⚠ Supercycle Margins Will RevertHigh threat

An eighty-five-percent gross margin on memory chips is one of the most extraordinary — and most certainly temporary — numbers in the market, and the central risk in Micron is that these margins will revert, perhaps sharply. No physical-goods commodity sustains software-like margins, because such margins are a flashing signal to add capacity, and capacity, once added, competes the excess profit away. Memory's entire history is a cycle of exactly this: extraordinary boom margins draw extraordinary investment, the investment creates overcapacity, and margins collapse — sometimes to losses. Micron's own margins were negative not long ago, in the fiscal 2023 down-cycle¹; the swing from there to eighty-five percent is the measure of how cyclical this business is, and how far the current numbers sit above any sustainable level.

The only real question is how far and how fast the reversion comes. The bull case holds that HBM's genuine differentiation and the structural, durable demand of the AI buildout have permanently raised memory's floor, so that the next trough is far higher than past ones — margins settling at, say, a healthy fifty or sixty percent rather than collapsing to zero. The bear case holds that a peak is a peak, that all three makers are building the capacity that will end the shortage, and that when the AI-investment cycle digests, memory will do what it always does. Both cannot be fully right, and the truth is likely between them. But no honest investor should model eighty-five-percent margins as a baseline; they are the zenith of the greatest memory boom in history, and the discipline that owning Micron demands is to value it on mid-cycle economics, not peak ones — because the peak, however glorious, is the one thing about memory that is guaranteed not to last.