SK hynix says it shipped samples of its 12-stack HBM4E (High Bandwidth Memory) to major customers, marking the next step in the company’s AI memory roadmap. The company said the samples were delivered on schedule and that it will work closely with partners toward timely mass production.
HBM4E builds on the company’s HBM portfolio with improvements in both bandwidth and power efficiency aimed at AI training and inference workloads. SK hynix said the new memory delivers up to 16Gbps per pin while improving power efficiency by more than 20% compared to previous-generation products.
The company also introduced a new interface and design optimizations intended to reduce data transfer latency and improve stability under high-bandwidth operating conditions. These enhancements are designed to support AI data centers and large-scale computing platforms that require increasingly higher memory throughput.
The 12-stack HBM4E package reaches 48GB of capacity using SK hynix’s Advanced MR-MUF (Mass Reflow Molded Underfill) packaging technology. MR-MUF injects a liquid molding material between stacked dies to improve structural integrity and protect internal circuitry. According to the company, the updated design also improves heat resistance by 17% over HBM4, helping maintain reliability in high-performance AI systems.
SK hynix said the new product leverages manufacturing and packaging experience gained through previous generations, including HBM3, HBM3E, and HBM4. The company expects HBM4E to address increasing memory bandwidth requirements as AI infrastructure continues to scale.
Ahn Hyun, President and Chief Development Officer at SK hynix, said the company has laid the foundation to strengthen its AI leadership with HBM4E, built on its technological capabilities and manufacturing expertise. He added that SK hynix will work closely with partners to deliver the value the market needs while reinforcing its technology leadership as a full-stack AI memory creator.
Amazon