n Friday, Samsung Electronics announced a slower sale of their artificial intelligence chips due to US export restrictions when it comes to China, also
triggering a slower launch of their improved version of the high-end chips.
Even more so, the advanced chips that are used for AI, have been the go-to when it comes to weak memory from the chip market. Yet, theri competitors such as SK Hynix, are Nvidia’s main supplier of high-bandwidth memory (HBM) chips that are used in AI graphics processing units.
In December, a fresh crackdown in the semiconductor industry from China restricted HBM chip sales. Samsung was the company that was expected to be hit harder than its peers due to its reliance on Chinese customers for nearly 20% of HBM sales.
Kim Jae-June, the executive vice president of Samsung’s memory business, stated on Friday that “There will be some temporary restrictions in our HBM chip sales in the first quarter,".
He also added, “We expect some impact on our HBM demand from not just U.S. restrictions on exports of high-end chips but also a shift in demand for improved chips by major clients”.
Subscribe to our newsletter
Samsung also said that it planned to launch an improved version of the HBM3E this March, coming with an 8-layer and 12-layer. Jensen Huang, Nvidia CEO, also said this month that Samsung has to “engineer a new design” in order to supply HBM chips to his company, reported Reuters.
Back in October, Samsung reported that it was making progress in supplying HBM chips to Nvidia, yet, no further information was released to the public.
By
Adam Brown
•
January 31, 2025 10:10 AM
.webp)
