Samsung Unveils Next-Gen AI Memory and 256TB SSDs

Samsung Electronics unveiled a series of next-generation memory and storage technologies designed for AI data centers and agentic AI during industry summits in San Jose and Shenzhen. At the 2026 Open Compute Project Global Summit, the company focused on computing speed, memory expansion, and data persistence. A key highlight included a joint demonstration with Advanced Micro Devices of a Compute Express Link-based memory pooling solution using CMM-D technology to allow flexible resource sharing during AI inference.

Continuing its rollout at the 5th GMIF Innovation Summit, Samsung introduced the V10 BV-NAND, the first V-NAND with over 400 layers, which provides 33% faster I/O and 58% higher bit density than previous versions. To support KV cache demands in AI inference, the company debuted two server SSDs: the PCIe Gen6 PM1763 performance flagship and the BM1773, the world's smallest 256TB SSD. For on-device AI, Samsung presented zNAND-O, an architecture offering 10 times the bit density of DRAM.

Other showcased technologies included HBM5, 2-nanometer gate-all-around wafers, and the MAX Lab for real-world AI workload testing. These developments follow a March memorandum of understanding with Advanced Micro Devices and a recent visit to Korea by AMD CEO Lisa Su to discuss foundry and memory cooperation. Samsung Semiconductor received the Outstanding AI Memory & Storage Industry Leadership Award during the Shenzhen summit.