Nvidia’s Huang asked SK Hynix to delay shipments of HBM4 chips by 6 months, says SK chairman

By Hyunjoo Jin and Joyce Lee

SEOUL (Reuters) – Nvidia Chief Executive Jensen Huang has asked memory chipmaker SK Hynix to push back deliveries of its next-generation high-bandwidth memory chips called HBM4 by six months, SK Group Chairman Chey Tae-won said on Monday.

South Korean company SK Hynix said in October it aims to deliver chips to customers in the second half of 2025. An SK Hynix spokesman said Monday that the timeline was faster than the original target, without elaborating.

Huang’s demand for faster delivery underscores the demand for faster, more power-efficient chips used in Nvidia’s GPUs to develop artificial intelligence technologies. Nvidia controls more than 80% of the global artificial intelligence chip market.

SK Hynix is ​​leading the global race to meet growing demand for HBM chips, which help process massive amounts of data for training artificial intelligence technology and are critical to Nvidia’s chipsets. But the company faces growing competition from rivals such as Samsung Electronics and Micron.

SK Hynix, which plans to ship the latest 12-layer HBM3E to an undisclosed customer this year, also plans to ship samples of the more advanced 16-layer HBM3E early next year, its CEO Kwak No-Yung said at the SK AI SUMMIT. 2024 in Seoul.

Samsung said last week it was making progress on a supply agreement with an unnamed major customer after delays, adding that it was in talks with major customers to release “enhanced” HBM3E products in the first half of next year. Samsung also plans to launch next-generation HBM4 products in the second half of next year.

SK Hynix shares jumped 5.1%, while Samsung shares rose 1%. The market as a whole grew by 1.6%. (This story has been corrected to say “16-layer HBM3E” rather than “18-layer HBM3E” in paragraph 5.)

(Reporting by Hyunjoo Jin and Joyce Lee; Editing by Ed Davis and Himani Sarkar)