When Nvidia CEO Jensen Huang arrived in South Korea, global attention turned to a viral "chimaek" (fried chicken and beer) dinner on October 30 featuring Samsung Chairman Lee Jae-yong and Hyundai Executive Chair Chung Eui-sun. Yet, the spotlight shifted the following day to SK Group—and particularly its memory arm, SK Hynix, which co-headlined a landmark AI infrastructure deal that analysts have called a strategic win.
The question hanging over the spontaneous Gangnam gathering was straightforward: why was SK Hynix, Nvidia's largest supplier of high-bandwidth memory (HBM) for AI chips, conspicuously absent?
Strategic sequencing: why SK missed the photo op
The absence of SK Group Chairman Chey Tae-won from the headline-grabbing chimaek summit was less a diplomatic slight than a matter of logistics and timing.
Sources familiar with the matter said the dinner was never meant to be a formal "Korea Inc." meeting. Huang himself described it as simply wanting to "eat chimaek with friends," not hold a press conference, and the informal venue could accommodate only a small circle.
Moreover, Chairman Chey was in final preparations for the APEC CEO Summit in Gyeongju, some 250 kilometers away, on the same evening of October 30. Travel constraints made attendance impractical. SK Group later confirmed that Chey had prioritized summit obligations and already had a separate, formal meeting with Huang scheduled for October 31.
In the end, while Samsung and Hyundai enjoyed what many in Korea dubbed "cultural theater"—a photo-friendly display of camaraderie—SK hynix secured the real substance. As one online comment succinctly put it, "SK hynix is in the GPU, not the photo. They're fine."
The next day's focus: HBM and the 50,000-GPU AI factory
While others were sharing chicken, SK hynix was securing its role at the heart of Korea's AI future. At the APEC Summit on October 31, Nvidia announced an expanded partnership with SK Group to build an AI factory dedicated to semiconductor research, development, and production.
The facility will deploy more than 50,000 Nvidia GPUs and is expected to become one of Korea's largest AI factories when completed in late 2027.
At its core, the collaboration cements SK Hynix's position in the advanced AI platform supply chain:
HBM advancements: Joint efforts will focus on developing next-generation SK Hynix high-bandwidth memory (HBM) and other advanced memory solutions for Nvidia GPUs.
AI-driven chip design: Using Nvidia's CUDA-X technologies and the PhysicsNeMo framework, SK Hynix will apply AI-based physics simulation to accelerate chip design and verification, enabling faster, more precise product development.
Digital transformation: The AI factory will support SK subsidiaries such as SK Hynix and SK Telecom. SK Hynix is building autonomous fab digital twins via Nvidia's Omniverse platform to simulate and optimize operations in real time, advancing toward self-optimizing manufacturing.
Workforce productivity: AI-powered agents built with Nvidia NIM microservices will help enhance productivity across SK Hynix's 40,000-plus employees, from R&D engineers to fab operators.
Huang called SK Group a "vital memory technology partner," enabling the creation of the world's most advanced GPU computing platforms.
Chairman Chey underscored the shared vision, saying the Nvidia AI factory will form the foundation to "build the infrastructure that powers the next generation of memory, robotics, digital twins, and intelligent AI agents."
Article edited by Jack Wu
