Jensen Huang shouts 'Everyone loves HBM' at Seoul barbecue as Nvidia announces Korea R&D center and previews four new products

Jensen Huang’s Seoul announcement of a Korean R&D center is a strategic move to lock in HBM supply chain alignment. Technically, it accelerates co-optimization between CoWoS and HBM4, pressuring Samsung and SK Hynix to fast-track 2.5D/3D stacking and compelling TSMC to expand Korean advanced packaging capacity. On compliance, tighter U.S. export controls on advanced packaging tools make localized R&D a hedge against disruption—but raise IP protection and talent competition costs. In response, AMD may deepen ties with Japan’s Rapidus to advance chiplet alternatives, while Intel could leverage EU subsidies to push Foveros as an HBM bypass. Over the next 18 months, HBM will become the epicenter of AI semiconductor geopolitics, elevating Korean memory giants’ influence yet exposing systemic overreliance on U.S.-centric architectures.