AI growth faces severe physical bottlenecks
The guest argued that the AI build-out is hitting a physical roadblock due to power constraints and a lack of data center shells, while LLMs face rapid commoditization.
The argument
The guest noted that only a third of announced data centers have broken ground, and companies like Microsoft have previously slowed spend due to a lack of physical shells for GPUs. Furthermore, low-cost models like DeepSeek and a shift to token-based pricing are squeezing margins in a commoditized market lacking network effects.
The thesis, stress-tested
✓ What validates it
- ✓Further cancellations of compute contracts by major AI developers
- ✓Data center construction starts remaining below announcements
▸ Risks discussed
- ▸Hyperscalers find alternative energy sources
- ▸Breakthroughs in model efficiency reduce compute demand
Hear it yourself
"think with this new Iran war, it's becoming clearer to people that, you know, we are in an inflationary age. You know, the twenty twenties are an inflate a new inflationary era that is gonna persist for for quite some time. Right now is a really vulnerable time for the bond market where we can see interest rates in the long end break out to new highs."
00:05
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE