Samsung unveils zHBM prototype at FMS 2026: HBM stacked on AI accelerators for up to 8x bandwidth
Kyrannio · x · 2026-09-14
Samsung showcased mockups of zHBM and zNAND-O, plus 400-plus-layer V10 BV-NAND, at FMS 2026 in California.
- zHBM architecture: vertically stacks HBM on top of AI accelerators (xPUs) along the z-axis, minimizing data travel distance to boost bandwidth and power efficiency — Samsung claims up to 8x performance versus conventional layouts, targeting large-scale AI training and inference.
- Context: traditional designs place HBM around the processor; the new approach moves to wide-bus, on-package low-latency designs. Samsung aims to strengthen its position in the fast-growing AI memory market.
- Epic CEO Tim Sweeney amplified the news, calling the evolution of flash toward HBM-like high bandwidth and on-package low latency exciting.
More from Infra
- Your AI agent would cost 260% more without prompt caching, Arena data shows — arena · 2026-10-30
- Nebius runs auction-driven GPU pricing, lifting rates ~20% amid tight supply to 2028 — BenBajarin · 2026-10-30
- macOS update silently re-adds eMudhra root certs users had removed — steipete · 2026-10-30
- Bittensor's Terza Inference routes one API across miners' models on subnet 103 — const_reborn · 2026-10-30
- SGLang Summit 2026 sells out sponsor booths and Full Passes ahead of Nov 12–13 event — ying11231 · 2026-10-30
- OpenRouter rebuilds provider routing: 46% fewer capacity errors, 16.9% faster TTFT — steipete · 2026-10-30