The easiest mistake with Micron is to treat the next quarter's earnings as a durable baseline.
The next mistake is to look at memory's history and assume every boom must end exactly like the last one.
Both shortcuts avoid the real question.
AI has made high-bandwidth memory strategically scarce, and Micron has signed agreements intended to make future demand more visible. But at $1,085.02, the market already expects that structural improvement to survive beyond today's extraordinary pricing.
What Micron actually controls
Micron is one of the few scaled producers of DRAM and NAND, and it has become a serious supplier of the HBM used beside AI accelerators.
That matters because a GPU without enough fast memory cannot deliver its theoretical performance. HBM is not an accessory. It is part of the compute system.
In fiscal Q3 2026, Micron's revenue increased 74% sequentially. Its two data-center businesses produced about 61% of company revenue, HBM4 entered high-volume shipments, and data-center SSD revenue more than doubled.
Management's fiscal Q4 guidance implied about $31 of non-GAAP EPS at the midpoint for one quarter.
Why does that matter?
Because Micron is producing earnings that would have looked impossible in an earlier memory cycle.
The agreements may change the cycle, not abolish it
Micron has disclosed 16 strategic customer agreements, most of which collectively represent roughly $100 billion of minimum-price revenue and generally extend through 2030. Management ultimately expects agreements of this kind to cover at least half of company revenue.
This could be the structural change.
Memory producers historically deserved low multiples because volume, price, and margin could fall together. Multi-year commitments and pricing floors may soften that downside and improve investment visibility.
But floors can coexist with ceilings. Customer deposits are financing, not profit. Micron still must spend heavily to build capacity. The agreements can reduce the depth of the cycle without turning memory into software.
What $1,085.02 requires
I use fiscal 2030 because the customer agreements generally extend through that period. With a 12% required annual return and a 15 times terminal multiple, today's price requires about $114 of normalized fiscal 2030 EPS.
The fiscal Q4 midpoint annualizes to roughly $124, but that is a one-quarter run rate during exceptional conditions. It is not management guidance for 2030.
The market is therefore assuming that normalized earnings remain close to today's extraordinary level.
That is a much harder claim than saying next quarter will be strong.
Three possible outcomes
Conservative: about $381 today
Micron earns $50 per share in fiscal 2030 and receives a 12 times terminal multiple.
This assumes AI demand remains real but memory pricing normalizes sharply, capital spending rises, and the market keeps treating Micron as cyclical.
Durable case: about $763 today
Micron earns $80 per share in fiscal 2030 and receives a 15 times terminal multiple.
This assumes strategic agreements improve the cycle, HBM leadership persists, and normalized earnings remain far above prior trough economics.
Upside case: about $1,258 today
Micron earns $110 per share in fiscal 2030 and receives an 18 times terminal multiple.
This requires AI memory demand, pricing discipline, and contract protection to support a structural re-rating.
At $1,085.02, MU trades about 42% above my durable case and only 14% below the upside case.
The price is no longer offering the structural thesis for free.
Capital intensity still decides who gets paid
Micron ended fiscal Q3 with roughly $30.2 billion in cash and investments against about $5.7 billion of debt.
That is a strong balance sheet.
It also expected roughly $27 billion of fiscal 2026 capital spending and indicated that quarterly capital expenditure in fiscal 2027 should exceed the fiscal Q4 level.
The tension is obvious: scarce supply creates the profits, but building more supply consumes the cash.
If Micron spends into a demand slowdown, shareholders can be right about AI and wrong about returns.
What to watch out for
- Fiscal Q4 results on September 30, 2026 - because this research snapshot will age quickly after the release.
- HBM4 and HBM4E qualification, volume, and customer breadth - because leadership must persist across generations.
- Strategic-agreement coverage, floors, and ceilings - because contract detail determines how much the cycle actually changes.
- Gross margin and capital expenditure together - because margin without after-investment cash flow is incomplete.
- Debt, deposits, and diluted shares - because financing and dilution determine what reaches each share.
What could change the thesis
The thesis would strengthen if contract-backed revenue expands toward half the business, HBM share strengthens across multiple customers, and free cash flow remains robust after the fiscal 2027 investment step-up.
The thesis would weaken if memory prices roll over faster than floors protect them, price ceilings surrender too much shortage upside, or capital spending keeps climbing as end demand slows.
The strongest evidence would be a higher through-cycle cash-flow floor, not another peak quarter.
Bottom line
Micron's AI memory thesis is working.
Business quality: strengthening rapidly.
Balance sheet: strong.
Cycle visibility: better, but not proven through a downturn.
Valuation: much closer to the upside case than the durable case.
At $1,085.02, MU needs roughly $114 of normalized fiscal 2030 EPS at a 15 times multiple to deliver a 12% annual return. That is possible. It is not conservative.
The practical principle: never confuse peak earnings with normalized value, even when the cycle has genuinely improved.
Sources
- Micron fiscal Q3 2026 results
- Micron fiscal Q3 2026 Form 10-Q
- Micron fiscal Q3 prepared remarks
- Micron fiscal Q4 reporting date
Scenario values are analytical estimates, not company guidance or price targets. This article precedes Micron's scheduled September 30 earnings release. Educational only - not financial advice.