Nvidia CEO Jensen Huang says China should not have Blackwell or Rubin AI GPUs

At the Milken Institute Global Conference in Los Angeles, NVIDIA CEO Jensen Huang stated that China should not have access to the company's latest AI GPU technologies, including the Blackwell and Rubi... Read original →

Huang’s stance reveals a strategic pivot toward enforcing U.S. AI supremacy, not mere compliance. Blocking Blackwell and Rubin access forces Chinese AI developers onto outdated Hopper or Ampere architectures, slashing training efficiency by over 30% and delaying domestic AI infrastructure upgrades. TSMC’s N4/N5 capacity is prioritized for U.S. clients, rendering H200 shipments to China effectively void; the upcoming 3nm EUV-based Rubin will widen the compute gap further. AMD’s MI308X lacks the software stack to fill NVIDIA’s void. Within 12 months, China will accelerate Chiplet and in-memory computing to bypass advanced-node restrictions. Paradoxically, NVIDIA’s zero market share in China erodes its global pricing power—excluding the world’s largest AI growth market risks premature saturation of its high-end GPU ROI curve. U.S. export controls are shifting from targeted sanctions to systemic decoupling, but techno-nationalism ultimately undermines the innovation feedback loop.