Nvidia founder to meet SK chief Chey Tae-won during CES 2025
Korea Herald Correspondent
LAS VEGAS -- Nvidia founder and CEO Jensen Huang said Tuesday that Samsung Electronics needs to redesign its high bandwidth memory chips when asked why his company is taking time to adopt its HBM products, but was optimistic the Korean company would succeed.
"They have to engineer a new one (HBM), a new design. But they could do it, and they are working very fast," Huang said during a Q&A session held on the sidelines of CES 2025, the world's largest tech show being held in Las Vegas.
"It is not (been) that long. Of course, Korea is very impatient, which is good."
The demand for Nvidia's advanced graphic processing units skyrocketed amid the global AI boom. Demand for HBMs has also surged, as the advanced DRAM products has become a crucial component for AI processing GPUs.
Among the three memory chip makers that can produce HBMs, SK hynix, the world's second-largest memory chip maker, became Nvidia's main supplier.
Samsung, which is the world's largest memory chip maker by revenue, failed to secure an earlier edge in the lucrative market, and has been struggling to pass Nvidia's product qualification test.
Huang did not elaborate on the reason Samsung has to redesign the HBM chips. But he emphasized that Samsung is "working on it" and that, "there is no question" that Samsung will succeed.
“Remember Samsung created HBM originally? The very first HBM memory that Nvidia ever used was from Samsung. They will recover, it's a great company," the Nvidia chief said.
“I have confidence that Samsung will succeed with HBM memory. I have confidence like tomorrow is Wednesday."
During the session, Huang also confirmed that he will meet with SK Group Chairman Chey Tae-won during the CES event, which runs from Tuesday to Friday. SK hynix, the HBM supplier, is an affiliate of SK Group, Korea's second largest conglomerate.
Chey and Huang are expected to discuss their collaboration on the next-generation HBM4, which is projected to become the main chip in demand this year. The cutting-edge chip is expected to support Nvidia's Rubin GPU architecture, the successor of the Blackwell, which is currently in high demand.
In November, Chey revealed that Huang had asked him to expedite the supply of HBM4 by six months for Nvidia's Rubin chips, slated for launch in 2026.
