AI infrastructure remains the dominant trade over software
The bull case for AI infrastructure, particularly memory and semiconductors, is supported by a structural supply deficit that cannot meet exponential token demand.
The argument
The speaker argued that while SaaS and seat-based software face structural disruption and multiple compression, hardware and memory providers like Micron benefit from years of underinvestment in capacity. He compared the current dynamic to fracking, noting that unlike oil, the supply side of AI compute cannot easily scale to match exponential demand.
The thesis, stress-tested
✓ What validates it
- ✓Continued earnings beats and positive guidance from major memory and semiconductor manufacturers
- ✓Hyperscalers expanding their contracted compute backlogs beyond current levels
▸ Risks discussed
- ▸Potential cyclical downturns in semiconductor demand
- ▸Over-extrapolation of short-term capital expenditure by hyperscalers
Hear it yourself
"They've continued to dominate the AI trade, but in two years, they've now gone up 48%. Good returns. I mean, you're you're getting 20 plus percent returns. You're outperforming the s and p, but it's not the 12 times that happened. And that's where we are in AI right now for all of these names."
00:00 / 00:22
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE