SK Group Chairman Chey Tae-won and NVIDIA CEO Jensen Huang are holding their second 'Kkanbu meeting'. This isn't just a friendly dinner; it's a pivotal moment for reshaping the global AI supply chain, centered on two key pillars: HBM4 memory and Physical AI.
The immediate backdrop is NVIDIA's preparation for its next-generation AI accelerator, the 'Rubin' platform. To power Rubin, a stable supply of cutting-edge HBM4 memory is essential. Recently, Jensen Huang confirmed that all three major memory suppliers—SK hynix, Samsung, and Micron—have been certified to produce HBM4. This move secures NVIDIA's supply chain by avoiding reliance on a single vendor, and it firmly places South Korea at the heart of its strategy.
So, what led to this strategic meeting? First, SK hynix has been sending strong signals of its commitment. Chairman Chey announced ambitious plans to double wafer capacity within five years and increase capital expenditures, along with pursuing an ADR listing in New York to fund this massive expansion. This proactive stance showed NVIDIA that SK hynix is ready to meet its immense demand.
Second, the collaboration is expanding beyond just hardware. SK Telecom recently unveiled a digital twin of an SK hynix factory, built on NVIDIA's Omniverse platform. This 'Physical AI' initiative aims to optimize the complex semiconductor manufacturing process, potentially improving HBM production yields and speed. This transforms the relationship from a simple supplier-buyer dynamic to a deep technological partnership.
Finally, the geopolitical landscape plays a significant role. The U.S. government has been tightening regulations to prevent advanced AI chips from reaching China. This creates uncertainty for NVIDIA's China-related business, making it more critical to secure reliable, high-volume production in 'approved' markets like South Korea. The meeting with SK's top leadership is a direct move to solidify this crucial, non-China-centric supply line.
In short, this meeting is about aligning strategies for the next era of AI. It's where SK's production capacity and technological innovation meet NVIDIA's roadmap, securing a stable foundation for the AI revolution to come.
- HBM (High Bandwidth Memory): A type of high-performance memory that stacks DRAM chips vertically to achieve faster data transfer speeds, essential for AI accelerators.
- Digital Twin: A virtual model of a physical object or process, used to run simulations and optimize performance before changes are made in the real world.
- ADR (American Depositary Receipt): A certificate issued by a U.S. bank representing a specified number of shares in a foreign company's stock, allowing it to be traded on U.S. exchanges.