An excellent, financially pristine top-three memory maker at a peak-cycle price - a cheap-looking P/E on record earnings is a value trap, not a bargain. Real long-term optionality, but you are buying a rich multiple into a cyclical peak.
Micron at $837.33 on 31 July. This is a quality business at the wrong point in its cycle: it screens cheap at about 19 times trailing earnings, but those earnings are struck at the top of an AI-driven memory shortage. On mid-cycle earnings the framework rates it Hold in every horizon.
Micron is one of only three companies in the world that make the memory chips modern computing runs on, and pricing is the best in its history. Quality scores 80. The balance sheet is pristine - net cash of about 19.6 billion dollars, a current ratio of 3.4 times and interest coverage of 258 times - with a gross margin of 84.6 percent on 41.5 billion dollars of quarterly revenue. Nothing here is broken. The Hold is entirely about the price you pay, and where memory sits in its cycle when you buy it.

Memory is a commodity cycle. Margins swing from the low thirties at the trough to the mid eighties at the peak, and today it sits at the peak. So the roughly 19-times trailing multiple is struck on record earnings of about 44 dollars a share that history says will not last. Score it on normalised mid-cycle earnings and it trades near 31 times, about 1.48 times what rates and growth warrant, and above the 28-times semiconductor guardrail. Price to book is 9.4 times against book value near 89 dollars - memory tops here and bottoms nearer one times book. Fair value lands around 650, below the 837 price.

The operating cycle is still hot, but the equity has already turned. Micron is down about 33 percent from its 1,255 high set in June and trades below a falling 50-day average near 962. The AI and high-bandwidth-memory up-cycle scores 68 as a driver - a genuine tailwind, and the reason the longer-term case exists - but it sits at a peak level and cannot amplify a stock that is already Expensive. For a deep cyclical at the top of its cycle, a tape that has rolled over is exactly the signal the framework weights most.

The risks here are not hypothetical, and they carry equal weight to the franchise. First, this is a commodity cycle: gross margins near 85 percent mean-revert toward 30 to 40 percent when new capacity lands into softer demand, and a Samsung price war would hit the number-three supplier first. Second, the stock is Expensive at the peak - about 31 times mid-cycle earnings, 1.48 times warranted. Third, United States and China export controls on advanced memory are a chronic overhang. The scenario spread is unusually wide: a 1,300 bull if the super-cycle extends, against a 440 bear if the cycle turns and the whole AI-capex cohort de-rates together. That is real optionality, and real downside.

Against the current US$837.33, the report frames a bull case at US$1,300 (+55%), a base case at US$720 (-14%) and a bear case at US$440 (-47%). See the full report for the probability weight behind each path.
An excellent, financially pristine top-three memory maker at a peak-cycle price - a cheap-looking P/E on record earnings is a value trap, not a bargain. Real long-term optionality, but you are buying a rich multiple into a cyclical peak.
Read the full report on donatien.ca →