Micron Technology Stock Surged On A Shortage The Company Had Already Described

The memory shortage behind the run was laid out in Micron’s own supply disclosures long before the price caught up.

Micron Technology (MU) stock is up 708% over the past year, and the easy explanation is that it caught the AI trade. The narrower truth is that the shortage that repriced memory was set out by the company itself, in disclosures about its own wafer capacity, months before the price moved.

Three Wafer Starts For Every One Of Ordinary DRAM

The earliest of those disclosures came in December 2024, when management put the manufacturing arithmetic on the record: producing a given quantity of bits as HBM3E, the high-bandwidth memory that feeds AI accelerators, takes about three wafer starts where conventional DRAM needs one. Every bit sold as HBM therefore consumed the wafer capacity of roughly three bits of ordinary DRAM. Management said at the same time that HBM was already pressuring non-HBM supply.

The Company Said Its Own Supply Would Fall Behind

By March 2025 the same thread had hardened into a forecast. Micron said its own supply growth in calendar 2025 would run below industry demand growth in both DRAM and NAND. The company’s HBM output for calendar 2025 was already sold out. Micron was also shrinking the smaller of its two businesses on purpose, taking NAND wafer capacity down over 10% structurally by the end of fiscal 2025. Those were statements about Micron’s own supply, not forecasts of demand: a supplier telling the market it had chosen to make less of the ordinary product.

The Quarter Where Scarcity Became Pricing Power

The arithmetic reached the income statement in fiscal Q3 2026. Revenue reached $41.5 billion, up 346% year over year, as DRAM prices rose in the low 60s percentage range sequentially, and Micron has guided fiscal Q4 2026 to a record $50 billion, give or take $1 billion. Scarcity is now written into contracts as well: 16 strategic customer agreements are signed, with $22 billion of customer cash deposits and related financial commitments projected under them. Memory does not usually hold economics like this. On a trailing-twelve-month basis as of fiscal Q3 2025, just before the run, net margin was 18.4% against a three-year average of -1.3%, a good reading sitting on three years that averaged a loss. Durable margins and cash generation are among the things the Trefis High Quality Portfolio insists on in its holdings, and memory has historically been a hard place to find either.

Readable In Advance, Impossible To Size

Were the signs actionable, or only obvious now? The mechanism was readable: anyone who took those disclosures seriously knew memory was heading into a shortage Micron’s own supply plan would deepen rather than relieve. The magnitude was not. A stock does not multiply on a wafer ratio, and nothing said in December 2024 or March 2025 sized a move of this order. The peer returns point the same way: over the same window the S&P 500 returned 23% and NVDA returned 25%, while WDC returned 490%. This was memory scarcity repricing, not the AI trade lifting every chip name alike. The repeatable part is the habit, not the trade: read a company’s own supply and outlook language before the price agrees with it, which is what a screen of companies whose guidance is climbing is for.

Memory Cycles Turn Both Ways

The same disclosures that made this shortage legible will eventually describe its end, and a position sized for scarcity is a bet on how long one lasts. A rules-based basket like the Trefis High Quality Portfolio spreads that timing problem across many businesses instead of resting it on one. That portfolio has a track record of outpacing the three major indices – the S&P 500, S&P Mid-cap, and Russell 2000.