Nonlinear trade-offs in GenAI application pricing: a game-theoretic analysis of API and psychological costs
研究了生成式AI应用提供商在订阅制和按使用付费制之间的定价策略选择，发现API成本和消费者心理成本通过非线性关系影响利润和市场竞争。
In the rapidly evolving generative artificial intelligence (GenAI) ecosystem, downstream application providers face persistent challenges in designing optimal pricing strategies. These challenges arise from dual cost pressures: application programming interface (API) costs imposed by upstream model platforms and psychological frictions experienced by end consumers. This study develops a game-theoretic model to analyze application providers’ strategic choice between subscription-based and usage-based pricing under both monopolistic and duopolistic market structures. The analysis reveals that the optimal strategies are shaped by a nonlinear interplay among psychological costs, model API costs, and consumer valuation. Under low psychological-cost conditions, API costs exert an inverted U-shaped effect on competing providers’ profits. By contrast, under high API-cost regimes, subscription-based and usage-based models exhibit asymmetric sensitivities to psychological costs: profits under usage-based pricing decline with rising psychological costs, whereas subscription-based profits stabilise. Fthermore, elevated API and psychological costs attenuate competitive intensity, driving duopolistic equilibria toward monopoly-like outcomes. These findings remain robust after incorporating a broad set of realistic market frictions. The results provide actionable insights for GenAI application providers balancing cost pass-through, user engagement, and competitive positioning in pricing decisions.