,75]. In practice, it is advisable to pay attention to the integration of AI into KM practices that aim to achieve the greatest level of sustainability.

This literature is based on the previous literature, which centers on the benefits of digitalization in terms of efficiency [74,76]. Without a well-established KM, SMEs adopting AI can become the victims of pilot project traps: digital tools create insights that do not become sustainable practice. Such research demonstrates the highest level of influence of AI in the case of integration with organizational knowledge practices, which contribute to sustainability in general.

6.2. The Conversion Engine Is Knowledge Management

The second implication is that KM is the core of the organization in terms of its transforming digital insights and entrepreneurial intent into quantifiable sustainability results, which supports the results of previous studies [12]. The Knowledge-Based View suggests that knowledge is the most strategic asset of the firm. These results, similarly to those obtained by other literature sources, support the idea that successful green KM, i.e., the acquisition, sharing, and utilization of knowledge, is the key to eco-innovation, embracing a circular economy, and gaining a competitive advantage, as mentioned in previous studies [77]. This implies that organizations should work on systematized KM to transform digital opportunities into sustainability initiatives [78].

Significantly, this study has shown that KM is the best predictor of SE, which proves the necessity of systematic knowledge practices in sustainability transitions, as argued in a previous study [79], in addition to entrepreneurial motivation or technology acceptance. This supports the fact that the ability to absorb and integrate green knowledge is the determinant of environmental competitiveness [80]. Practically, SMEs investing in strong KM systems are in a better position to integrate the insights derived from AI into practical interventions, which include eco-design, efficient production, and inclusive models, which also strengthens the results of previous studies [81]. Moreover, the creation of powerful KM systems is key to the viable sustainability benefits, aligning with previous research [31].

This contribution reinforces the opinion that KM [82] is to be regarded more as a strategic means of maintaining performance, as opposed to playing a supporting role. KM is the connection between digital transformation and sustainability outcomes, and this is why some SMEs successfully implement sustainability and others only achieve surface-level implementation. The implication is that strategic KM fills the gap in the meaningful integration of sustainability.

6.3. EO and GPS: Situational Boundary Conditions

The third implication is that context is important. EO and GPS affect the way knowledge is converted to the sustainability outcomes, but differently. EO reinforces the interaction between KM and SE, which proves the idea that proactive, innovative, and risk-taking firms are more prone to take on sustainability knowledge, which aligns with the previous study [83]. However, EO cannot ensure sustainability, and one should not rely on entrepreneurial spirit only, which argues against previous studies [84]. This implies that although EO may facilitate action, it needs to be complemented by strong KM and policy frameworks to prevent superficial or insincere implementation.

Another key policy that is instrumental in determining the outcomes of sustainability is the GPS, which offers external legitimacy and minimizes uncertainty, supporting previous studies [85]. Although policy incentives have the potential to scale eco-innovation, where the goals of SMEs are aligned with national goals, excessive use of government incentives may result in compliance-based sustainability practices as opposed to transformative practices. Hence, policy support must be employed to strengthen, and not to replace, internal organizational initiatives towards sustainable entrepreneurship. These findings form part of the overall discourse on contextual conditions, in the sense that they show that EO provides the internal mindset and GPS provides the external scaffolding; they are both required to ensure that KM will translate into real sustainability outcomes.

The findings underscore that AI and knowledge management are most effective when supported by entrepreneurial leadership and favorable policy environments. SME managers should focus on building organizational processes that encourage experimentation, cross-functional learning, and the systematic integration of digital insights into sustainability strategies. Policymakers can maximize impact by pairing incentives with capability-building programs, ensuring support extends beyond compliance to genuine transformation. These actionable insights inform both managerial practice and policy design for sustainable digital transformation, as shown in

Table 6. Summary of key contributions, findings, and future research.

Key Contribution	Summary of Findings/Novelty	Directions for Future Research
Integration of Digital, Organizational, and Institutional Factors	Demonstrates that AI capabilities, when combined with knowledge management (KM), entrepreneurial orientation (EO), and government policy support (GPS), most effectively advance sustainable entrepreneurship (SE) in emerging market SMEs.	Explore multi-level or cross-sector models in other emerging economies; test for differences in combinations of internal/external drivers.
Mediation Role of Knowledge Management	Establishes KM as a critical mediator that converts digital (AI) potential into actionable sustainability practices.	Conduct qualitative studies on KM processes and barriers in SMEs; investigate specific KM practices most linked to sustainability.
Moderating Impact of EO and GPS	Reveals that both EO and GPS magnify the KM → SE link, highlighting cultural and institutional contexts as amplifiers.	Longitudinal studies on how EO and GPS evolve; examine EO/GPS in various institutional settings or policy regimes.
Direct Effect of Government Policy Support (H8)	Confirms that government policy directly fosters SE, not merely as a moderator, validating the Institutional Theory’s claims in emerging contexts.	Assess long-term effects of policy dependency; compare effectiveness of different policy instruments or approaches.
Qualitative Managerial Insights	Shows from open-text responses that genuine SE outcomes require both internal champions and enabling environments—policy alone can foster compliance without transformation.	Case studies to understand “what works” on the ground; behavioral research on attitudes toward compliance versus innovation.
Saudi Context and Vision 2030 Blueprint	Offers empirical evidence for Vision 2030’s effectiveness in catalyzing SME sustainable transformation, with lessons for similar emerging markets.	Comparative policy analyses; adapt the model for various national development agendas and stages of digital maturity.

.

These findings show the theoretical contributions of and practical implications for SMEs, policy implications, compatibility with the SDGs, relevance to society, limitations, and future research opportunities. This overall summary describes how research can contribute to the current theories and provides useful advice to managers and policymakers. It also explains the relationship with international sustainability objectives, including the SDGs and Vision. These findings are visible to the academics and practitioners according to the table that contains them.

While these findings are situated within the ambitious policy context of Saudi Vision 2030, the pathways identified here may have broader relevance for emerging economies pursuing sustainable entrepreneurship via digital transformation. Nevertheless, outcomes should be interpreted considering contextual differences such as regulatory environments and levels of digital readiness. The country’s state-driven model with substantial policy scaffolding may not mirror environments with less government intervention or fewer resources. Adaptation and context-specific analysis are therefore recommended before broadly applying these findings.

Comparative analysis across emerging or developed economies would be valuable for testing the robustness of these mechanisms under diverse institutional settings. Particularly, exploring countries with varying levels of digital readiness, policy support, and market maturity could illuminate the conditions under which these pathways to sustainable entrepreneurship are most effective, and highlight whether policy-driven models offer unique advantages or constraints.

Ultimately, recommendations for practice and policy must be applied with careful regard for local circumstances. Strategies that succeed in the Saudi context, characterized by strong government support and ambitious economic reforms, may require adjustment in other environments. Researchers and practitioners should continually assess how institutional, cultural, and economic conditions shape the effectiveness of digital and sustainability interventions.

7. Conclusions

This study was conducted in the context of Saudi Arabia’s Vision 2030—a far-reaching national initiative to foster economic diversification, promote digital transformation, and advance sustainability goals. The policy environment is distinguished by active government involvement, large-scale incentives, and a strong push toward institutional modernization, making it an ideal setting for examining how digital capabilities interact with entrepreneurial practices and policy support in SMEs.

This research advances theory in three ways. First, it extends the Knowledge-Based View by demonstrating that AI capabilities drive sustainability primarily through knowledge management mechanisms rather than through technological efficiency alone. Second, it integrates Dynamic Capabilities Theory with Institutional Theory, showing that both internal capabilities (sensing through AI, transforming through KM) and external institutional support (GPS) shape sustainability outcomes. Third, the non-significant direct effect of EO provides important boundary conditions, suggesting that entrepreneurial orientation operates through knowledge-mediated pathways rather than directly influencing sustainability.

For SME managers, the findings suggest prioritizing KM infrastructure alongside AI adoption, cultivating entrepreneurial culture to amplify knowledge effectiveness, and measuring authentic sustainability outcomes. For policymakers, the results indicate that technology subsidies without KM capacity-building yield limited transformation; effective policies should integrate training, knowledge infrastructure, and incentive structures rewarding genuine sustainability rather than adoption metrics. For Vision 2030 implementation, the study validates current programs combining technology and knowledge development (e.g., KAMIN) while cautioning against approaches emphasizing entrepreneurial spirit without corresponding knowledge systems.

7.1. Limitations and Directions for Future Research

Several limitations warrant acknowledgment. The cross-sectional design precludes causal inference; longitudinal research is needed to establish temporal precedence. Self-reported data may introduce common method bias, although statistical tests (Harman’s single-factor) suggest acceptable levels. The Saudi Arabian context, while relevant for Vision 2030 research, may limit generalizability to other emerging economies or developed markets. Finally, the quantitative approach, while enabling hypothesis testing, cannot capture the nuanced implementation processes that qualitative methods might reveal.

Future research should address these limitations through (1) longitudinal panel studies tracking AI-KM-SE relationships over time to establish causality; (2) multi-country comparisons across GCC and other emerging economies to test generalizability; (3) sector-specific analyses examining whether manufacturing, services, and trade SMEs exhibit different pathway strengths; (4) qualitative case studies providing rich implementation narratives; (5) examination of negative outcomes (greenwashing, compliance-driven adoption) in firms lacking KM foundations; and (6) investigation of additional moderators, including firm size, industry turbulence, and digital maturity levels.

7.2. Final Remark

This study does not contradict the United Nations Sustainable Development Goals because it explains how SMEs can use AI and knowledge management to create economic, environmental, and social value. The lessons provide practical advice to policymakers, practitioners, and businessmen to support resilient, inclusive, and responsible business models, thus improving the welfare of society as a whole and sustainable development.