Nvidia CEO Jensen Huang Says AI Memory Needs Are Rising During Taiwan Trip, Backs TSMC's Global Expansion, Dismisses China H200 Rumors

Core Insights - Nvidia's CEO Jensen Huang emphasized that the increasing demand for AI is driving a significant rise in the need for advanced memory solutions [1][2] Group 1: AI Demand and Memory Needs - The future of artificial intelligence will be influenced by memory requirements as much as by computing power, with modern AI models necessitating high-speed processing and increased memory capacity [2] - Nvidia is collaborating with all major high-bandwidth memory suppliers, including SK Hynix, Samsung Electronics, and Micron Technology, to meet the soaring demand for memory this year [3] Group 2: Semiconductor Manufacturing and Partnerships - Huang refuted claims that 40% of Taiwan's semiconductor manufacturing capacity has been transferred to the U.S., asserting that global chip production is expanding with new capacities being added in various regions while Taiwan remains a crucial manufacturing hub [4] - Taiwan Semiconductor Manufacturing Co. (TSMC) is regarded as Nvidia's irreplaceable foundry partner, with Huang highlighting its technology leadership and the need for TSMC to scale capacity significantly over the next decade [5] Group 3: Regulatory and Geopolitical Context - Huang dismissed rumors regarding the approval of Nvidia's H200 AI chips in China, stating that no orders have been placed and final clearance is still pending [6] - The H200 chip has become a focal point in U.S.-China tech tensions, with Washington approving shipments while Beijing has yet to fully clear imports [8]