MU at a $1 Trillion Valuation: Supercycle or Cycle Top?

The question on every memory investor's mind: with Micron through $1T, is this a durable supercycle or the part of the cycle where you get hurt?

Start with the numbers

DRAM average selling prices rose over 60%, NAND is up sharply, and gross margins are guided near 68%, numbers this industry almost never sees. Micron's HBM4 is in volume supply to Nvidia, and 2026 HBM is entirely sold out under long-term contracts. SK Hynix, the upstream leader, is up 186% this year and holds about 70% of Nvidia's HBM orders.

Why this cycle may be different

Past memory cycles crashed because supply flooded the spot market. This time, HBM is sold on multi-year contracts to a handful of AI buyers, and supply growth is disciplined, which can keep margins high far longer than skeptics expect. The risk is that memory is still memory, and every supercycle felt permanent right before it broke.

My take: the contract visibility is real and I am constructive on the multi-year demand, but I would not chase a name at a trillion dollars right after this run. I would rather scale in on pullbacks. Not investment advice.