Samsung Electronics Co., Ltd. unveils next-generation memory strategy, including HBM5, in the U.S.

[Financial News] Samsung Electronics Co., Ltd. has unveiled a strategy for innovating AI infrastructure that spans next-generation high-bandwidth memory (HBM), memory expansion technology and storage solutions at a global information technology (IT) infrastructure event. It also jointly demonstrated next-generation memory expansion technology with Advanced Micro Devices, Inc. (AMD) and showcased collaboration results with global AI companies, including NVIDIA Corporation, broadening its technology partnerships beyond memory supply to AI systems as a whole.
Samsung Electronics Co., Ltd. said on the 13th that it had taken part in the Open Compute Project (OCP) Global Summit 2026, which opened on the 12th (local time) in San Jose, California, and unveiled next-generation memory and storage technologies as well as solutions to optimize AI systems.
At the event, it presented three technological approaches to improving AI infrastructure performance: high-performance memory such as HBM5 and custom HBM; memory expansion technology based on Compute Express Link (CXL); and data storage and management technology using next-generation solid-state drives (SSDs). It also jointly demonstrated a CXL-based memory expansion solution with Advanced Micro Devices, Inc. (AMD) and showcased collaboration results with global AI companies by exhibiting a Groq LPU wafer and an NVIDIA Corporation Groq 3 LPX system, among other products.
In a presentation that day, Vice President Shim Ho-jun of Samsung Electronics Co., Ltd.'s AE Team stressed, "As AI models grow larger and the amount of data and KV cache capacity that must be handled during inference increases, capacity expansion and data storage and management technologies are becoming as important as memory bandwidth for improving AI system performance." Shim Ho-jun also presented a three-pronged technology strategy to meet these AI infrastructure demands: speed, memory expansion and data persistence.
Samsung Electronics Co., Ltd. is also expanding collaboration with global customers and partners on co-design and technology validation from the early stages of development, aiming to optimize AI systems as a whole—including central processing units (CPUs), graphics processing units (GPUs), memory and storage—rather than focusing solely on improving individual memory products. At this OCP event, it plans to jointly demonstrate a CXL-based memory expansion solution with Advanced Micro Devices, Inc. (AMD) and showcase the results of its technology collaboration to flexibly expand memory capacity in AI systems.
As the spread of AI agents increases the amount of data and memory required for inference, the aim is to offer a way to address the memory capacity limits of existing systems using CXL technology. In particular, the company will showcase memory pooling technology based on CXL Memory Module DRAM (CMM-D), which enables efficient use of memory resources across multiple systems and flexible access to the memory capacity needed in AI environments.
Samsung Electronics Co., Ltd. said, "This joint demonstration is significant because it validates the technology in a real-world system environment with a global computing platform company and shows that we are expanding our collaboration to improve AI infrastructure performance."
Samsung Electronics Co., Ltd. is also running two exhibition zones at the event to introduce memory and storage solutions required for AI computing and inference. The “Tokenization Solution” exhibit, which features high-performance AI memory technologies, will showcase HBM4E, HBM5, custom HBM and LPDDR5X-PIM. The company will also exhibit a wafer made using a 2-nanometer Gate-All-Around (GAA) process, a Groq LPU wafer and an NVIDIA Corporation Groq 3 LPX system. The “KV Cache Solution” exhibit will present ways to efficiently store and manage the large volumes of data generated during AI inference using CXL memory, enterprise SSDs and KV cache optimization technologies.
Samsung Electronics Co., Ltd. is also strengthening co-design and technology validation partnerships with global customers. Through the Samsung MAX Lab, it validates next-generation memory and storage technologies in real-world AI environments and incorporates customer requirements from the early stages of product development. It is also participating in open-source software development and standardization activities through GOST, its dedicated organization for the global open ecosystem, broadening collaboration beyond hardware to software optimization.
This year, Samsung Electronics Co., Ltd. has been working to strengthen its competitiveness in AI semiconductors by successively unveiling next-generation memory technologies, including HBM5 and zHBM, at major global events such as NVIDIA Corporation's GTC, Computex and FMS. According to market research firm Counterpoint Research, Samsung Electronics Co., Ltd. ranked first in the global DRAM market in the second quarter of this year, with a 39% share by revenue. It also retained the top spot in the NAND flash market over the same period, with a 29.3% share.
Samsung Electronics Co., Ltd. said, "We will step up our efforts to capture high-value AI memory markets, including HBM, server DDR5 and enterprise SSDs, leveraging our technological capabilities and manufacturing base."

[email protected] Im Su-bin Reporter