Physical bottlenecks drive the AI scarcity trade
The bull case presented is that the AI build-out has shifted from pre-training to inference and agentic AI, creating massive scarcity and bottlenecks across physical hardware, servers, and chemicals.
The argument
The guest argued that the market has reached a severe compute and hardware shortage. Investors should focus on physical bottlenecks - including semiconductors, servers, memory, and specialized chemicals - rather than software.
The thesis, stress-tested
✓ What validates it
- ✓Continued breakout price action and earnings beats from server manufacturers like Dell and HPE
- ✓Rising prices for memory (HBM) and specialized semiconductor components
▸ Risks discussed
- ▸Overbuilders could eventually lead to a supply glut in hardware and memory
- ▸High capital expenditure budgets from hyperscalers could be reined in if AI monetization stalls
Hear it yourself
"And I said that if you look at what's performed so far this month within the IGV, which is the software ETF, almost all of the top 10 names were related to crypto. So let's separate Bitcoin for a second. What is happening is the miners were kind of the ones that led way before Bitcoin. And I mentioned them last week, I think here, but definitely on my weekly, where I just said we've reached a point of compute shortage. And if you can scramble and get anything related to AI, that's where you wanna go. We have bottlenecks throughout the world now."
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE