The real risk this week isn't the earnings miss.10 月 26 日美股低开，所有人盯着周三微软、Alphabet、周四亚马逊的财报，却问错了问题。真正该问的不是这几家会不会超预期，而是这轮 AI 基建被放大的下行，最后由谁来扛。

It's who bears the amplified downside of an AI build-out financed by debt and circular deals. Start with scale: hyperscalers' 2026 AI capital spending runs to roughly $800 billion, increasingly funded by debt and vendor financing rather than profit.

Then the structure, which is turning into a loop: I buy your chips, you take a stake in me, a third party underwrites my compute. Nvidia has lined up more than $500 billion in capital pools, with partners like Apollo, BlackRock and Blackstone, to finance its own customers; OpenAI signed a roughly $300 billion, five-year cloud commitment with Oracle. In the same week, Meta fell more than 9% after its first-ever negative free cash flow, and the Nasdaq had its biggest one-day drop since June.

This is leverage cutting both ways. The firms holding the lever ride up weightlessly. But if compute demand disappoints, the amplified loss doesn't land on them. It lands on bondholders, on the chipmakers' own balance sheets, and on retail that bought in through leveraged ETFs. An earnings beat proves today's demand; it doesn't answer who owns the risk. Two different questions, and you can't use one to answer the other.

The strongest case against me: this is real, multi-year demand, cloud revenue still grows double digits, the circular deals are a small slice of total spending, and most capex still comes from operating cash flow. We'd revise this if, by 2027, most of these commitments convert into booked revenue without more borrowing. Until then, be clear who carries the downside for whom.

Research, not investment advice.