Skip to content

Nvidia acknowledges shift: SK hynix seizes position as world's leading DRAM producer, ending Samsung's reign after more than three decades

Samsung Electronics, the long-standing global leader in DRAM production, has been dethroned after nearly 3 decades. SK hynix took over the top spot, propelled by rising demand for AI memory chips and an exclusive contract with Nvidia, securing their supply.

Nvidia acknowledges shift: SK hynix tops Samsung as global leader in DRAM production for the first...
Nvidia acknowledges shift: SK hynix tops Samsung as global leader in DRAM production for the first time in over thirty years.

Nvidia acknowledges shift: SK hynix seizes position as world's leading DRAM producer, ending Samsung's reign after more than three decades

In a modal window presented to readers, this article explores the latest trends and developments in High Bandwidth Memory (HBM) technology, with a particular focus on its role in supporting Artificial Intelligence (AI), machine learning (ML), and high-performance computing (HPC) applications.

Chapter 1: Market Growth and Demand

HBM shipments are set to grow by approximately 70% annually in 2025, driven by the high bandwidth and low latency requirements of AI processors and data centers performing data-intensive tasks. This surge is expected to propel the overall HBM market from $2.36 billion in 2024 to $3 billion in 2025, with a long-term compound annual growth rate (CAGR) of around 26.8% through 2029, fueled by AI, 5G, edge computing, IoT, VR/AR, and HPC.

Chapter 2: Technological Roadmaps

Major industry players, including Samsung, SK hynix, and Micron, are progressing toward HBM4 and beyond, with HBM4 anticipated in 2026 and HBM4E expected a year later. Future plans include extremely wide interfaces (e.g., 16,384-bit for HBM8 by 2038) and embedded NAND, indicating ongoing scaling in bandwidth and capacity.

Chapter 3: Industry Innovation

SK hynix is collaborating with Sandisk to advance the standardization and development of High Bandwidth Flash (HBF) memory. The goal is to deliver memory solutions with bandwidth comparable to HBM but significantly higher capacity and at similar cost, targeting AI inference workloads.

Chapter 4: Near-Memory and In-Memory Computing

HBM technology is viewed as a critical step toward the eventual "in-memory computing/processing" era, leveraging its 3D stacked DRAM design with high bandwidth and energy efficiency for HPC, GPUs, AI, and data centers.

Chapter 5: Market Drivers

The surge in AI/ML workloads, increasing data center demands, and evolving edge AI applications are primary drivers of innovation and adoption in HBM technology, with Samsung and SK hynix leading these advancements.

Chapter 6: LG's Role

While LG is a major semiconductor player, recent developments or announcements in HBM technology by LG were not highlighted in the latest sources. The primary focus for HBM innovation lies with Samsung Electronics and SK hynix among Korean manufacturers, alongside global competitors like Micron and emerging players such as CXMT.

In conclusion, SK hynix and Samsung Electronics continue to pioneer advances in HBM technology aimed at greater speed, capacity, and energy efficiency to meet the booming AI and HPC market demands, supported by collaborations and strategic standardization efforts. The HBM market is rapidly evolving with a bright growth outlook through the end of the decade.

The article has captions settings, and the dialog can be closed or minimised. It was originally written in Korean and translated into English using bilingual reporter and generative AI tools. The translation was reviewed and refined by a native English-speaking editor. The Escape key cancels and closes the window.

  1. The rapid evolution of High Bandwidth Memory (HBM) technology, driven by companies like Samsung and SK hynix, is anticipated to be fueled by data-intensive business tasks and the increasing demands in the technology sector, particularly in data-and-cloud-computing, AI, 5G, edge computing, IoT, VR/AR, and HPC.
  2. To address the rising AI inference workloads and support the growing market, SK hynix is collaborating with Sandisk on High Bandwidth Flash (HBF) memory, aiming to deliver solutions with bandwidth comparable to HBM and higher capacity at similar costs, thereby contributing to the data-and-cloud-computing industry.

Read also:

    Latest