This is a useful distinction: AI is becoming consensus as a growth narrative, but not yet as a rates narrative.
The unresolved question is whether AI capex extends nominal growth through investment, power demand, data centers, and hardware spending before the productivity gains arrive — or whether the longer-term disinflation/productivity story ultimately dominates.
That is why the split between T. Rowe Price and Candriam is interesting. One treats AI and related structural forces as part of a higher-rate regime, while the other still sees enough risk to justify long duration.
So the real debate is not whether AI matters. It is whether AI keeps neutral rates and inflation risk higher for longer, or eventually becomes the productivity shock that supports bonds.
Tony, agreed, and the pick-and-shovel history is instructive here. In the California gold rush, the miners mostly broke even or went bust. The money went to Levi Strauss and the hardware merchants.
The AI consensus right now is heavily weighted toward the infrastructure layer: semiconductors, hyperscalers, data centers. That is the shovel trade. Whether the productivity gains eventually flow to the users, and what that does to growth, inflation, and rates, is exactly what Invesco is flagging when they say it's still hard to "sort the AI wheat from the cyclical chaff."
The consensus knows AI matters. It has not yet priced where the value lands.
This is a useful distinction: AI is becoming consensus as a growth narrative, but not yet as a rates narrative.
The unresolved question is whether AI capex extends nominal growth through investment, power demand, data centers, and hardware spending before the productivity gains arrive — or whether the longer-term disinflation/productivity story ultimately dominates.
That is why the split between T. Rowe Price and Candriam is interesting. One treats AI and related structural forces as part of a higher-rate regime, while the other still sees enough risk to justify long duration.
So the real debate is not whether AI matters. It is whether AI keeps neutral rates and inflation risk higher for longer, or eventually becomes the productivity shock that supports bonds.
The consensus on AI is easy.
The harder question is what AI means for growth, inflation, and ultimately interest rates.
That debate may matter more for portfolio returns than AI itself.
Tony, agreed, and the pick-and-shovel history is instructive here. In the California gold rush, the miners mostly broke even or went bust. The money went to Levi Strauss and the hardware merchants.
The AI consensus right now is heavily weighted toward the infrastructure layer: semiconductors, hyperscalers, data centers. That is the shovel trade. Whether the productivity gains eventually flow to the users, and what that does to growth, inflation, and rates, is exactly what Invesco is flagging when they say it's still hard to "sort the AI wheat from the cyclical chaff."
The consensus knows AI matters. It has not yet priced where the value lands.