(Photo : NVIDIA Newsroom)
Nvidia's Urgent Call for Advanced HBM4 Chip Supply

• Nvidia CEO Jensen Huang has requested SK hynix to expedite the delivery of its next-generation high bandwidth memory (HBM), named HBM4.
• The request comes amid increasing demand for Nvidia's AI accelerators, which heavily rely on HBM chips for data and energy-intensive tasks.
• SK hynix, a key supplier of HBM chips to Nvidia, has initiated mass production of its advanced 12-layer HBM3E chips.
• The recent request by Nvidia's CEO underscores the increasing demand for AI accelerators and the critical role of HBM chips in meeting this demand.

In a recent development that underscores the escalating demand for artificial intelligence (AI) computing chips, Nvidia CEO Jensen Huang has requested SK hynix, a leading South Korean semiconductor manufacturer, to expedite the delivery of its next-generation high bandwidth memory (HBM), named HBM4, by six months. This information was disclosed by SK Group Chairman Chey Tae-won during his keynote speech at the SK AI Summit 2024 in Seoul.

Huang's request comes in the wake of a burgeoning demand for Nvidia's AI accelerators, which heavily rely on HBM chips for data and energy-intensive tasks as AI technology advances beyond text-based models. The request to accelerate the supply of HBM4 chips is a testament to the increasing reliance of AI accelerators on HBM chips for data and energy-intensive tasks.

SK hynix, a key supplier of HBM chips to Nvidia, has been at the forefront of HBM technology. The company made history in March when it became the first in the world to supply fifth-generation eight-layer HBM3E chips to Nvidia.

SK Hynix's Role in Nvidia's Operations

It has also recently initiated mass production of its advanced 12-layer HBM3E chips. The company had previously announced plans to supply the upcoming HBM4 chips in the second half of next year, but Huang's request may see this timeline brought forward.

During a session of the SK AI Summit, Huang emphasized the importance of the partnership with SK hynix in a video message. He described the latter's development plan as super aggressive but super necessary. This statement underscores the critical role that SK hynix plays in Nvidia's operations and the broader AI chip industry.

In response to the rising demand for higher-performance HBM, SK hynix CEO Kwak Noh-jung announced the development of 16-layer HBM3E chips, with plans to produce it in the first half of next year. This development is indicative of the rapid pace of innovation in the semiconductor industry and the increasing demand for high-performance memory solutions.

Building the Supply Chain of the Global AI Chip Industry

In his keynote, Chey highlighted the trilateral partnership among SK hynix, Nvidia, and Taiwan Semiconductor Manufacturing Co. (TSMC) in building the supply chain of the global AI chip industry. He explained that as Nvidia develops more advanced GPUs, it requires more HBM chips. This necessitates SK hynix to work hard to keep up. The development of advanced HBM products and maintaining high yields is challenging, necessitating closer cooperation with Nvidia.

Chey also emphasized the importance of a close partnership with TSMC, the world's largest foundry, or semiconductor contract manufacturing company. This partnership is crucial in ensuring a robust supply chain for the global AI chip industry.

The SK AI Summit 2024 also saw participation from other industry leaders. Microsoft CEO Satya Nadella and TSMC CEO C.C. Wei delivered video messages, while Huang participated in an online conversation with Professor David Patterson from UC Berkeley.