‹ Micron
Ch 6 · The Cycle Question
The Finale · Risk
Memory was always cyclical.
✦ The bottom line
Memory chips have always boomed and busted. The whole question now: is AI different, or just delayed?
↓ the brief below
Micron · money in by fiscal year
$38B
FY18
FY19
FY20
FY21
FY22
FY23
FY24
FY25
Boom-bust-boom-bust. Every previous cycle ended. The question for the next chapter: does AI break the pattern, or just delay the next bust?
Source · 10-K · Historical annual revenue (FY18-FY25) · Updated May 26, 2026
From the 10-Q · the China block
Following the May 2023 decision of its cybersecurity review of our products sold in China, the CAC determined that critical information infrastructure operators in China may not purchase Micron products.
↳ A whole class of Chinese state-linked customers has been blocked from buying Micron for two years. That's revenue Micron simply can't get back without geopolitics shifting.
Source · 10-Q · Risk Factors · Q2 FY26 · Filed Mar 19, 2026
✦ Teach me
Who buys their stuff — and how much it matters
If most of a company's money comes from a small number of customers, that's customer concentration. Lose one big buyer, lose a lot of revenue all at once. Micron's biggest customer is 13% of revenue. Their top ten are over half. That's a lot riding on a few phone calls.
Wall Street calls this
Customer concentration
Tells you how *fragile* the boom is. A few customers cutting orders moves the whole story.
From the 10-Q · who buys from Micron
Revenue from one customer was 13% of total revenue for the first six months of 2026... over half of our total revenue came from our top ten customers... approximately one-half of our total revenue was concentrated in the data center end market.
↳ Half their money from data centers. Top 10 customers = over half. AI demand IS the story — but it's a concentrated story.
Source · 10-Q · Segment Information · Q2 FY26 · Filed Mar 19, 2026
Watch
China demand structurally blocked. Cyclicality still the bigger question.
You just finished
Chapter 6 · RISK
The Cycle Question
you now read: evidence-based prediction
Up next