SK Group Chairman Chey Tae-won revealed that Jensen Huang, CEO of Nvidia, has requested an acceleration of the supply schedule for the sixth-generation high-bandwidth memory (HBM), HBM4, by six months.
During his keynote speech at the SK AI Summit 2024 held in Seoul on November 4th, Chey described a recent meeting with Huang, stating, "Huang is an engineer to his core, and he's almost like a Korean. He wants everything to be done quickly." Chey added, "Every time Nvidia releases a new GPU, they ask SK Hynix for more HBM and always request an earlier delivery schedule. In my last meeting with Huang, he asked us to expedite the supply of HBM4 by six months."
"I replied, 'I can't do it; the person who can should do it,'" Chey recounted. "When I asked SK Hynix CEO Kwak No-jung if it was possible, he said he would do his best."
Chey also expressed his apprehension about future meetings with Huang, saying, "I'm a bit afraid to meet him again. He might ask me to speed up the schedule even more." He praised Nvidia's leadership in the AI chip market, attributing the company's success to Huang's emphasis on speed.
SK Hynix began supplying Nvidia with HBM3E 8-layer in March and started mass production of HBM3E 12-layer products last month. The company plans to ship HBM3E 16-layer products by the end of the year and provide samples of HBM3E 16-layer in early 2025.
Originally scheduled for release in 2026, the HBM4 12-layer product is now planned for release in the second half of next year, in response to Huang's request.
In a meeting with reporters after the keynote speech, Chey emphasized that the agreement to accelerate the HBM4 supply schedule was based on mutual commitment. He stated that the expedited timeline would only be meaningful if the product passed quality tests. This suggests that SK Hynix is confident in the quality of its HBM4 product.
During his keynote speech, Chey highlighted SK's commitment to collaborating with global partners to address bottlenecks in AI development. He specifically mentioned TSMC, stating, "No matter how good a chip is designed, it's meaningless if it can't be actually manufactured. SK, along with Nvidia, is working closely with TSMC to address the global shortage of AI chips."
Other global partners, including Nvidia's Jensen Huang, TSMC's Wei Zhejia, and Microsoft's Satya Nadella, also expressed their commitment to strengthening partnerships with SK.
[Copyright (c) Global Economic Times. All Rights Reserved.]
