Samsung Semiconductor Highlights 400+ Layer V-NAND and zNAND-O at GMIF2026 as Foundation for Agentic AI

SHENZHEN, China, Oct. 15, 2026 /PRNewswire/ -- The 5th GMIF Innovation Summit (Global Memory Innovation Forum 2026, GMIF2026), co-hosted by the Shenzhen Memory Industry Association and the School of Integrated Circuits at Peking University, recently concluded in Shenzhen. Themed "The Future Built on AI Memory and Storage," the summit drew more than 3,000 registered professional attendees for discussions on AI inference, enterprise storage, and cloud-edge-device coordination as AI infrastructure enters a new cycle.

Kevin Yoon, CVP & CTO of Samsung Memory China at Samsung Electronics, delivered a keynote titled "The Memory Revolution: Building the Foundation of Agentic AI," presenting Samsung's full-stack portfolio from server SSDs to on-device zNAND-O. As AI has moved from generative to agentic, multiple agents collaborate on complex tasks and massive numbers of tokens have been generated, driving up KV cache capacity and reuse needs. Once KV cache outgrows GPU memory, an independent shared storage tier becomes inevitable.

PM1763 and BM1773: A Two-Track Server SSD Lineup

Yoon introduced two server SSDs for KV cache shared storage. The performance flagship PCIe Gen6 SSD PM1763 combines V-NAND TLC with an enhanced cooling system and will sample in Q4 2026. The capacity flagship BM1773, built on V9 QLC, spans 16TB to 256TB and is billed as "the world's smallest 256TB SSD."

"With these two SSDs, we will drive both the performance and capacity markets," Yoon said.

V10: The Industry's First 400+ Layer V-NAND

At the NAND process level, Yoon introduced Samsung's latest V10 BV-NAND, the industry's first V-NAND with over 400 layers. Compared with V9, it delivers 58% higher bit density, 33% faster I/O, and 25% lower power consumption. The key, Yoon said, is advanced bonding technology: memory cells and peripheral circuits are manufactured separately and then bonded vertically, enabling high-density stacking.

zNAND-O: 10x the Bit Density of DRAM for On-Device AI

With AI models expected to exceed one trillion parameters by 2030, keeping all weights in DRAM under the traditional UMA would sharply raise on-device costs. Samsung's answer is zNAND-O, a next-generation NAND architecture designed for on-device AI, now in development in 4-layer and 8-layer versions. It is co-packaged with the MPU and connected through the UCIe interface, delivering 10x the bit density of DRAM and 7x the read bandwidth and read energy efficiency of NAND flash.

In a simulation using Moonshot AI's Kimi K2-1T model, an NPU-optimized architecture with zNAND-O cut memory cost to 1/4 at equal performance and fit a model nearly twice as large.

Outstanding AI Memory & Storage Industry Leadership Award

GMIF annually recognized companies driving core technology breakthroughs and compute ecosystem building across the industry chain. Samsung Semiconductor received the Outstanding AI Memory & Storage Industry Leadership Award for its sustained leadership in DRAM and NAND, including the zNAND-O 3D architecture, 400-plus layers V10 NAND, and CXL memory, setting a benchmark for the global memory industry and providing important support for the build-out of global AI infrastructure.

Closing

As an annual gathering for the memory industry, GMIF brought together leaders from across the value chain, and Kevin Yoon's keynote gave the industry a clear view of where memory and storage are heading in the agentic AI era. As KV cache outgrows GPU memory, Samsung is answering at every layer, from PCIe Gen6 and high-capacity server SSDs for shared KV cache, to 400+ layer V10 NAND, to zNAND-O for on-device AI. With its second consecutive GMIF award, Samsung has shown that it is not only keeping pace with agentic AI's demands on memory & storage, but helping define the path forward.

Carina Gu, [email protected] 

SOURCE Shenzhen Memory Industry Association (SMIA)