Close Menu

    ACCESS Newswire

    Latest News

    AI Memory Innovations and CXL Collaboration Highlighted at OCP Summit by Samsung and AMD

    Technology

    Samsung Electronics expands its memory and storage strategy for next-generation AI infrastructure. Samsung centered its strategy on enhancing speed, scalability, and data durability. The first pillar emphasizes high-performance memory, with the company showcasing customized high-bandwidth memory and HBM5, its eighth-generation HBM. The second pillar utilizes CXL technology to extend memory capacity beyond what standard servers typically support. The third focuses on data management and storage through cutting-edge enterprise SSDs and software solutions. Samsung asserts that these three elements work synergistically to improve the overall efficiency of AI systems. The company linked its strategy to the increasing demand for memory driven by AI inference workloads and larger models. As models grow in size, they generate more data during inference, necessitating greater memory capacity. Current server configurations often limit the amount of memory accessible to each processor. Samsung suggests that CXL can help overcome these restrictions without requiring a complete system redesign. The company also emphasized key-value cache optimization, which allows AI models to reuse data previously processed instead of recomputing it, thereby improving efficiency. Samsung and AMD Showcase CXL Memory Expansion Solutions At the summit, Samsung and AMD demonstrated a CXL-based memory expansion solution that enables processors to share memory resources and scale capacity flexibly across AI systems. Samsung confirmed that this technology has been validated in real-world system environments. The two companies signed a memorandum of understanding in March 2026, focusing on next-generation AI memory and computing innovations. Additionally, AMD Chief Executive Lisa Su visited South Korea on October