盈利表现并非纯投机性：期初财报显示，盈余每股较一年前成长约 83%，由持续需求与结构性供应约束推动，后者抬高价格与利润。这股动能已从 Nvidia 扩散至较平凡的晶片领域。然则在半导体产业，超高利润率往往吸引更多产能进场，增加过剩风险。净利润率目前约为 44%，而历史周期显示利润率会大幅起伏：金融危机后曾扩张、后续降温；Covid 初期再度飙升，但 2022 年在需求回补完成与库存累积后又回落。
文章认为投资人可能高估了现阶段的领跑者。高盛的 Jim Covello 最近建议较偏好 hyperscalers 而非晶片制造商。Microsoft、Oracle 与 Meta 仍低于近期高点，部分因市场担心 AI 支出尚未转化为充分报酬。若其 AI 投资最终带来正向现金流，hyperscalers 仍有机会跑输过去的半导体股；若他们削减支出，短期或有利于自由现金流与股价，但对晶片股将形成利空。第二、第三梯队的半导体公司要继续上涨，需求必须持续异常且供应限制仍在。然而极端需求终将被新的产能吸收，主要回调风险因此并非可忽视。
Semiconductors have outpaced almost everything in equities, including the hyperscalers that led AI enthusiasm from 2022–2025. The S&P Semiconductors Select Industry Index (equal-weighted) rose 144% over 12 months, versus 19% for the equal-weighted S&P 500 and 36% for a hyperscaler basket, its biggest outperformance in the AI era. Since the ChatGPT launch period, semiconductors are positioned to overtake hyperscalers as the largest AI beneficiaries. In 2026, the S&P 500 was up about 6%, but would have been only about 2.2% without semis, implying semis contributed roughly 3.8 percentage points.
Key performance is not purely speculative: early-season results show EPS growth running around 83% from a year earlier, supported by durable demand and tight structural supply that is lifting prices and profits. This strength has spread beyond Nvidia to broader, less glamorous chip categories. Yet ultra-high margins in semiconductors historically attract fresh capacity, increasing overhang risk. Net margins are now around 44%, and prior cycles show margins can swing sharply: they expanded after the financial crisis then cooled, surged during early Covid but narrowed in 2022 when demand was pulled forward and inventories swelled.
The article argues investors may be overpaying for today’s leaders. Jim Covello of Goldman Sachs recently preferred hyperscalers over chipmakers. Microsoft, Oracle and Meta are below recent highs, partly due to worries that AI spending has not yet produced returns. If their spending generates cash-flow gains, hyperscalers could still outrun semiconductors; if they cut spending, shares may gain short-term from higher free cash flow, but chipmakers could lose momentum. For second- and third-tier chip firms to continue their run, demand must stay extreme and supply constraints must persist. But high demand can eventually be met by new capacity, and that makes a major pullback increasingly plausible.