Micron Technology stock has risen nearly 600% through the last week of September to trade around $1,080 a share, leaving the memory maker at a forward price-to-earnings ratio of 6.8 times fiscal 2027 analyst estimates, a multiple the market typically assigns at peak earnings, not during the early innings of an upcycle.
The supercycle mechanics
Demand from AI data centers has pushed both DRAM and NAND prices sharply higher, lifting Micron’s revenue and gross margins. The company derives roughly 75% of revenue from DRAM and 25% from NAND. At the same time, the three major memory makers, Micron, SK Hynix and Samsung, have directed most of their capital and extreme ultraviolet lithography capacity toward high-bandwidth memory, a specialized DRAM that stacks alongside GPUs. HBM requires about three times the wafer area of conventional DRAM, and it competes for the same EUV tools needed for advanced logic. The result, analysts note, is that ordinary DRAM and NAND prices have risen more than already-elevated HBM pricing.
Why Micron differs from its rivals
Because Micron generates the smallest share of revenue from HBM among the big three, its earnings have benefited disproportionately from the run-up in commodity memory. Gross margins have expanded faster than at SK Hynix, the HBM market leader. The dynamic is ironic: the industry’s sprint toward the highest-margin product has tightened supply of the lower-margin products that still dominate Micron’s sales mix.
The analyst view on peak and trough
Consensus forecasts do not assume the historical boom-bust pattern has vanished. Analysts project earnings peaking in fiscal 2028, which ends August 2030, at more than $180 per share, then troughing at $48.21 in fiscal 2030 before a moderate recovery. The spread between peak and trough exceeds 70%, underscoring that cyclicality remains embedded in the model even with long-term AI supply agreements in place.
What the multiple implies
At 6.8 times forward earnings, Micron is priced as if the current profit level is the ceiling. A more normalized environment, the analysis suggests, would support a 12-to-15-times multiple. The gap between today’s valuation and that range measures the market’s skepticism that AI-driven demand can sustain commodity memory pricing once HBM capacity catches up.