Memory Cycle Peak Verification — Contract Price & Earnings Gauntlet
Scenario Comparison
TrendForce Q2 CY2026 DRAM contract price forecast shows +10-30% QoQ; Samsung/SKHynix Q1 earnings confirm strong ASP realization; MU Q3 beats 81% GM guide
TrendForce Q2 CY2026 DRAM contract flat to +5% QoQ (sharp deceleration from +90-95%); Samsung/SKHynix guide cautiously; MU Q3 meets but Q4 guidance softens
TrendForce Q2 CY2026 DRAM contract -5 to -15% QoQ; Samsung undercuts on HBM pricing; spot-contract spread inverts; MU Q3 misses 81% GM or cuts Q4 outlook
All Scenarios
4Positioning Suggestions
→ DO NOT add to MU at current levels until Q2 CY2026 contract price direction is confirmed (early April TrendForce report). The 5.5x forward P/E is a cycle-peak signature, not a buy signal — identical to 2018 peak (5.6x).
→ If TrendForce Q2 contract forecast shows +10% QoQ or better: MU at $321 offers +40% to base case $450. Consider initiating position with tight stop at $280 (-13%).
→ If TrendForce Q2 contract forecast shows flat or negative: stand aside entirely. Wait for E revision cycle to complete before re-evaluating.
→ Samsung earnings (late April) is the second gate: watch for HBM pricing aggression. Samsung historically starts price wars to gain share — any signal of this kills the supercycle thesis.
→ Memory WFE names (AMAT, LRCX) are leading indicators — if they guide down in April/May earnings, memory capex peak is confirmed regardless of MU's Q3 results.
→ Use the April-June gauntlet as a discipline framework: each data point either confirms or invalidates the structural HBM thesis. No single data point is sufficient alone.
Predictions
12TrendForce Q2 CY2026 DRAM contract price forecast will show positive QoQ growth (>0%)
If positive → MU $321 is a buy with $450 base target. If negative → cycle peaked, stay away.
Q2 CY2026 DRAM contract price QoQ increase will decelerate to <30% (from 90-95% in Q1)
Deceleration is expected and already priced in. The key question is the MAGNITUDE of deceleration, not the direction.
DRAM spot prices will decline >15% from Q1 CY2026 peak for 4+ consecutive weeks before June 25 (Bear Trigger #1)
If triggered → cycle peak confirmed by leading indicator. Reduce all memory exposure. This is the monitoring config bear trigger.
Samsung will report Q1 CY2026 DS division revenue growth >20% QoQ with HBM revenue exceeding $3B
Strong Samsung = double-edged: validates memory demand but increases three-vendor competitive pressure on HBM ASPs.
Samsung will signal HBM pricing aggression (undercut or match MU/SKHynix HBM4 pricing) on Q1 earnings call
If Samsung starts price war → bear case for MU. Samsung historically uses scale to undercut and grab share in downturns.
SK Hynix will maintain or increase HBM revenue guidance for CY2026 (no downward revision)
Expected given sold-out supply. A surprise downward revision would be extremely bearish for the entire memory sector.
At least 2 of 4 hyperscalers (MSFT/GOOG/META/AMZN) will maintain or raise AI capex guidance in Q1 2026 earnings
If maintained → demand side intact. If 2+ cut → memory cycle thesis weakened regardless of supply dynamics.
MU Q3 FY2026 will achieve >78% non-GAAP GM (within 300bp of 81% guide)
If 81% materializes → E is still accelerating, 5.5x P/E was a screaming buy at $321. If <75% → peak margin thesis cracking.
MU Q4 FY2026 revenue guidance will be >$33B (maintaining or exceeding Q3 run-rate)
If Q4 > Q3 → supercycle intact. If Q4 < Q3 → sequential decline = cycle peaked, E downward revision begins.
The 'Rubin Ultra 4→2 die' rumor will be explicitly denied by NVIDIA or proven false by official specs at Computex/earnings
Expect no change to published specs. If confirmed true → HBM demand per rack reduced ~50%, material bear catalyst.
MU stock will trade below $300 at some point before June 25 (continued cycle-fear selloff)
If $300 breached on sentiment (no fundamental deterioration) → potential entry point. If breached on contract price decline → stay away.
MU stock will recover above $400 before June 25 (sentiment normalizes ahead of Q3 earnings)
Recovery before fundamentals are verified = speculative. Do not chase. Wait for contract price data.
Key Questions
- TrendForce Q2 CY2026 DRAM contract price forecast direction — sustained growth, plateau, or decline?
- Samsung Q1 CY2026 earnings: DRAM ASP realization vs guidance? HBM capacity utilization? Any aggressive pricing signals?
- SK Hynix Q1 CY2026 earnings: HBM4 yield update? NVIDIA allocation changes? Revenue mix shift?
- DRAM spot price trajectory through April — does the current softening stabilize or accelerate into a trend?
- Hyperscaler Q1 2026 earnings (MSFT/GOOG/META/AMZN in late April): AI capex guidance maintained, raised, or cut?
- MU Q3 FY2026 (June 25): Does 81% GM materialize? What does Q4 FY2026 guidance imply for FY2027 trajectory?
- Is the Rubin Ultra '4→2 die' rumor confirmed or denied at Computex or NVIDIA earnings?
- Channel inventory levels — do DRAM/NAND weeks-of-supply rise from 2-3 weeks toward 4-6 weeks (normalization)?