Investment case
Sandisk is a leading NAND flash supplier with vertically integrated technology, Flash Ventures manufacturing, enterprise SSD products and a rapidly increasing datacenter mix. AI inference, KV-cache offload, RAG and large data estates are increasing the importance of high-capacity flash in AI infrastructure.
> The PowerFund thesis is that Sandisk's New Business Model agreements and datacenter mix can structurally improve the quality of NAND earnings without eliminating cyclicality. The September evidence raises confidence that the downside is better protected than in prior NAND cycles, but the investment still requires a meaningful expected-return premium because roughly one-third of future business remains exposed to prevailing market prices, supply can respond, and current margins are extraordinary.
New September evidence
- Management said datacenter has become more than half of the NAND market, changing customer behavior and demand visibility.
- Roughly two-thirds of FY2028 bit supply is covered by NBM agreements; management agreed that a simplified downside framing of that NBM-covered portion at about an 80% gross-margin floor is directionally reasonable.
- NBM contracts can run up to five years with detailed volume schedules and financial guarantees from third-party institutions if customers fail to perform.
- Some NBM customers have returned for additional volumes outside existing agreements, a positive signal on underlying demand.
- TrendForce's 21 Sep update said North American CSPs raised enterprise-SSD demand forecasts and that 4Q26 orders could exceed 3Q26, with QLC a material demand driver and pricing still moving higher in enterprise SSDs.
- Offset: consumer/spot NAND remains weaker, and prospective supply competition from Chinese suppliers remains an important long-term risk.
Updated valuation scenarios
These are PowerFund scenarios, not company guidance. They deliberately normalize earnings below peak quarterly run-rates and do not simply capitalize management's ~80% long-term gross-margin target.
24 months
| Case | Weight | Core assumptions | Implied value |
|---|
| Bear | 30% | ~$60 normalized EPS; 15x P/E; NBM cushions but does not prevent a material NAND downturn | $900 |
| Base | 50% | ~$150 normalized EPS; 18x P/E; enterprise SSD/QLC growth plus NBM supports structurally higher through-cycle earnings | $2,700 |
| Bull | 20% | ~$205 normalized EPS; ~22x P/E; AI-storage demand, disciplined supply and NBM economics remain unusually strong | $4,500 |
Probability-weighted working value: ~$2,520.
At $1,887.04: weighted 24-month CAGR ~15.6%; bear-case downside ~52%; base-case upside ~43%.
60 months
| Case | Weight | Core assumptions | Implied value |
|---|
| Bear | 30% | ~$80 normalized EPS; 14x P/E | $1,120 |
| Base | 50% | ~$190 normalized EPS; 18x P/E | $3,420 |
| Bull | 20% | ~$300 normalized EPS; 22x P/E | $6,600 |
Probability-weighted working value: ~$3,366, or roughly 12.3% annualized from $1,887.04.
Entry framework
The prior single $1,300–1,350 trigger is retired as the minimum acceptable entry. It is better treated as a deep-dislocation zone.
- >$2,000: patience unless new evidence raises fair value again.
- ~$1,850–1,950: legitimate small starter zone if thesis is intact; weighted 24-month CAGR is roughly mid-teens.
- ~$1,700–1,800: attractive second-tranche territory; weighted CAGR moves toward ~18–22%.
- ~$1,550–1,650: aggressive dislocation territory if the thesis remains intact; preserve capacity to add rather than assuming this level must occur.
The purpose is not to predict the bottom. Capital should increase as expected return improves.