Nvidia courts Korea's industrial giants ahead of COMPUTEX - digitimes

NVIDIA CEO Jensen Huang's pre-Computex meetings in Taipei are drawing significant attention from South Korean companies aiming to secure a larger role in the global AI supply chain. This development r... Read original →

Jensen Huang’s pre-COMPUTEX outreach to South Korean conglomerates signals a strategic recalibration driven by AI chip manufacturing bottlenecks. With sub-3nm nodes reliant on EUV and TSMC as NVIDIA’s sole high-volume foundry for H200/B100 GPUs, Korean firms face an existential push beyond memory dominance into advanced logic and co-packaged optics. This triggers a technical cascade: Samsung may accelerate GAA and backside power delivery to compete for next-gen AI contracts. Yet, under U.S.-China tech decoupling, deeper integration involving Taiwan, China’s advanced capacity invites export control scrutiny. Over the next 12–24 months, regionalized redundancy—U.S., Korea, and Japan each building backup fabs—will emerge, but yield and cost gaps will widen against TSMC’s scale. Ultimately, geopolitical alignment, not just transistor density, will dictate who controls the AI compute stack.