是说芯语·2025-09-11 05:21Core Viewpoint - The article discusses the significant changes in the North American AI hardware supply chain, highlighting four major expectation gaps due to rapid application growth and competition among tech giants [4]. Group 1: Expectation Gaps - Expectation Gap 1: A "computing power squeeze" has emerged due to the rapid explosion of applications, leading to shortages in critical materials such as advanced process capacity from TSMC, high-end PCBs, and optical modules, exceeding expectations [5][6]. - Expectation Gap 2: The competition among major players like OpenAI, Google, Meta, and Amazon for computing resources is intensifying, with companies scrambling to secure TPU and GPU resources [10][11]. - Expectation Gap 3: Product upgrades are expected to significantly enhance profit margins, particularly for high-end products like PCBs and optical modules, which are in high demand [13]. - Expectation Gap 4: The fierce competition among North American tech giants is expected to lead to price increases, as companies prioritize securing production capacity over reducing prices [14]. Group 2: Industry Dynamics - The supply chain's position is improving, which is a necessary condition for valuation increases, as major players like Google, Oracle, AWS, and OpenAI seek suppliers from the current mainstream companies in the supply chain [12]. - The competition is expected to drive up profit margins due to the scarcity of high-end products and the aggressive resource acquisition strategies of major clients [16]. - The technological uncertainty is anticipated to decrease amid intense competition, delaying significant advancements in areas like CPO until after the current competitive cycle [17]. Group 3: Market Outlook - The article suggests that the current innovation cycle favors domestic supply chains, with a bullish market outlook for AI leading to a natural increase in valuations [17]. - The expectation is that new capital influx will drive marginal pricing, with potential for significant growth in valuations for high-demand products [16].