Remember the left-tail?
If not, no one will blame you. It was, err, left for dead, what with nominal US GDP sustaining “boom” rates on AI investment and a resilient consumption impulse in the upper-half of the “K,” where the American dream’s alive and well thanks to buoyant asset prices.
Although the left-tail’s share of the “most likely macro outcome” vote rose to the highest since the start of the Iran war in this month’s installment of BofA’s Global Fund Manager survey, the fact that “hard landing” managed just 6% of the vote even after tripling speaks volumes.
As the figure shows, half of professional capital allocators still expect “no landing” (i.e., the macro right-tail), while four in 10 expect some version of a Goldilocks outcome.
The problem with consensus is that everything hinges on the enormous outlays (i.e., capex) driving the investment boom and earnings growth for so-called AI “enablers.”
That nods to a question that’s lurked in the background all year long, but which is now coming to the fore amid rising rates and jitters in credit markets: Would the first evidence of capex restraint by the big spenders be viewed as positive (i.e., a welcome sign of prudence and renewed discipline) or negative (i.e., a domino-tip moment)?
If Meta’s midweek update was any indication, the hyper-scalers won’t be announcing capex cuts anytime soon. But the mini-panic in neocloud names and attendant widening in data center CDS and junk AI spreads suggests mounting concerns with this global, leveraged buildout.
The figure above, which shows you Oracle CDS and AI-related high yield spreads, speaks for itself.
As alluded to above, this risks a domino effect. “The escalation of the ‘AI trajectory rethink’ which kicked off last week in earnest has continued,” Nomura’s Charlie McElligott said Thursday, noting that financing costs for speculative names look increasingly “non-linear, as credit spreads widen in somewhat stressy fashion, and with a lot more paper coming down the pike besides.”
This matters enormously for the overarching US macro zeitgeist. The burgeoning AI narrative shift (Charlie’s “trajectory rethink”) is “clearly impacting larger structural assumptions regarding capex trickle down into the economy,” he wrote, pointing to the nascent bond rally as clear evidence that “we are indeed repricing the left side of the 2027 US growth implied probability distribution.”
Interesting timing ahead of the midterms as well. I expect Trump to throw a few more hail mary’s especially if the market starts to go south. He’s got a bit of a conundrum though in that he probably doesn’t want to be seen as propping up the AI and data center market.
If Trump would get out of his own way with tariffs and middle eastern wars, he could get those rate cuts he’s been coveting which could delay the bottom coming out on semiconductors, but we all know he doesn’t possess the forethought to do that.
“Would the first evidence of capex restraint by the big spenders be viewed as positive (i.e., a welcome sign of prudence and renewed discipline) or negative (i.e., a domino-tip moment)”
Bond UST investors would probably take it to be a good sign because it would signal that the hyperscalers and the related industries would not be issuing quite so many bonds to fund expansion. And, as you alluded, it might be taken to mean that the AI-buildout-driven economy will slow as a consequence, taking off some of the upward pressure on some prices.
Stocks? Who knows. “Strategists” and X pundits will quickly find reasons to spin a capex slowdown as a buying opportunity. Perhaps because a slowdown in the AI arms race would allow the big guys to divert free cash flow to funding increased share buybacks.
** nicely done on that bond call. **