SEOUL SEOUL: Nvidia CEO Jensen Huang has asked SK Hynix to deliver its next-generation high bandwidth memory (HBM) called HBM4 six months ahead of schedule amid surging demand for artificial intelligence (AI) computing chips, SK The group's chairman, Che Tae-won, said on Monday. “The last time we met, he asked me to bring forward the HBM4 supply schedule by six months, which was already set,” Che said at SK AI in Seoul. Said in his keynote address at Summit 2024.
SK Hynix is a major supplier of HBM chips to Nvidia. It became the world's first company to supply fifth-generation eight-layer HBM3E chips to Nvidia in March, and recently began mass production of its advanced 12-layer HBM3E chips, Yonhap news agency reported. The company had previously announced plans to supply upcoming HBM4 chips in the second half of next year.
Huang's request highlighted growing demand for Nvidia's AI accelerators, which rely heavily on HBM chips for data- and energy-intensive tasks as AI technology moves beyond text-based models. During one session, Nvidia's CEO emphasized the partnership with SK hynix in a video message, calling the latter's development plan “super aggressive” but “super necessary.”
In response to the increasing demand for high-performance HBM, SK hynix CEO Kwak Noh-jung announced the development of 16-layer HBM3E chips, which is planned to be produced in the first half of next year.Meanwhile, a keynote speech In 2017, Che emphasized the trilateral partnership between SK hynix, Nvidia and Taiwan Semiconductor Manufacturing Company (TSMC) in building the supply chain of the global AI chip industry. “When Nvidia comes out with newer, more advanced GPUs, it needs more HBM chips,” Che explained. “So SK Hynix is working hard to stay ahead of the curve. Developing advanced HBM products and maintaining high yields is challenging, so close cooperation with Nvidia is essential.” Che said. Also emphasized on close partnership with contract manufacturing company TSMC.
Comments are closed.