When trading stocks, look to the research reports by the Golden Qilin analysts—authoritative, professional, timely, and comprehensive—helping you uncover potential themes and opportunities!
（Source: Zhitong Finance）
According to the Zhitong Finance app, as the AI computing power race moves deeper into uncharted territory, storage industry giants SK Hynix (SKHY.US) and SanDisk (SNDK.US) jointly unveiled the first set of standards for high-bandwidth flash (High Bandwidth Flash, HBF) at the “Future Memory & Storage Summit 2026” (FMS 2026) held in Santa Clara, California, USA.
This landmark achievement—driven jointly by the industrial alliance over approximately six months—marks the AI storage track’s official arrival of a “major player.” HBF is positioned as a new storage tier between high-bandwidth memory (HBM) and solid-state drives (SSDs). It aims to meet the growing AI inference demand with a solution that balances both high bandwidth and large capacity.
A technical breakthrough: solving the “memory wall” problem in the AI era
As AI expands from model training to large-scale inference applications, traditional storage architectures are facing serious challenges: HBM offers extremely high bandwidth but is expensive and limited in capacity, while SSDs have large capacity but insufficient bandwidth. HBF is designed to address this “memory wall” bottleneck.
According to the released specification, HBF achieves a balance between bandwidth and capacity in its technical parameters:
Capacity and stacking: supports 8-layer and 16-layer NAND chip stacking configurations, with a maximum capacity of up to 512GB.
Bandwidth tiering: divided into three performance grades (Grade 1~3), with data transfer capability spanning an extensible range of approximately 0.4TB/s to 3.0TB/s.
An open interconnect standard: using the industry-standard universal chip-to-chip interconnect technology UCIe as the connection interface, laying the foundation for flexible integration of HBF with different types of processors such as GPUs and CPUs.
Notably, this specification is not an enterprise-private standard. Instead, it is made publicly available to the entire industry through OCP (Open Compute Project), the world’s largest open data center technology organization. This means HBF will become a widely adopted open industry standard, helping accelerate the maturation of the entire AI storage ecosystem.
Ecosystem consolidation: Google and Tenstorrent join in
Standardization is only the first step; building an industry ecosystem is the key. Currently, the HBF consortium has successfully attracted Google and AI chip startup Tenstorrent.
During the FMS 2026 conference, SK hynix will showcase its HBF strategy through multiple events:
Keynote speech: SK hynix solution development Senior Vice President Kim Cheon-sung and next-generation product planning Senior Vice President Kang Eokseong will jointly deliver a talk titled “Building efficient AI infrastructure with tiered memory in the era of agentic AI.”
Roundtable discussion: On August 6, SK hynix, SanDisk, and Google DeepMind will jointly participate in a special discussion on “Breaking through the memory wall with HBF.”
Strategic depth: the 10th-generation 375-layer 4D NAND makes its debut
As a “heavy-duty” companion for this release, SK hynix also unveiled for the first time at the FMS 2026 exhibition the 10th-generation (V10) 375-layer 4D NAND flash wafer and product currently under development.
According to SK hynix, compared with the previous generation, the product’s performance-per-power consumption ratio has improved by 2.5x. It is especially suitable for AI data center environments that demand extremely high performance and energy efficiency. The company plans to begin mass production in early 2027 of high-performance, high-capacity enterprise-class solid-state drives (eSSDs) based on this NAND flash.
From the release of the HBF standard to the debut of next-generation NAND technology, SK hynix is building a complete AI storage solution matrix spanning everything from storage media to new storage tiers.
“Rule maker” in the AI storage sector
For investors, the release of the HBF standard has multiple far-reaching implications.
First, SK hynix is evolving from an “HBM leader” to an “AI storage architecture shaper.” Building on its already dominant position in the HBM market, the introduction of HBF further expands its technical roadmap in AI storage.
Second, an open-standard strategy helps accelerate market penetration. By publishing open standards through OCP and attracting ecosystem partners such as Google, SK hynix is working to make HBF the mainstream storage solution for AI inference scenarios.
Third, this layout is expected to open up new incremental markets. As AI moves from training to inference, demand for storage solutions that balance bandwidth and capacity is projected to grow exponentially. The HBF is precisely targeting this emerging market.
SK hynix Executive Vice President Kim Cheon-sung said: “With the rapid adoption of AI applications, we are at a point where we must redesign the overall data processing architecture. Through HBF technology, SK hynix will expand the boundaries of memory and storage, contributing to a new architecture that improves the efficiency of the entire system.”