Abstract: As global electricity demand grows, power grid supply chain management faces increasing challenges, particularly in demand fluctuations and supply disruption risks. This study proposes a distributionally robust optimization (DRO)-based inventory management model for power grid supply chains. The model integrates two production plants (local and remote) with multiple distribution stations into a unified supply network. Demand uncertainty is characterized using moment-based ambiguity sets, while supply disruption risks are modeled through expectation constraints. In addition, a goal-oriented service level concept is introduced to balance cost and service quality. A two-stage decision framework is established, where the first stage determines production plans and local transportation strategies, while the second stage adjusts remote transportation strategies and service level shortfalls after demand and production conditions are realized. The model is solved using a column generation algorithm. Numerical experiments demonstrate that compared to deterministic, sample average approximation, and robust optimization models, the proposed DRO model achieves superior performance in total cost and service level stability. A real-world case study further confirms the model adaptability across distribution stations with varying demand characteristics, providing effective decision support for inventory management in power grid supply chains.