SK Group’s chairman has confirmed that its chipmaking affiliate, SK Hynix, will be bringing forward production of its high-bandwidth memory 4 (HBM4) chip by six months. In a statement published earlier this morning, Chey Tae-won said that the decision was made after a special request from Nvidia. Though Chey was not specific about the content of any communications between SK hynix and Nvidia, he did allude to the launch of a future GPU product as the motivating factor behind the request.
“Every time Nvidia launches new GPUs, they ask us for more HBMs and push us to accelerate production,” said the SK Group chairman. “Although I don’t make those decisions alone, I consulted SK hynix CEO Kwak Noh-jung, who committed to trying to meet the demand.”
Nvidia’s special SK hynix request
Founded in 1983 as a sub-division of Hyundai Electronics, SK hynix is now one of the world’s pre-eminent suppliers of memory chips. These components are also in short supply amid a general surge in interest among thousands of companies worldwide in AI applications, leading other major players in the chip assembly market to take steps to secure their own supply chains.
Nvidia is no different, incorporating SK hynix memory chips in their own GPU product lines. Last month, the former announced plans to begin supplying HBM4 chips to its customers in the latter half of 2025. The sheer size of Nvidia’s custom,, however, means that SK hynix is expected to adjust its production timelines to accommodate the GPU producer’s request, which could involve increasing production capabilities and reallocating resources.
SK hynix plans to deliver its 12-layer HBM3E chips to an undisclosed customer within this year. Additionally, CEO Kwak Noh-Jung stated at the SK AI Summit 2024 in Seoul that the company intends to ship samples of its 16-layer HBM3E chips in early 2025.
Following these developments, shares of SK hynix rose by 5.1%. The company also signed a non-binding preliminary memorandum of terms (PMT) with the US Department of Commerce (DOC) to secure up to $450m in federal incentives to set up a chip packaging facility in Indiana.
Growing demand for Nvidia’s GPUs amid AI boom
This development coincides with an unprecedented surge in demand for Nvidia’s GPUs, driven by growth in generative AI and deep learning models. AI companies and technology firms are actively acquiring Nvidia’s A100 and H100 GPUs to power new AI models.
Last month, Nvidia’s shares achieved a record high, bringing the chipmaker closer to overtaking Apple as the world’s most valuable company. Nvidia’s stock around mid-October, climbed 2.4%, closing at $138.07, as investor confidence grew due to strong demand for its AI processors. In June, Nvidia briefly held the title of the world’s most valuable company before being surpassed by Microsoft. For several months, the market valuations of Nvidia, Microsoft, and Apple have been closely aligned and competitive.