SK Hynix: HBM4 is ready for initial mass production, with power efficiency expected to improve by 40%
```
South Korean memory chip manufacturer SK Hynix has announced the completion of next-generation HBM4 memory development and has achieved the world's first conditions for large-scale mass production, marking a further consolidation of the company's technological leadership in the field of artificial intelligence memory.
On Friday, SK Hynix announced that the new product doubles bandwidth by utilizing 2,048 I/O terminals, improves power efficiency by more than 40% compared to the previous generation, and achieves an operating speed exceeding 10Gbps, far surpassing the JEDEC standard requirement of 8Gbps.
The company expects that after the application of HBM4 products, AI service performance can be improved by up to 69%, which will help solve data bottleneck issues and significantly reduce power costs for data centers.
With the sharp increase in AI demand and the surge in data processing needs, the demand for high-bandwidth memory continues to rise, while the increase in data center operating power consumption has made memory power efficiency a key customer requirement.
Dual Enhancement of Performance and Efficiency, Lowering Risk with Mature Processes
SK Hynix's HBM4 achieved significant breakthroughs in technical specifications. The product uses 2,048 I/O terminals, doubling the number compared to the previous generation, thereby achieving a doubling of bandwidth. At the same time, power efficiency improves by over 40%, and operating speed exceeds 10Gbps, far surpassing the industry standard of 8Gbps.
The company predicts that once HBM4 products are applied, AI service performance can be increased by up to 69%. This performance boost will directly address the data bottleneck issues currently faced by AI systems and bring significant savings in power costs for data centers.
To ensure the stability of large-scale mass production, SK Hynix used market-proven Advanced MR-MUF and 10nm processes in HBM4; these mature processes are aimed at minimizing technical risks during mass production.
Joohwan Cho, head of HBM development at SK Hynix, stated:
The completion of HBM4 development will become a new milestone for the industry. By providing products that meet customer needs in terms of performance, power efficiency, and reliability in a timely manner, the company will achieve timely market entry and maintain its competitive position.
Justin Kim, President of AI Infrastructure at SK Hynix, pointed out that HBM4 is "a symbolic turning point that surpasses the limitations of AI infrastructure," and will become a core product for overcoming technological challenges. The company plans to develop into a full-stack AI memory provider by supplying optimal quality and diverse performance memory products needed for the AI era in a timely manner.
With the dramatic growth of AI demand and a surge in data processing needs, the demand for high-bandwidth memory required for faster system speeds is soaring. SK Hynix believes that HBM4, with higher bandwidth and power efficiency, will be the optimal solution to meet customer needs.
Risk Warning and DisclaimerThe market has risks, and investment should be cautious. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are appropriate for their specific situation. Any investments made based on this information are at your own risk. ```