This study explores how artificial intelligence (AI) enables organizational ambidexterity to support sustainability-oriented business models. Building on the knowledge-based view of the firm and organizational ambidexterity theory, it conceptualizes AI as a knowledge-enabling infrastructure that transforms how organizations generate, integrate, and apply knowledge. AI is framed as a dual capability that simultaneously supports knowledge exploitation – through optimization, automation, and efficiency – and knowledge exploration – through experimentation, simulation, and the generation of new insights.
Although prior research on AI and sustainability has expanded, it remains fragmented and often focused on linear applications of AI, overlooking its configurational nature. To address this gap, the study adopts a configurational perspective and investigates how different combinations of AI-enabled exploitation and exploration shape sustainability-oriented business models. A qualitative multiple-case study design is used, based on four multinational firms operating in different industries: Enel, Siemens, Unilever, and Nestlé. Data were collected from secondary sources, including sustainability reports, annual reports, and corporate strategy documents, and analyzed through within-case and cross-case comparison.
The findings identify four distinct archetypes of AI-enabled knowledge ambidexterity: (1) an ambidextrous sustainability transformer characterized by high levels of both exploitation and exploration; (2) an efficiency-oriented sustainability optimizer dominated by exploitation; (3) an innovation-oriented sustainability explorer driven mainly by exploration; and (4) a limited AI adoption configuration with weak integration of both processes. Each configuration corresponds to a different sustainability-oriented business model logic, ranging from system-level transformation to incremental efficiency gains and exploratory but less scalable innovation.
The results show that sustainability outcomes are not determined by AI adoption alone, but by how AI-enabled knowledge processes are configured and aligned with organizational structures and business model logic. The study highlights equifinality, demonstrating that different configurations can lead to sustainability through distinct pathways.
Theoretically, the paper contributes by reframing AI-enabled ambidexterity as a configurational knowledge capability and integrating ambidexterity theory with sustainable business model research. Practically, it suggests that managers should move beyond fragmented AI initiatives and instead design integrated systems that balance exploitation and exploration within coherent organizational and strategic architectures.