Most investors are approaching AI as a sector.
I don’t think that framework holds up anymore.
What’s actually emerging is a system—where demand in one layer forces investment into the next. Capital isn’t moving randomly. It’s moving through constraints.
The way I’ve been positioning reflects that:
---
Demand — Compute (MU)
AI starts here. Memory and compute demand continue to accelerate.
Flow — Networking (ANET)
Compute is useless without moving data efficiently at scale.
Power — Infrastructure (PWR)
None of this works without reliable, scalable energy.
Validation — Testing (TER)
Every layer above has to be proven, tested, and brought into production.
---
The takeaway isn’t “pick the best AI stock.”
It’s that returns are increasingly coming from owning multiple layers of the same system, not just one narrative.
Curious how others are thinking about positioning across the stack.