Samsung will achieve HBM chip supply goal, but needs new design, says Nvidia CEO

Nvidia CEO Jensen Huang speaks throughout a press Q&A session at Fontainebleau Las Vegas, Tuesday (native time). Korea Occasions photograph by Nam Hyun-woo

LAS VEGAS — Nvidia CEO Jensen Huang expressed optimism, Tuesday (native time), about Samsung Electronics’ efforts to produce its most superior high-bandwidth HBM reminiscence chip to his firm. Nevertheless, Huang careworn the necessity for the Korean chipmaker to „engineer a brand new design.“

Throughout a press Q&A session on the sidelines of CES 2025, the unreal intelligence (AI) guru stated Samsung is “engaged on” Nvidia’s qualification checks for HBM3e and there’s “no query they are going to succeed.”

“I’ve confidence that Samsung will succeed with HBM. I’ve confidence like tomorrow is Wednesday,” Huang stated.

The remarks have been made in response to a query about why he selected to focus on Micron’s G7 reminiscence chip, which is utilized in Nvidia’s new RTX 50 collection graphics playing cards, unveiled throughout his CES keynote speech. The RTX 50 collection is powered by Micron’s Blackwell-based AI applied sciences, and following Huang’s feedback, the U.S. chipmaker’s inventory noticed a big rise.

2025-01-08 15:41  |  Tech

“Bear in mind, Samsung created HBM,” he stated. “Initially, the very first HBM reminiscence that Nvidia has ever used was from Samsung.”

Nvidia CEO Jensen Huang provides a keynote presentation at CES 2025 in Las Vegas, Monday (native time). UPI-Yonhap

Samsung Electronics has been struggling to get its HBM3e chips to move qualification checks for Nvidia’s AI processors, whereas its rival, SK hynix, has change into the dominant provider, offering 12-layer HBM3e chips to Nvidia.

In October, Samsung stated in an earnings convention name that it has achieved “significant progress” in assembly qualification necessities for a “main buyer,” assumed to be Nvidia. However the chipmaker has but to obtain an official verification from Nvidia for its eight-layer and 12-layer HBM3e chips.

The challenges surrounding HBM have raised issues about Samsung’s competitiveness because the world’s largest reminiscence chipmaker. As HBM continues to account for less than a small share of Samsung’s chipmaking enterprise, a decline within the conventional DRAM market has damage the corporate’s total earnings.

Towards this backdrop, Huang expressed confidence that Samsung will get well and stated that Nvidia’s testing of its HBM3e chips shall be accomplished quickly. Nevertheless, he emphasised that the corporate must develop a brand new design.

“Korea could be very impatient, which is an effective factor,” Huang stated when requested in regards to the delays in Samsung’s chip checks. “However they should engineer a brand new one, new design. However they will do it. They’re working very quick and really dedicated to do it.”

Thus far, trade observers have speculated that the delay in qualifying Samsung chips is because of a scarcity of optimization with Nvidia’s chip specs, moderately than a difficulty with the chip high quality itself. Nevertheless, Huang’s feedback recommend that additional delays in Samsung’s bid to produce Nvidia might happen until there’s a basic change within the reminiscence chip design.

Huang additionally famous that Samsung and SK hynix are two of Nvidia’s largest suppliers, praising them as “wonderful reminiscence firms” which is able to “proceed to succeed” given the excessive demand for reminiscence chips in Blackwell structure. Throughout his keynote speech, Huang showcased the GB200 Grace Blackwell Superchip, which is powered by a number of HBM chips.

After the Q&A session, Huang talked about that he’ll meet SK Group Chairman Chey Tae-won throughout CES, however didn’t reveal any additional particulars in regards to the assembly.