Smarter, not harder: the AI capability paradox in emerging-market SMEs

Purpose This study aims to identify configurations of artificial intelligence (AI)-related organisational capabilities that lead to superior performance in small and medium-sized enterprises (SMEs) operating in an emerging market, moving beyond the assumption that “more AI usage is better”. Drawing on resource orchestration theory, the authors conceptualise how governance, skills, ethics, leadership and data infrastructure jointly enable value creation from AI. Design/methodology/approach This study relies on a cross-sectional survey of Russian SMEs (October 2024 to January 2025). Of 384 firms, 47 that reported AI use were analysed. Using fuzzy-set qualitative comparative analysis (fsQCA), the authors examined how AI usage intensity combines with internal enablers, AI governance, data infrastructure, employee AI/digital skills, top management team (TMT) involvement and AI ethics preparedness, to explain four outcomes: operational efficiency, strategic decision quality, product/service innovation and customer responsiveness. The authors calibrated the conditions using the direct method and explored the robustness of configurations across alternative consistency and frequency thresholds. Findings Across all outcomes, high AI usage intensity was not a core condition. Instead, multiple high-performance pathways featured AI governance as a central ingredient, frequently complemented by ethics preparedness and either employee training or active TMT involvement. Where governance was weaker, strong employee capabilities could serve as a substitute. These results show that SMEs can achieve strong performance with moderate AI intensity when organisational capabilities are well-aligned. In emerging-market SMEs, this points to an “AI capability paradox”: under the dual constraints of limited resources and weaker institutional environments, more intensive AI use does not necessarily yield better outcomes unless complemented by appropriate capability bundles. Originality/value The authors shift the debate from “how much AI” to “how AI is governed and supported”. By applying a configurational lens in an emerging-market SME context, the authors reveal equifinal capability bundles, highlighting governance and ethics, paired with skills and leadership, as more decisive than sheer adoption intensity. The authors extend AI-related capabilities research to the focus on SMEs in emerging markets. Methodologically, the authors use fsQCA to identify multiple, empirically grounded resource and capability configurations associated with superior performance.