SK Hynix announces HBM4E: Next generation of high-performance memory coming in 2026

HBM as a key component for AI progress

In the AI industry, HBM is considered a critical component for performance advancements due to its immense importance in modern AI accelerators. The widespread adoption of the HBM3E standard in newer AI GPUs such as the Blackwell B100 and Instinct MI300X has recently resulted in a significant performance boost. However, SK Hynix shows that this is just the beginning: the Korean memory manufacturer plans to launch HBM4E on a large scale over the next two years.

Accelerated innovation cycles

In a recent announcement, Kim Gwi-wook, Head of HBM Advanced Technology at SK Hynix, emphasized that progress at HBM has reached new heights. While previous generation changes took place after two years, industry demand has shortened this period to one year. This indicates that SK Hynix is indeed planning to introduce HBM4E by 2026.

HBM4E: Details and improvements

The unveiling of HBM4E marks a milestone in memory technology. SK Hynix has not only confirmed the existence of the standard, but has also revealed the first details of the process. The bandwidth of HBM4E is said to be 1.4 times higher than the previous generation, HBM4, while offering significantly improved energy efficiency. These improvements give hope for enormous performance increases in next-generation AI accelerators.

MR-MUF technology and TSMC partnership

The first developments around HBM4 emerged some time ago when the Korean giant announced the MR-MUF development technology. This technology aims to integrate logic and memory semiconductors in a single package. To achieve this goal, SK Hynix has already formed an alliance with TSMC.

HBM4 as an “iPhone moment” for the industry

Industry experts predict that HBM4 will be the segment’s “iPhone moment” and set new benchmarks for future standards. The combination of significantly increased bandwidth, improved energy efficiency and innovative integration technology from SK Hynix and TSMC promises groundbreaking advances in the field of AI accelerators and paves the way for new applications and opportunities in the world of artificial intelligence.

Source: EtNews