Samsung Semiconductor’s Post

The exhibit hall doors are officially open at the 2026 OCP Global Summit! We’re thrilled to welcome attendees to booth C41 to experience firsthand how Samsung is scaling the next generation of AI infrastructure. Here is a look at the cutting-edge solutions we are showcasing: Scaling efficiency of KV caching with CMM-D MD310, PM1763, BM1773, and SOCAMM2: 💠CMM-D MD310: Samsung’s CXL 3.2-based 256GB memory expander supporting the PCIe Gen6 x8 interface with blazing bandwidth up to 72GB/s. 💠PM1763: Samsung’s next-generation PCIe Gen6 enterprise SSD designed for AI servers and AI infrastructure.  💠BM1773: A 16-channel PCIe Gen5 NVMe SSD featuring 9th-generation 2Tb QLC V-NAND, delivering massive capacities of up to 256245.76TB with top-tier performance for data-intensive workloads. 💠SOCAMM2: Our next-gen LPDDR-based server memory module, powered by LPDDR5X technology, purpose-built for high-density AI computing environments. Scaling throughput for maximized tokenization with HBM4E, LP5X-PIM, and Groq 3 LPU: 💠HBM4E: Our next-generation high-bandwidth memory solution engineered to support future AI accelerators and large-scale AI infrastructure. 💠LP5X-PIM: Integrates a Processing-in-Memory block into each of the 16 DRAM banks, enabling highly parallel computation and delivering up to 8× higher effective bandwidth than conventional LPDDR5X (as demonstrated in evaluations with Samsung Edge AI accelerator SoCs). 💠Groq 3 LPU: Built on Samsung Foundry’s 4nm process technology for low-latency AI inference, supporting massive AI workloads. Our products and technologies are designed to scale the future of AI infrastructure. If you're at OCP, stop by booth C41—we’d love to connect and show you what's next! https://lnkd.in/eiyB_FpV #SamsungSemiconductor #SamsungMemory #OCPSummit2026 #KVCache #Tokenization #SamsungMAXLab #SamsungGOST