Samsung Electronics a worldwide chief in superior reminiscence know-how, at the moment showcased its newest AI reminiscence improvements at Future of Memory and Storage (FMS) 2026 in Santa Clara, California.
With broad experience spanning DRAM, NAND, enterprise storage and related superior semiconductor applied sciences, Samsung introduced its newest AI reminiscence portfolio and know-how roadmap, together with a preview of its zHBM and zNAND-O idea fashions, the industry-first introduction of its V10 BV-NAND with 400-plus layers and a complete vary of options designed to advance future AI infrastructure. Samsung can be that includes its AI cloud server-inspired sales space at FMS 2026, the place roughly 30 reminiscence and storage applied sciences are on show.
At the opening keynote occasion, Samsung’s Memory Business executives Jin-Yub Lee, Executive Vice President and Head of Flash Product & Technology and Kyungryun Kim, Vice President and Project Leader of the DRAM Design Team, delivered their deal with on next-generation 3D buildings in future reminiscence options.
Titled “Driving the Wave of AI Revolution: 3D Innovations in Memory & Storage Architecture,” the keynote introduced Samsung’s imaginative and prescient to maximise AI system efficiency and energy effectivity whereas addressing the quickly rising calls for of high-performance computing (HPC) and AI infrastructure.
The firm additionally unveiled V10 BV-NAND, an structure enabled by a brand new wafer bonding know-how — for the primary time within the {industry}.
Samsung’s Vision for Next-Generation 3D Memory
At FMS 2026, Samsung unveiled the {industry}’s first idea fashions of zHBM and zNAND-O, presenting its imaginative and prescient for the subsequent technology of 3D reminiscence architectures.
Designed for superior AI computing, zHBM presents a brand new reminiscence structure that vertically stacks HBM instantly above AI accelerators, transferring past typical designs wherein HBM is positioned alongside the processor.
By minimizing the space that knowledge travels between the AI processor and reminiscence, zHBM is designed to ship larger bandwidth and improved energy effectivity, enabling the ultra-fast knowledge processing required for large-scale AI coaching and inference.
A next-generation interface system incorporating zHBM is predicted to ship roughly eight instances the efficiency of HBM5. With next-generation wafer bonding know-how, zHBM can obtain greater than 10 instances the reminiscence density of HBM5 whereas enhancing vitality effectivity threefold and lowering thermal resistance by greater than half, maximizing the soundness and energy effectivity of high-capacity, high-bandwidth methods.
zHBM additionally helps customer-specific designs, enabling personalized IP to be built-in into the interlayer between the reminiscence and AI accelerator to broaden reminiscence capability and improve accelerator efficiency.
Building on shut collaboration with prospects, Samsung plans to additional optimize the combination of AI processors and reminiscence by way of zHBM to maximise AI system efficiency.
The firm additionally launched zNAND-O, a next-generation high-performance NAND resolution constructed on its V-NAND know-how and in growth in four- and eight-layer variations. By combining excessive area effectivity, improved I/O efficiency and low latency, zNAND-O is optimized for edge AI environments supporting real-time, data-intensive AI functions.
Industry-First V10 BV-NAND
Samsung unveiled V10 BV-NAND, that includes its new Bonding V-NAND structure. The announcement comes 13 years after the corporate launched the {industry}’s first V-NAND know-how on the 2013 Flash Memory Summit, marking one other milestone within the firm’s NAND innovation.
With greater than 400 layers, V10 BV-NAND delivers larger efficiency and energy effectivity whereas considerably growing storage density. By making use of wafer bonding know-how to stack reminiscence cells, Samsung has elevated reminiscence density by roughly 58% over the earlier technology (V9).
Beyond the…
