Jensen Huang knows how to work a dinner table. The Nvidia CEO gathered top South Korean tech executives at a traditional Taiwanese restaurant in Taipei on June 1, hosting what’s being called the first-ever “Korean Partner Night.”
The guest list read like a who’s-who of Korea’s tech elite: SK Hynix CEO Kwak Noh-Jung, plus executives from Samsung Electronics, LG Electronics, and Naver. The setting was Computex, the annual trade show that has become ground zero for AI hardware announcements.
Why Korean chipmakers matter to Nvidia’s AI ambitions
Nvidia designs the GPUs that power everything from ChatGPT to autonomous vehicles, but it doesn’t manufacture the memory chips those GPUs depend on. That critical component, high-bandwidth memory (HBM), comes primarily from two sources: SK Hynix and Samsung.
SK Hynix has been the dominant force in HBM production, essentially becoming Nvidia’s preferred partner for the memory that goes into its data center GPUs. Samsung has been working to close the gap, and both companies are racing to develop next-generation memory chips that can keep pace with Nvidia’s increasingly power-hungry AI accelerators.
