Silicon Motion Technology (SMI) has released a portfolio of AI-optimized storage controllers and products ahead of Ccmputex 2026, targeting edge inference, AI PCs, enterprise AI infrastructure and automotive AI systems. The company said it will demonstrate new Edge SSD controllers, embedded UFS and eMMC controllers, enterprise NVMe solutions and automotive-grade storage aimed at improving data movement, latency and sustained workload performance for AI deployments.
For AI PCs and edge AI devices, SMI announced the SM2524XT, a PCIe Gen5 DRAMless SSD controller it described as optimized for inference and key-value cache workloads. The firm said the Edge SSD controller lineup is designed to accelerate model response speed and reduce latency in situations where local inference and fast cache access are critical.
In embedded storage, the company presented new UFS and eMMC controller designs intended for always-on AI applications and next-generation smart edge devices. SMI stated these embedded controllers offer higher bandwidth, lower latency and improved power efficiency, positioning them for use in mobile devices, edge gateways and other constrained platforms that require persistent, low-power AI processing.
Furthermore, the company outlined an enterprise SSD controller portfolio and a next-generation AI infrastructure platform that includes PCIe NVMe boot SSDs and controllers. SMI said that multiple AI infrastructure suppliers have adopted these solutions for servers, DPUs, network systems and AI storage nodes, with emphasis on reliable low-latency boot performance, high endurance, data integrity and enterprise-grade security for long-term stable operations.
In the automotive segment, SMI promoted Ferri-branded automotive storage solutions engineered for smart vehicles and autonomous systems. The company said these products focus on high reliability, functional safety, cybersecurity protection and durability for mission-critical edge computing in vehicles and other embedded platforms. According to the company, the combined portfolio is intended to support the shift of AI workloads from cloud training to on-device inference and physical AI systems.
Article edited by Emily Kuo
