Exclusive-Nvidia releases Samsung HBM3 chips for use in Chinese market processors, sources say

By Fanny Potkin and Heekyong Yang

SINGAPORE/SEOUL (Reuters) – Samsung Electronics’ fourth-generation high-bandwidth memory or HBM3 chips have been cleared by Nvidia for use in its processors for the first time, three people briefed on the matter said.

But it’s a silent green light, as Samsung’s HBM3 chips will, for now, only be used in a less sophisticated Nvidia graphics processing unit (GPU), the H20, which was developed for the Chinese market in compliance with the US export controls. people said.

It was not immediately clear whether Nvidia would use Samsung’s HBM3 chips in its other AI processors or whether the chips would have to undergo additional testing before that happens, they added.

Samsung also has not yet met Nvidia’s standards for fifth-generation HBM3E chips and testing of those chips continues, added the people, who declined to be identified because they were not authorized to speak to the media.

Nvidia and Samsung declined to comment.

HBM is a type of dynamic random access memory or DRAM standard first produced in 2013, in which chips are stacked vertically to save space and reduce power consumption. A key component of GPUs for artificial intelligence, it helps process large amounts of data produced by complex applications.

Nvidia’s approval of Samsung’s HBM3 chips comes amid growing demand for high-end GPUs created by the generative AI boom that Nvidia and other AI chipset makers are struggling to meet.

There are only three main HBM manufacturers – SK Hynix, Micron and Samsung – and with HBM3 also in short supply, Nvidia is eager to see Samsung set its standards so it can diversify its supplier base.

Nvidia’s need for more access to HBM3 is also likely to grow as SK Hynix – the clear leader in the field – plans to increase its HBM3E production and produce less HBM3, two of the sources said.

SK Hynix declined to comment.

Samsung, the world’s largest memory chip maker, has been trying to pass Nvidia’s tests for HBM3 and HBM3E since last year, but has faced difficulties due to heating and power consumption issues, Reuters reported in May, citing sources.

Samsung said after the Reuters article was published in May that claims Nvidia failed tests due to heating and power consumption issues were false.

H20

Samsung could begin supplying HBM3 for Nvidia’s H20 processor as early as August, according to two of the sources.

The H20 is the most advanced of three GPUs that Nvidia adapted for the Chinese market after the US tightened export restrictions in 2023, aimed at preventing advances in supercomputing and AI that could benefit the Chinese military.

The story continues

In line with US sanctions, the H20’s computing power was significantly limited compared to the version sold in markets outside of China, the H100.

The H20 initially got off to a weak start when deliveries began this year and the US company priced it below a rival chip from Chinese tech giant Huawei, Reuters reported in May.

But sales are now growing rapidly, separate sources said.

Unlike Samsung, SK Hynix is ​​the main supplier of HBM chips to Nvidia and has been supplying HBM3 since June 2022. It also began supplying HBM3E in late March to a customer it declined to identify. The shipments went to Nvidia, sources said.

Micron also said it will supply HBM3E to Nvidia.

(Reporting by Fanny Potkin in Singapore and Heekyong Yang in Seoul; Editing by Edwina Gibbs)