TL;DR: NVIDIA CEO Jensen Huang met with Samsung and Hyundai executives at the APEC summit, discussing key partnerships amid Samsung's HBM3 memory challenges. NVIDIA plans to use Samsung's advanced HBM4 and HBM4E memory in upcoming AI GPUs, strengthening its semiconductor supply chain alongside TSMC and Intel for future AI innovations.
NVIDIA CEO Jensen Huang has been enjoying his time in South Korea, enjoying some Korean fried chicken with his "executive friends" from Samsung and Hyundai.
Jensen delivered an important keynote at the APEC summit, where he met with Samsung chairman Lee Jae-yong, and the president of Hyundai Motors, Chung Eui-sun. After the important stuff was over, the executives were spotted in casual clothing, enjoying some Korean fried chicken, what looks like some mozzarella sticks, and some beer at a South Korean restaurant.
This year's APEC summit was an important one, with President Trump meeting with Chinese president Xi Jinping, which is the reason Jensen was in South Korea, in support of the US. After the business side of things, the NVIDIA CEO was hanging out in South Korea talking to the public, and enjoying his dinner with his executive friends.
We're sure that Jensen would've been enjoying some fresh discussions with Samsung, as the South Korean memory giant has been having issues with its HBM3 memory, but HBM3E and next-gen HBM4 memory from Samsung will be used inside of NVIDIA AI GPUs.
Samsung Foundry has also recently been included in NVIDIA's coveted NVLink ecosystem, alongside Intel, with rumors of a "huge partnership" between NVIDIA and Samsung being discussed right now. NVIDIA's push into AI GPU dominance means it needs the best semiconductor foundries in the world -- TSMC for now, but Samsung and Intel are options -- as well as the best DRAM chips in the market.
HBM4 and HBM4E will be used on NVIDIA's next-gen Rubin and beefed-up Rubin Ultra, while it will be moving to future-gen HBM5 memory for its future-gen Feynman GPU family.