Morning Bid: Nvidia chief to Asia: ‘We’re still supply constrained’ - WHTC

NVIDIA CEO Jensen Huang stated at a conference in Taipei that while the company has sufficient supply to support robust growth linked to the AI boom, constraints still persist. This comes shortly afte... Read original →

Jensen Huang’s remarks in Taipei, China expose structural bottlenecks beneath AI chip euphoria: despite NVIDIA’s lead in 3nm EUV processes, its Hopper and Blackwell platforms remain constrained by TSMC’s advanced packaging capacity. This directly delays server deployments and forces cloud providers to recalibrate AI capex. Geopolitical risk has shifted from a latent variable to a tangible cost—Middle East tensions disrupt logistics and energy pricing, while U.S. export controls compel multinationals to build redundant, higher-cost supply chains. AMD and Intel will accelerate MI300X and Gaudi 3 customer adoption, targeting Europe and Northeast Asia where supply reliability trumps peak performance. Over the next 18 months, the industry enters an era of asymmetric capacity expansion: NVIDIA locks in TSMC CoWoS output as a moat, while rivals pivot to chiplet-based heterogenous integration to bypass leading-edge node shortages. The true long-tail impact? AI infrastructure is becoming less about transistors and more about technological sovereignty.