Seoul: South Korean chipmaker SK hynix stated on Friday that it has arrange a strategic partnership with Taiwan Semiconductor Manufacturing Firm (TSMC) aimed toward bolstering its capabilities in producing premium high-bandwidth reminiscence (HBM) chips and superior packaging applied sciences.
Beneath the Memorandum of Understanding agreed just lately in Taipei, Taiwan, the 2 semiconductor giants will collaborate in creating sixth-generation HBM4 chips, set for mass manufacturing in 2026, in keeping with SK hynix.
Each, SK hynix and TSMC are key prospects of Nvidia, a significant participant within the synthetic intelligence (AI) semiconductor market with its graphics processing models, studies Yonhap information company.
“The collaboration between the worldwide chief within the AI reminiscence house and TSMC, a high world logic foundry, will result in extra improvements in HBM know-how,” SK hynix stated in a press release.
“The collaboration can be anticipated to allow breakthroughs in reminiscence efficiency by way of trilateral collaboration between product design, foundry, and reminiscence supplier,” it added.
The SK hynix-TSMC collaboration’s preliminary focus shall be on enhancing the efficiency of the bottom die, which is mounted on the very backside of the HBM bundle, SK hynix stated.
HBM chips, integral elements used for AI computing, have garnered rising consideration with the rise of functions similar to generative AI, exemplified by fashions like ChatGPT.
SK hynix additionally stated it plans to leverage TSMC’s superior logic course of for the bottom die of HBM4, strengthening its performance to satisfy diversified buyer calls for for enhanced efficiency and energy effectivity.
The Korean chipmaker makes use of a proprietary know-how to make base dies as much as the present HBM3E.
Furthermore, SK hynix and TSMC can even work collectively to optimise the combination of SK hynix’s HBM and TSMC’s packaging course of, CoWoS know-how.
“We anticipate a powerful partnership with TSMC to assist speed up our efforts for open collaboration with our prospects and develop the business’s best-performing HBM4,” stated Justin Kim, head of AI Infra at SK hynix.