I like to say failure comes from a failure to imagine failure. Let’s imagine some.
There’s a discipline, older than modern finance, before daily CNBC and fintwit that consists of nothing more than watching what money does when it thinks no one is narrating. The bond market doesn’t give interviews. It doesn’t post. It prices risk and moves, and when it moves in a direction that contradicts the story equities are telling, the historically correct response is to ask why the less emotional of the two is frightened.
It is possible what is happening in markets is just a rotation. That the software selloff Saaspocalypse is sector-specific, that the treasury rally is temporary, and the economy remains sound, that consumer weakness is noise and that the capex intensive AI scaling thesis is intact.
It’s also the kind of story markets tell themselves in the time between something quietly breaking and when the break becomes loud or noticeable and impossible to ignore.
Right now the bond market is telling a different story, maybe quietly, through price. And what price is saying, across a ton of instruments: Treasuries, credit spreads, leveraged loans, and credit default swaps all at the same time, is that the consumer is weakening, a multi-trillion-dollar credit structure built in part on peak-vintage software valuations is fracturing and the largest AI infrastructure buildout in history is being funded on terms that assume a future that will surely materialize for some but also surely not materialize for all.
20 basis points of TRUTH
The 10-year Treasury yield fell from 4.22% to 4.05% in a week. In bond market terms it’s less a slide and more a shout. The catalyst came in a sequence starting with retail sales missing expectations, jobs softening, home sales falling by the most since 2022. Every economic release no matter how much one might discount them said pretty much the same thing: the consumer isn’t where equities say they are.
End of year retail sales were flat against expectations of modest growth. Employers added just 28,000 jobs per month since December which is a drip compared to 400,000 during the 2021–2023 recovery. January had a payroll bounce and then equities pounced on it like translating it as weakness being transient.
The bond market wasn’t persuaded. The yield curve was negative from October 2022 through December 2024 and has now un-inverted. That's historically a signal before onset of recession.
Equity investors may be seeing the data as mixed while bond investors may be seeing it as directional. Equity investors are biased to find reasons to stay invested while bond market investors are biased to avoid losing principal. The bond market may also be anticipating political games around Kevin Warsh confirmation and risks around future policy and Fed independence. When the two markets diverge and disagree the question is this: which one has more to lose from being wrong?
When a Stock Crash Becomes a Loan Problem
Software represents something like 13% of the $1.5 trillion leveraged loan market and roughly 20% of private credit BDC investments. A lot of those loans were originated at peak valuations between 2021 and 2023, when private credit snagged 40–70% of buyout financing which is thethe riskiest tranche at the most expensive moment, many with fewer covenants than any prior generation of lenders may have accepted.
Evidence? We have seen Blue Owl's software BDC see 15% asset redemptions. And over $7 billion in redemption requests hit software-heavy BDCs in late 2025. UBS put out a note predicting $75–120 billion in defaults by year-end.
Like cockroaches, there’s never just one. When you find one borrower in distress, the question is never whether others exist but whether the structure can absorb whatever losses are coming without triggering a negative feedback cycle with redemptions forcing sales, sales depressing marks, depressed marks triggering more redemptions…It hasn’t happened yet but doesn’t mean it won’t.
The Oracle Seismograph
While the Oracle of Omaha stays away from all of this (and steps into the sunset of his amazing career), we may this year sadly see a poetic confirmation of his passing in parallel with rumors of the death of value investing itself being greatly exaggerated. The other Oracle (led by Ellison) may be where the AI narrative meets credit reality. Its 5-yr CDS spread tripled from 43 to 139 basis points in two months. Total debt exceeds $100 billion, with $248 billion in new lease commitments tied to data centers. At least $300 billion of its remaining performance obligations are tied to one counterparty: OpenAI.
Poetically, Oracle embodies a question the market has not yet answered: what happens when the largest infrastructure buildout since the postwar highway system is financed on terms that assume exponential demand growth and the demand curve flattens before the debt matures?
2000 or 2007?
In 2000 the .com bust and tech crash was an equity event. It was painful but structurally contained because the credit system wasn't complicit. By contrast in 2007, losses in the single sector of housing spread through a complex, opaque, leveraged credit architecture into the entire financial system.
This time is different? There’s roughly $3 Trillion of private credit with significant software exposure, opaque marks, and $7 billion in redemption pressure is structurally more similar to 2007's transmission mechanism than 2000's equity-only losses though also at significantly smaller scale relative to GDP. This does not mean 2007 is the outcome. High-yield spreads at 284 basis points (the tightest 5% of readings in 25 years) have little room to absorb bad news and enormous room to reprice. If spreads widen while equities hold steady, then the bond market's murmurs are confirmed. If private credit markdowns exceed say 3–5%, the redemption cycle intensifies.
The Message
The bond market doesn’t narrate like the equity market, it just prices. And what it’s pricing across lots of relevant instruments is a clear thesis: the consumer is decelerating, credit structure is under stress, the largest infrastructure financing cycle in a generation is bumping into friction, and the institution responsible for managing the response may have political distraction at the worst possible time.
In the long history of disagreements between the stock and bond markets, the price has been right more often than the story. Not because bond investors are smarter (though some are) but because they’re paid to be afraid and to fear loss. And fear is a more reliable guide to reality than hope.
