They are completely sold out of their highest-margin product for a year, yet the street treats this as a growth cap rather than a pricing weapon.
Thesis: Micron has achieved the holy grail of commodity manufacturing: scarcity in a high-value tier. The 'capacity ceiling' bear case ignores the margin implication. When you are sold out of HBM through 2025 with yields beating plan, you don't chase volume; you squeeze price. The 190bps operating margin expansion proves the cost curve is working. The street is modeling a cyclical memory peak, but this is a structural mix-shift to high-margin AI infrastructure. You buy the bottleneck owner.
Verdict: LONG — Conviction: HIGH
Catalyst: HBM share parity with DRAM share in H2 2025. Hitting this target validates the yield ramp and unlocks the next leg of margin expansion.
Key Risk: HBM4 transition in 2026. If competitors qualify faster or Micron's 1-beta node hits physical limits against hybrid bonding alternatives, the premium evaporates.
The Tell: When Harlan Sur asked about 2026 committed supply, Mehrotra shifted from 'sold out' (the 2025 answer) to 'working closely with customers.' This confirms the order book for 2026 is not yet hard-locked, exposing them to potential cycle volatility if AI CapEx cools.
Friction Level: MODERATE_FRICTION — Bears see 'Sold Out' as a revenue ceiling. Bulls see it as the ultimate pricing leverage in a supply-constrained market.
Report not found
The report data is no longer available. Please return to the archive.