Samsung has become one of the first HBM manufacturers to announce progress around HBM4E at the OCP, and by the looks of it, the memory will feature gigantic generational improvements.
Samsung's HBM4E Process Becomes One Of The Fastest Memory Types Out There, Reaching 13 TB/s Pin Speeds Per Stack
The Korean giant has been quite busy with its HBM business over the past few days, particularly after securing decisive contracts with companies like NVIDIA and AMD for its memory products. At the Open Compute Project (OCP) Global Summit, Samsung has showcased the future of its HBM products, detailing both HBM4 and HBM4E. In particular, with the latter process, the Korean giant plans to deliver speeds of up to 13 Gbps per stack, reaching a bandwidth of 3.25 TB/s. There's also an important development around HBM4, which we'll discuss ahead.
Samsung's HBM4E module will also debut with excellent power efficiency figures, which are claimed to be almost double those of the current HBM3E counterpart. More importantly, the HBM4 process of the Korean giant has achieved a pin speed of 11 Gbps, which is significantly higher than the standards set by authorities such as JEDEC. The push comes after NVIDIA's request to ramp up HBM4 solutions to boost the performance of its Rubin architecture, and Samsung has become the first to achieve this feat, according to local media reports.
Samsung Electronics, which fell behind its competitors in HBM3E, has been targeting higher bandwidth than its competitors since the early stages of HBM4 development. While the 'speed battle' it waged in HBM4 is on the verge of success, its strategy is to move quickly in the next generation as well to turn the tables.
- SedailyApart from the ramped up performance, Samsung is also claimed to be working on the price structure of HBM4 supply to NVIDIA and others, offering a tag that might be 'too attractive to decline' according to reports. Since a crucial part of HBM4 features semiconductor (4nm) onboard, Samsung has its own foundry division to cater to the HBM manufacturing, which means that it can control profit margins around the memory module.
The Korean giant is claimed to factor in little margins to ensure that it becomes a major HBM supplier, which means that both SK hynix and Micron have severe competition coming ahead. As far as when we could see HBM4E debuting in the industry, the timeline is set for early 2026, right alongside the mass production of HBM4.
News Source: Sedaily
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
