Core Insights - Elon Musk agrees with TSMC's concerns about chip oversupply, predicting that the limiting factor for the AI industry will shift from chip manufacturing to the ability to power and operate these chips, with key bottlenecks in power supply, transformer configuration, and cooling system deployment [1][3] Group 1: AI Infrastructure Challenges - Musk emphasizes that deploying AI chips involves more than just transporting GPUs to power plants; it requires addressing three core issues: gigawatt-level power supply, high-voltage power conversion, and efficient cooling systems [2] - The data center industry is undergoing a critical transition from air cooling to liquid cooling, which carries significant risks, such as potential losses of up to $1 billion if cooling systems fail [2] Group 2: Power Supply vs. Chip Production - Musk predicts that by Q3 2026, the core bottleneck will transition from chip manufacturing to the capability to operate these chips, as AI chip production is expanding exponentially while the supporting power infrastructure is only growing linearly [3] - The disparity between the rapid increase in chip output and the slow growth of power supply means that many high-performance AI chips may remain unused due to inadequate power conversion and cooling systems [3]
台积电担忧芯片过剩？马斯克：他们是对的，电力液冷都跟不上
Hua Er Jie Jian Wen·2026-01-07 08:29