In Interview with AP, Nvidia's Jensen Huang Urges Americans to Use AI More - Broadband Breakfast

In an interview with the Associated Press, NVIDIA CEO Jensen Huang emphasized that society must adapt as the era of artificial intelligence (AI) unfolds. While expressing optimism about AI's transform... Read original →

Huang’s remarks are a veiled warning about America’s AI infrastructure bottleneck: compute scaling is hitting hard limits in power availability and regulatory inertia. Technically, surging demand for inference chips will accelerate adoption of chiplet-based designs and memory-compute integration, making TSMC’s CoWoS capacity a de facto strategic asset. On compliance, tighter U.S. export controls would force NVIDIA to diversify assembly/test operations to India and Vietnam, raising operational costs by 15–20%. Competitively, AMD and Intel are exploiting this moment—pushing MI300 and Gaudi3 with open software stacks to erode CUDA lock-in. Over the next 18 months, the U.S.-China AI race will pivot from raw performance to watts-per-trillion-operations; energy efficiency, not just transistor count, will dictate who sets the price of next-gen AI infrastructure. Huang’s real message to Washington: while containing China, don’t first exhaust your own grid—and time.