CONNECT WITH US

Nvidia's Jensen Huang requests early shipment of HBM4 from SK Hynix

Amy Fan; Jerry Chen, DIGITIMES Asia 0

Credit: AFP

Nvidia CEO Jensen Huang emphasized during a video call at the SK AI Summit in Seoul's Coex that he aims to strengthen collaboration with SK Hynix in the high bandwidth memory (HBM) sector. Specifically, he requested an expedited supply of the sixth-generation HBM (HBM4).

Advancing the Timeline According to ZDNet Korea, SK Group Chairman Chey Tae-won announced during his keynote speech at the SK AI Summit on November 4 that the company will work to advance the supply timeline of HBM4 by six months in response to Huang's request.

Speaking at the SK AI Summit, Huang emphasized that as machine learning technology continues to evolve, expanding memory bandwidth and improving power efficiency are particularly crucial. He highlighted that through collaboration with SK Hynix, Nvidia has been able to technically surpass the progress dictated by Moore's Law, indicating that more support from SK Hynix's HBM will be needed in the future.

Currently leading the market, SK Hynix supplies Nvidia with cutting-edge HBM, having successfully commercialized the fifth-generation HBM3E. While the next-generation HBM4 was originally slated for mass production in 2026, the timeline may be accelerated due to Nvidia's request.

During the meeting, Chey revealed that Huang personally requested to expedite the supply of HBM4 by six months. When asked about the feasibility of this timeline, SK Hynix President Kwak Noh-Jung assured that they would make every effort to meet the request.

In a notable exchange, Chey shared that when he inquired about Huang's desired timeline, the Nvidia CEO responded by asking how soon SK Hynix could deliver. This interaction highlights Nvidia's significant demand and expectations for HBM4.

Industry observers in South Korea expect SK Hynix to actively respond to Nvidia's needs and accelerate the supply pace of the next-generation HBM.