Baek Seung-ho

NVIDIA's competitive advantage in physical AI comes from its software ecosystem (CUDA, Omniverse), not its hardware. This software moat creates high switching costs and network effects, ensuring continued dominance as physical AI scales. Companies will build on Omniverse as the standard simulation platform, driving GPU demand and reinforcing NVIDIA's position.

As physical AI generates 100x more data than current AI, memory (DRAM, NAND) becomes even more critical than GPU. SK Hynix, a leader in HBM and NAND, is well-positioned to benefit from the transition to HBF and the explosive growth in memory demand driven by physical AI. Memory will be a bottleneck and thus a growth area.