7月以来美股半导体板块经历大幅波动，反映出市场对AI资产定价方式的根本性转变[1]。"闭眼买AI"的投资阶段已经过去，市场心态从无条件看好转向要求更多商业回报证据[1]。
虽然AI需求仍然保持强劲，但成长股的核心交易逻辑已发生改变，关键在于是否能带来"比预期更好"的表现[1]。此前云厂商资本支出（Capex）的上修曾是市场利好，如今却成为需要解答的问题——投资者开始要求看到这些AI投入转化为实际商业回报的证据[1]。为适应市场变化，投资者需要重新分类持仓资产，将AI半导体相关投资划分为四个类型：收税型资产、卖铲型龙头、周期弹性资产和纯叙事资产[1]。
The semiconductor sector in U.S. markets has experienced significant fluctuations since July, prompting investors to fundamentally reconsider how artificial intelligence assets should be valued [1]. The era of indiscriminate enthusiasm for AI investments has largely concluded, as market sentiment has shifted from unconditional optimism toward demanding concrete evidence of commercial returns [1].
While demand for AI technology remains robust, the dynamics governing growth stock valuations have changed materially [1]. Investors now require proof that performance exceeds expectations rather than simply meeting baseline forecasts. This shift has transformed cloud infrastructure providers' increased capital expenditures from a positive signal into a critical question mark [1]. The market is demanding visibility into tangible business returns generated by these substantial AI infrastructure investments [1].
To navigate this changed landscape, market observers have categorized AI semiconductor-related holdings into four distinct asset classes: revenue-generating assets positioned to capture value, leading suppliers selling essential infrastructure, cyclical positions with earnings elasticity, and purely narrative-driven plays [1]. This classification framework reflects the market's transition toward a more differentiated and critical assessment of the diverse competitive positions and business models within the artificial intelligence ecosystem [1].
