With the lightweighting of large models and the maturity of end-to-end inference frameworks, edge AI is accelerating its implementation in security, retail, and industrial scenarios. Localized inference requires higher storage capacity and bandwidth, and end devices are generally upgrading to storage solutions with larger capacity and lower power consumption. From LPDDR to UFS, storage is becoming a key bottleneck and breakthrough in the end-to-end AI experience. Jieke can provide storage selection recommendations for edge scenarios, helping customers strike a balance between power consumption, cost, and performance.