SK hynix has shipped samples of its 12-layer HBM4E, a next-generation high-bandwidth memory product, to major customers. The company delivered the 12-high HBM4E samples on schedule, citing its expertise in HBM development and production. It plans to collaborate with partners to prepare for timely mass production. The 12-layer HBM4E reaches a maximum speed of 16Gbps per pin. It provides improvements in both performance and power efficiency, with power efficiency increasing more than 20 percent compared to previous models. These specifications support data processing for AI training and inference.
The product reduces data transfer latency through its latest interface and design optimization, while maintaining stable operation in high-bandwidth environments. This is intended to help customers improve efficiency in AI datacenters and large-scale computing systems.
SK hynix applied Advanced MR-MUF (Mass Reflow Molded Underfill) technology to the HBM4E. This process stacks semiconductors by injecting liquid protective materials between chips. It enables a 48GB capacity in the 12-layer configuration and improves heat resistance by 17 percent compared to the preceding HBM4, supporting structural stability in high-performance computing.
The company has previously supplied HBM3, HBM3E, and HBM4 products. It states that its experience in mass production and supply will help address bottlenecks in next-generation AI infrastructure.
“SK hynix has laid the foundation to strengthen its AI leadership with HBM4E based on its market-leading technological capabilities and manufacturing expertise,” said Ahn Hyun, President and Chief Development Officer. “Through close collaboration with our partners, we will deliver the value needed in the market while reinforcing our technology leadership as a full-stack AI memory creator.”
