Investment Memo · Semiconductor Memory (DRAM / NAND / HBM) · Only Pure-Play Memory Stock on US Exchanges
Publish Date 2026-07-01 | Data as of: FQ3 FY2026 Earnings (Through 2026-05-28), Price Snapshot as of 2026-07-01 | Currency USD ($); Fiscal Year Sept–Aug
I. Investment Thesis on One Page (Tearsheet)
| Investment Rating | Hold (Cautious; extreme cycle peak, poor risk/reward after +700% rally) |
|---|---|
| 12-Month Target Price | $935 (probability-weighted) vs Current $1,064 → −12% |
| Upside / Downside | Upside $1,350 (+27%) / Downside $500 (−53%) |
| Risk Level | HIGH – Extreme cycle peak (record quarterly revenue $41.5B, gross margin 84.6%); severe mean reversion risk |
| Time Horizon | HBM4 ramp & pricing → Memory spot/contract prices → FQ4'26 ($50B guidance) delivery → Supply discipline |
| Position Sizing | 3–5% of personal portfolio max (strong cyclical, high vol; better to control size or use options instead of stock) |
| Liquidity/Concentration | Extremely liquid; AI crowded trade with sentiment and memory prices in sync, large intraday swings |
| Market Cap | TTM P/E | FQ3'26 Revenue / YoY | 1-Year Stock Performance |
|---|---|---|---|
| ≈ $1.22 trillion | ≈ 24x (peak earnings) | $41.5B / +346% | +700% |
| FQ3'26 Gross Margin | FQ3'26 EPS | FQ4'26 Revenue Guidance | |
| 84.6% (historic) | $24.67 (beat) | ≈ $50B | Sold out through 2026+ |
Note: Market cap/P/E are public market data as of 2026-07-01 ($1,064, ~1.145B shares). The 84.6% gross margin and $41.5B quarterly revenue are AI-HBM supercycle extremes, far above historical memory peaks (~55–60% gross margin) and not linear extrapolable. Target prices are modeled/projected.
1.1 Core Argument (Weighted Framework)
| Scenario | Weight | Key Assumptions | Target Price / Move |
|---|---|---|---|
| ▲ Bull | 30% | HBM structural repricing: AI demand permanently lifts memory profit margin midpoint; supercycle extends; FY27E EPS ~$90 × 15x | $1,350 +27% |
| = Base | 40% | Cycle consolidates at high then mild mean reversion; FY27E EPS ~$68 × 14x | $952 −11% |
| ▼ Bear | 30% | Severe mean reversion (supply comes back + demand cools); FY27E EPS ~$40 × 12.5x | $500 −53% |
| Weighted Expected | 100% | 0.30×1350 + 0.40×952 + 0.30×500 | $935 −12% |
Target prices are modeled, derived from FY27E normalized EPS × cycle P/E (not FY26 peak EPS, which would give absurdly high targets). The three scenarios ranging $500–$1,350 reflect the binary debate: "HBM structural repricing vs cyclical mean reversion."
1.2 Key Monitoring Indicators (KPIs to Track)
| Indicator | Current Value | Re-evaluation Threshold | Frequency |
|---|---|---|---|
| DRAM/NAND spot & contract prices | Supercycle highs | Continued up → bullish; turn down → bearish | Weekly/monthly |
| HBM4 ramp/pricing/yield | 2x HBM3E speed, yield above expectations | Maintain share & premium → bullish | Quarterly |
| Blended gross margin | 84.6% (peak) | Stay >60% → repricing validation; drop toward 40% → reversal | Quarterly |
| FQ4'26 revenue delivery | Guidance ~$50B | Beat/raise → bullish | Event-driven |
| Industry supply discipline | Three oligarchs restrained expansion | Maintain restraint → bullish; arms race → bearish | Ongoing |
| Anthropic / hyperscaler long-term contracts | Strategic supply + investment signed | More long-term contracts → demand anchor | Event-driven |
| Valuation vs normalized earnings | Peak P/E ~24x / current price | Implied P/E vs mid-cycle EPS | Ongoing |
1.3 Operational Thoughts (Reference Only, Not Recommendations)
Not holding: Already +700%, market cap >$1 trillion, at extreme cycle peak – no need to chase at current price. Wait for cycle to cool or a meaningful pullback, then build a small position in batches; cap at 3–5% of portfolio.
Already holding (low cost basis): Enjoy the supercycle but set a reduction discipline – watch memory prices and gross margin. Once prices turn down and inventories rise, take profits in stages. Use covered calls / OTM puts to hedge extreme volatility.
Wanting to short: Valuation logic works but timing is extremely difficult – supercycle and AI demand still strong, HBM sold out. Fighting the trend is costly; use put spreads to cap losses, avoid naked shorts.
Absolutely avoid: Extrapolating the 84% gross margin and $41.5B quarterly revenue as the new normal. Or going full position/margin to chase the extreme cycle top due to FOMO.
专注投资分析、市场洞察与资产配置。不追短期波动,只理解真正驱动长期回报的东西。