The involvement of original equipment manufacturers (OEMs) in competitive service markets has brought about new challenges to its maintenance and spare parts inventory management. Under fierce competition from incumbent independent maintenance providers (IMPs) with advantages in cost efficiency and time-varying competitiveness, a new interaction between maintenance service strategy and spare parts provisioning strategy is generated. As a new entrant, an OEM can choose either independent operation or cooperating with an IMP. For each pattern, different from the traditional cost-oriented or efficiency-oriented perspective focusing on fixed service objects, a profit-oriented strategy considering the arrival of time-varying orders under dynamic competition is urgently needed for a service-oriented OEM to achieve success. In this paper, a dynamic maintenance service competition mechanism is established integrating competitors’ learning effects. Accordingly, a new framework of time-varying demand prediction for maintenance actions and spare parts is developed by capturing the dual dynamic features of service order arrival. Then, considering the differences in independent and sharing mechanisms between the two operation patterns, a profit-oriented optimization model for each pattern is proposed. And a Genetic Algorithm and Stochastic Dynamic Programming Combined Algorithm (GA-SDP) is developed to obtain the optimal collaborative maintenance service and spare parts provisioning strategy. Finally, the effectiveness of the proposed models is verified through numerical studies and interesting managerial insights into pattern selection under different conditions for OEMs are provided.