MU Micron Technology
Peak-cycle memory priced as a perpetual monopoly; great business, but margins will come down.
The story
Micron is one of three global DRAM suppliers and a top NAND player, now at the peak of an AI-driven memory supercycle in which HBM supply is sold out and contract prices have pushed TTM revenue to $90B at a 66% operating margin. HBM, plus DRAM oligopoly discipline, gives it a more durable position than in past cycles. Memory is still a capital-intensive commodity business, though: capacity additions from Micron, SK Hynix and Samsung (and later CXMT) will push pricing and margins back toward through-cycle levels. The market is pricing today's peak as permanent; I see a mature cyclical franchise with a structurally higher, but not peak, margin.
Year-one growth of 30% reflects contracted HBM volumes and elevated DRAM pricing into 2027. The 42% target margin sits well above the historical through-cycle level of about 25-30%, crediting HBM mix and oligopoly discipline, but well below the 66% peak because supply responds. Sales-to-capital of 1.2 reflects fab capex intensity, beta 1.4 reflects memory's cyclicality above the 1.3 semiconductor anchor, and the 8-year horizon fits a cyclical business, not a franchise. These drivers still value the stock well below the market. I accept that gap rather than inflate margins, and set confidence low because of it.
Value drivers
| Revenue growth (Y1) | 30.0% |
| Terminal growth | 3.5% |
| Forecast horizon | 8y |
| Target operating margin | 42.0% |
| Years to target margin | 5 |
| Sales-to-capital | 1.20 |
| Beta | 1.40 |
| Failure probability | 3.0% |
| Cost of capital (WACC) | 11.5% |
| Terminal WACC | 9.7% |
Valuation bridge
| PV of explicit FCFF | 319.84B |
| PV of terminal value | 614.61B |
| Equity value | 904.58B |
| ÷ shares → per share | $800.95 |
Projected free cash flow to the firm
| Yr | Revenue | Growth | Margin | NOPAT | Reinvest | FCFF | PV |
|---|---|---|---|---|---|---|---|
| 1 | 117.36B | 30.0% | 60.9% | 63.20B | 22.57B | 40.63B | 36.44B |
| 2 | 148.12B | 26.2% | 56.2% | 73.56B | 25.64B | 47.93B | 38.56B |
| 3 | 181.34B | 22.4% | 51.5% | 82.47B | 27.68B | 54.79B | 39.54B |
| 4 | 215.15B | 18.6% | 46.7% | 88.85B | 28.17B | 60.67B | 39.28B |
| 5 | 247.11B | 14.9% | 42.0% | 91.70B | 26.64B | 65.07B | 37.78B |
| 6 | 274.47B | 11.1% | 42.0% | 101.86B | 22.80B | 79.06B | 41.17B |
| 7 | 294.47B | 7.3% | 42.0% | 109.28B | 16.66B | 92.61B | 43.27B |
| 8 | 304.78B | 3.5% | 42.0% | 113.10B | 8.59B | 104.51B | 43.80B |
Key risks
- Capacity overbuild by Samsung, SK Hynix and CXMT collapses DRAM/HBM pricing as in 2019 and 2023
- AI capex digestion or hyperscaler spending pause cuts HBM demand
- Loss of HBM share or qualification slips at Nvidia/custom ASIC customers
Catalysts
- HBM4 ramp and multi-year contracted pricing that cuts earnings volatility
- Sustained supply discipline keeping DRAM contract prices elevated through 2027