Nvidia CEO Jensen Huang recently appeared on the popular Lex Fridman Podcast for a deep, over two-hour discussion covering AI scaling laws, compute and power bottlenecks, AI factories, and the future impact of AI on society. Among the topics drawing strong attention from Taiwan’s market was Huang’s revisit of his 2013 decision to decline an offer from TSMC founder Morris Chang to become CEO. He further analyzed the core factors enabling TSMC’s sustained leadership in the semiconductor industry.

Sources in the passive component supply chain report that Nvidia’s next-generation platform architecture, Vera Rubin, is scheduled to enter mass production in the third quarter of 2026 and has drawn significant attention from the global AI computing market. However, the design of the GPU compute tray has yet to be finalized, which could cause slight delays in actual shipment timelines.

While the barriers to entry in cloud AI ASICs remain out of reach for many chip designers, demand for specialized chips at the edge is rising sharply. Across a widening range of applications, customers are seeking greater control at the silicon level, aiming to reduce their reliance on traditional chip giants.

Taiwan foundry revenue growth expected to reach 30% 2026; strategic adjustments initiated for mature processes.