Research record
Not reviewedNo reviewer, review date, or supporting citations are recorded for this demo thesis. Scores are editorial opinions and have not been verified against cited evidence.
Revision history
No dated revisions are recorded for this thesis.
Investment thesis
The thesis in one line
Every AI GPU is stacked with high-bandwidth memory — and HBM is sold out. Micron is one of just three companies on earth that can make it.
Why memory became a growth story
Memory has always been a brutal commodity cycle. AI changed the mix. Modern accelerators need HBM — expensive, hard-to-manufacture stacks of DRAM bonded right next to the logic die. Demand has run ahead of supply, HBM capacity is booked out, and that scarcity is dragging the entire memory market's pricing higher. A commodity business suddenly has a premium, supply-constrained product at its centre.
The oligopoly
There are effectively three credible HBM suppliers worldwide. That concentration gives the survivors real pricing leverage during an up-cycle — a very different dynamic from the cut-throat commodity DRAM of the past.
Respect the cycle
Here is the discipline this name demands: memory is deeply cyclical, and reported P/E ratios are misleading at the turns (the multiple looks "expensive" at the bottom and "cheap" at the top — the opposite of intuition). The right mental model is to buy when sentiment is washed out and supply is being cut, and to take profits into euphoria. Get the cycle right and the returns are spectacular; get it wrong and the drawdowns are severe.
Stage of the cycle
A cyclical upswing driven by AI memory demand. The debate is about how long the up-cycle runs before supply catches up.
Scorecard
Upside 8/10 · Safety 5.5/10 · AI exposure 8/10. A cyclical bet on the scarcest ingredient in AI hardware.
