◆ What the Market Isn't Pricing In

Micron presents one of the most fascinating valuation puzzles in the market, and it is the mirror image of a wide-moat compounder's. Here is a company that has reached a market value of about $1.24 trillion¹, posting record revenue around ninety billion dollars, record net income near fifty billion, and gross margins approaching an almost unheard-of eighty-five percent — and yet it trades at only about nineteen times earnings and under eleven times sales. The whole investment question is what to make of that low multiple on such spectacular numbers, because it is not an oversight. It is the market's considered judgment, and decoding it is the key to the stock.

The low multiple says, unmistakably, that the market does not believe these earnings are sustainable. Memory is cyclical, these are peak-cycle earnings, and a nineteen multiple on peak earnings is the market pricing in reversion — expecting profits to fall, perhaps sharply, as the cycle turns and capacity catches up with the AI-driven demand. This is rational and rooted in a century of memory history: the market has seen Micron swing from a six-billion-dollar loss in fiscal 2023² to fifty billion in profit today, and it is not about to capitalize the peak as though it were permanent. So on the conventional reading, Micron is not cheap despite its low multiple — it is a cyclical correctly priced near its peak, where low multiples are a warning, not a bargain.

But there is a genuine bull case for what the market may be under-pricing, and it rests on the possibility that this time the structure has changed. The argument: HBM is a real, durable differentiator, not a passing shortage; the AI buildout represents a structural, multi-year step-change in memory demand, not a normal cyclical wiggle; and the three-player oligopoly has grown disciplined enough that the next trough will be far shallower than history suggests — margins settling at, say, a healthy fifty or sixty percent rather than collapsing. If that is right, the market's reversion assumption is too pessimistic, mid-cycle earnings will prove far higher than the bears expect, and a nineteen multiple on earnings that don't fall much is genuinely cheap. On this view, the market is anchored to memory's ugly past and missing a structural transformation in its present.

The bear case is the weight of that same history plus the specific risks: all three makers are pouring capital into capacity that will become tomorrow's oversupply; the AI-investment cycle driving demand is unprecedented and untested; China's subsidized entry threatens the oligopoly's discipline over time; and eighty-five-percent margins on a commodity are, almost by definition, a peak that reverts. What the market may be under-appreciating, in either direction, is the sheer width of the range of outcomes: memory at a historic peak is a business where the plausible cases span from 'structurally re-rated, and cheap here' to 'classic cyclical top, and about to halve.' The honest verdict is that Micron is a narrow-moat cyclical at the summit of the greatest boom in its history, priced at a multiple that already assumes reversion — so the question is not whether earnings fall, which they almost certainly will from this peak, but whether the new floor is high enough, thanks to HBM and AI, to make even the reverted earnings worth more than today's modest multiple implies. That is a real debate with serious arguments on both sides, and it is the entire investment case. The market has priced Micron as a cyclical near its peak. Whether it is instead a transformed business near a new, higher baseline is the one question that matters — and it is one that only the next downturn will truly answer.