Record profit is a rear-view mirror.SK海力士创纪录的利润和韩股的暴涨,被市场读成了『AI需求可持续』的证据——但这是在用一个回看的指标,回答一个前瞻的问题。

A record quarter and a soaring stock are being read as proof that AI memory demand is durable. On October 28 SK Hynix reported third-quarter operating profit up 631% to 83.3 trillion won (about US$62.7 billion) and a shareholder-return program worth over 100 trillion won (about US$75.3 billion). The KOSPI closed up 6.63% at a record 6,012, with SK Hynix up 11% and Samsung up 9%; in US pre-market trading SK Hynix's ADR rose more than 8%. Analysts called it proof the AI rally has not peaked.
But the signals are mismatched. Record profit measures memory already shipped last quarter: it is lagging. A surging stock measures sentiment. The real question is forward-looking: will the hyperscalers' capex still be there in 2027. Using realized earnings and a share price to answer that confuses one indicator for another.
Memory is the most cyclical link in tech. Capacity expands with a lag and prices swing with the supply gap, so record profits have historically clustered near cycle tops, not the safe middle, because that is when everyone extrapolates the realized into the sustainable.
The strongest counter: HBM is customized, locked into long-term contracts, and supply-constrained, so this cycle could run longer than past spot cycles. Maybe. But confirming it means watching HBM contract volumes and buyers' capex guidance, not realized profit.
We'd revise this if forward orders and 2027 capex guidance keep rising while supply stays tight.
So change the question: not how high are profits and the stock, but whether forward orders and buyer capex are rising or falling.
Research, not investment advice.
10月28日,SK海力士三季度营业利润同比飙升631%至83.3万亿韩元(约627亿美元),并抛出逾100万亿韩元(约753亿美元)的股东回报计划;KOSPI收盘涨6.63%、创6,012点新高,SK海力士涨11%、三星涨9%,美股盘前SK海力士ADR涨超8%。分析师的解读是:这缓解了对AI行情见顶的担忧。
问题在于指标错配。创纪录的利润衡量的是上个季度『已经卖出去的内存』,是滞后的;股价暴涨衡量的是情绪和预期。而真正要回答的是前瞻问题:2027年云厂商的资本开支还撑不撑得住。用已实现的业绩和股价去证明未来的需求,是把两种不同的指标混为一谈。
内存是科技链里最具周期性的一环:产能扩张滞后、价格随供需缺口剧烈波动,创纪录的利润在历史上往往出现在周期高点附近,而不是安全的中段,因为那正是人人把『已实现』外推成『可持续』的时刻。
二阶效应:当『利润创纪录=需求可持续』成为共识,资本以更高估值涌入扩产;一旦2027资本开支指引下调、或HBM长协放缓,已扩的产能叠加追高的杠杆头寸,下行会比上行更快。
最强反方:HBM是定制化、长协锁定、产能紧张,这一轮可能比过去的现货周期更长更稳。但要确认这一点,该看的是HBM长协量和买方的资本开支指引,而不是已实现的利润。
若前瞻订单和2027资本开支指引在产能偏紧时继续走高,这个判断就该修正。
所以把问题换一下:不是『利润和股价有多高』,而是『前瞻订单和买方资本开支是在升还是在降』。
研究,不构成投资建议。
Built on the WAKENEX concept:所用的 WAKENEX 概念: Match the feedback to the question反馈分工:不要用一种指标替代其他问题的答案
Sources来源
Add WAKENEX as a preferred source on Google在 Google 把 WAKENEX 设为首选来源


