SK hynix and Western Digital (Sandisk) have announced a joint initiative to standardize High Bandwidth Flash (HBF), a next-generation memory technology poised to revolutionize AI inference systems. This collaboration aims to improve the performance and efficiency of artificial intelligence workloads.
The two leading chipmakers officially launched the HBF Spec Standardization Consortium at Western Digital’s headquarters in California. They also unveiled plans to establish a dedicated workstream within the Open Compute Project (OCP), the world’s largest open data center technology community.
This workstream will concentrate on defining industry-wide specifications for HBF, a technology designed to bridge the performance gap between high-bandwidth memory (HBM) and solid-state drives (SSDs). While HBM excels in providing ultra-fast speeds and SSDs offer vast storage capacity, HBF seeks to integrate the strengths of both to better address the evolving requirements of modern AI infrastructure.
A workstream is a collaborative framework where industry participants jointly develop technical standards and specifications for emerging technologies.
According to SK hynix, HBF is specifically designed to meet the growing demands for increased memory capacity and enhanced power efficiency as AI workloads transition from model training to inference – the critical stage where AI systems process user requests in real-time.
Current memory architectures often struggle to achieve an optimal balance of speed, scalability, and energy efficiency in AI inference environments, the company stated.
By effectively bridging the gap between the high-speed capabilities of HBM and the storage density of SSDs, HBF has the potential to significantly improve system scalability and reduce the overall total cost of ownership (TCO) for AI data centers.
“By establishing HBF as an industry standard in partnership with Western Digital, our goal is to create a foundation for the collective growth of the AI ecosystem,” SK hynix stated in a press release.
Ahn Hyun, President and Chief Development Officer at SK hynix, emphasized the importance of broader collaboration in the age of AI.
“The key to AI infrastructure lies not only in competing over individual component performance but also in optimizing the entire ecosystem,” Ahn explained. “Through HBF standardization, we will establish a collaborative framework and introduce a memory architecture that is specifically optimized for the AI era.”
The companies are committed to pursuing both standardization and commercialization of HBF, anticipating that industry demand for advanced memory solutions like HBF will continue to grow significantly towards 2030 and beyond.
