中美大反转，中国AI调用量首超美国，A股嗨了，多板块掀涨停潮！华尔街知名分析师：中国算力路径颠覆传统认知

Group 1 - Nvidia's stock dropped 5.5% after reporting record earnings, resulting in a market value loss of nearly $260 billion (approximately 1.77 trillion RMB) [1][5] - In contrast, A-shares and Hong Kong stocks related to computing power leasing, cloud computing, and electricity surged, with notable gains including 20% for CloudWalk Technology and 19.91% for Jiawei New Energy [1][6][11] - The divergence in market performance highlights a significant shift in the perception of computing power value distribution [3][4] Group 2 - For the first time, China's AI model usage surpassed that of the U.S., with a weekly call volume of 4.12 trillion tokens compared to the U.S.'s 2.94 trillion tokens [2][20] - In the top five global models, four are from China, indicating a "cluster rise" in Chinese AI models [2][24] - The rapid growth in Chinese model usage is attributed to their efficient architecture, which reduces reliance on high-end GPUs and leads to exponential growth in domestic computing power demand [4][29] Group 3 - The market is reassessing the sustainability of AI capital expenditures, moving beyond Nvidia's short-term performance to long-term concerns about growth ceilings [5][27] - The "Mixture-of-Experts" (MoE) architecture used by Chinese models significantly lowers inference costs and reduces the need for high-end GPUs, with a 60% reduction in memory usage and a 19-fold increase in throughput [27][28] - The cost of processing tokens with Chinese models is substantially lower than that of their U.S. counterparts, with prices as low as $0.3 per million tokens compared to $5 for foreign models [29][30] Group 4 - The demand for domestic computing power infrastructure is surging due to the global acceptance of Chinese AI models, which are recognized for their performance and cost competitiveness [29][30] - Analysts predict a compound annual growth rate of 330% in China's token consumption from 2025 to 2030, leading to a 370-fold increase in just five years [29][30] - The combination of cost advantages and technological capabilities is driving exponential growth in domestic computing power demand [29][30]