Source: Wall Street See
Author: Zhao Ying
sk hynix revealed the progress of its next-generation products, planning to launch a 16-layer HBM3E chip in early 2025, with 12-layer HBM3E chips already in mass production since September, and it also plans to introduce HBM5 chips between 2028 and 2030.
Nvidia still can't provide enough GPUs to meet the demand, and it requires SK Hynix to provide HBM4 chips six months in advance.
On Monday, SK Hynix Chairman Choi Tae-young stated at the group's AI summit held in Seoul that SK Hynix is working with Nvidia to address supply bottleneck issues. He mentioned that Nvidia CEO Huang Renxun requested him in a recent meeting to accelerate the timetable for SK Hynix's next-generation high-bandwidth memory chip HBM4 by six months.
It is worth mentioning that in October, SK Hynix revealed that it is expected to supply HBM4 chips to customers in the latter half of 2025. Choi Tae-young mentioned that the timetable for HBM4 is faster than the initial target, but did not provide further details.
Huang Renxun's request to speed up delivery highlights the market's strong demand expectations for Nvidia's next-generation GPUs, which will include new HBM chips. Currently, Nvidia holds over 80% of the market share in the AI chip market.
Furthermore, on Monday, SK Hynix also revealed the latest progress on its next-generation products. SK Hynix will launch 16-layer HBM3E chips in early 2025, with mass production of 12-layer HBM3E chips starting in September. At the same time, it plans to introduce HBM5 chips between 2028 and 2030.
Editor/Jeffy