The market spent this week redrawing the line between what AI can do and what AI can earn.
SpaceX posted $4.4 billion in net income, its superintelligence segment growing 830% to $7.6 billion and becoming its largest business; the stock rose after hours. Netflix beat on earnings and revenue, then guided below expectations on ads and cut its full-year target from $3.0 billion to $2.7 billion; the stock fell after hours. Both did well. The split verdict is about slope, not size: one is being repriced on acceleration, the other on deceleration.
There's a macro layer underneath. Ten-year Treasury yields sit near 5%. AI infrastructure is long-duration and capital-hungry. When the risk-free rate resets upward, the discount rate resets with it, and the market stops paying for stories that can't yet show free cash flow.
My working assumption: the AI demand thesis isn't broken. What broke is the assumption that cheap financing was permanent. What survives this repricing isn't the loudest narrative—it's whoever can turn capex into cash flow while rates stay high.
今晚是这轮财报周最密的一晚:特斯拉和 Meta 盘后交卷,前面还有 IBM、泛林、TI。上周科技股刚被融资成本敲了一记,这周就是最考验成色的时候。
我不提前押方向,就好奇一件事:特斯拉和 Meta 的电话会,如果只能认真听一个问题,你们最想听管理层答清楚哪件事?Cybercab 什么时候出得州、HW3 老车主怎么办、Meta 明年资本开支到底多少——评论区聊聊,我今晚复盘。