Thematic Exposure -- 6/10
The AI memory super-cycle. Micron is a top-3 DRAM maker and #2 in HBM, riding the single hottest
secular demand driver in semis. The position is genuine -- a 3-player DRAM oligopoly (~90% top-3),
structurally tight supply "beyond calendar 2027," locked further by 16 take-or-pay Strategic Customer
Agreements (~$100B RPO floor). But Micron is the smallest of the three in its primary DRAM market
(~22% share, behind Samsung and SK Hynix), and #4-5 in a fragmented NAND market. That third-place
standing in the core segment is what caps an otherwise outstanding thematic setup at 6/10.
Weight: 35%
AI Memory Super-Cycle -- Elite Theme
Secular Tailwind -- Multi-Year Visibility
Micron sells the scarcest input to the AI buildout: HBM and data-center DRAM. HBM TAM is
projected ~$35B in 2025 to ~$100B in 2028 (~40% CAGR, per Micron). Supply is structurally tight
"beyond calendar 2027" with no line of sight to catch-up, and 16 signed SCAs (take-or-pay, 5-yr,
~$100B RPO floor, ~$22B deposits) lock in roughly half of future revenue -- a real shift from
spot-priced commodity memory toward contracted, partial price-setting on ~40% of revenue.
#3 in DRAM -- A Real Oligopoly, But Not the Leader
The Score Ceiling -- Third Player
DRAM is a textbook 3-player oligopoly -- Samsung 38.6% + SK Hynix 28.8% + Micron 22.4% ≈ 90%
combined (Omdia Q1'26). The structure is excellent, but Micron is the #3 of three, at ~22% share --
below the >30% leadership threshold the rubric reserves for the 7-8 band, and the firm the framework
explicitly says not to settle for over the leader. Strong theme, real oligopoly, but a #3 seat.
HBM #2 -- A Genuine Share-Gain Story
Overtook Samsung on NVIDIA HBM4
HBM is a duopoly tilting to a triopoly -- SK Hynix 62%, Micron 21%, Samsung 17%. Micron has
overtaken Samsung on NVIDIA HBM4 allocations (~20% of HBM4 vs SK Hynix mid-50s, Samsung mid-20s) --
the biggest competitive surprise of the cycle and a genuine share-gain story. HBM4 12-high is
ramping 2x faster than HBM3E, with >$1B already shipped. This is what lifts the score above the
fragmentation floor.
NAND -- The Soft Spot
#4-5, Fragmented
NAND (24% of revenue) has 5 credible players (Samsung ~28%, SK Hynix ~22%, Kioxia, Micron ~12-15%,
SanDisk), none with a commanding lead. This segment is fragmented and is where the thesis is weakest.
Segment Revenue & Positioning (FY26Q3)
| Segment | Revenue | % Rev | Market Share | Theme Growth |
|---|---|---|---|---|
| Cloud Memory (HBM-heavy) | $13,769 | 33% | #2 in HBM, ~21% (SK Hynix 62%, Samsung 17%) | Very high (HBM ~40% CAGR) |
| Core Data Center | $11,524 | 28% | #3 in DRAM (~22% blended) | >10% (DC bits >2x vs 2 yrs ago) |
| Mobile & Client | $11,521 | 28% | #3 in DRAM / #4 NAND | Low-to-mid teens (DRAM bits) |
| Auto & Embedded | $4,634 | 11% | top-3 supplier | High (content-driven, 5x by 2030) |
Oligopoly Gate
| Criterion | Result |
|---|---|
| Micron DRAM share | ~22.4% (#3 of 3) |
| Top-3 combined DRAM share | ~90% (≤3 players >70%) |
| Any segment >30% Micron share? | No |
| Customer replace Micron within 12 mo? | No (moat) |
| Gate result | PASS |
6/10 — The oligopoly hard gate is
satisfied: DRAM is controlled by ≤3 players holding >70% (~90% combined), and Micron is one of
those three, so the 5/10 fragmentation ceiling does NOT apply. Capital intensity (greenfield fabs,
EUV, qualification cycles) plus 16 take-or-pay SCAs mean no customer can replace Micron within 12
months. The theme is elite and compounding well above 10%. The score is held to 6 (not 7-8) by the
one principle that dominates here: Micron is the third player in its primary DRAM market at ~22%
share -- below the >30% threshold, and a #4-5 fragmented NAND tail. Strong theme, real oligopoly,
but not the leader.
Data sourced from Daloopa (company_id 136), Omdia/Counterpoint/TrendForce, and the MU FY2026Q3 earnings transcript.