Back to News
Market Impact: 0.3

Prediction: This Will Be Sandisk's Stock Price by Mid-2027 (Hint: It Implies a Big Move)

Artificial IntelligenceTechnology & InnovationCredit & Bond MarketsCompany FundamentalsCorporate EarningsAnalyst Estimates

Sandisk (SNDK) is up 570% YTD in 2026 on demand for enterprise storage supporting AI infrastructure. The company is expanding enterprise SSD/NAND offerings, with products shipping this quarter based on a new Stargate controller and new BiCS10 3D NAND that boosts bit density 59% and performance 33% while improving power efficiency. Q3 FY2026 revenue surged 251% to $5.9B and non-GAAP EPS rose to $23.41 vs a $0.30 loss prior year, while Wall Street forecasts revenue up 155% to ~$50B in FY2027, implying substantial upside if valuation compresses from 18x sales to 9x.

Analysis

The core market mechanism is not just “AI demand for memory,” but a rapid re-rating of a commodity supplier that has temporarily gained quasi-oligopolistic pricing power through qualification, controller differentiation, and supply discipline. That creates a near-term winner in SNDK, but it also raises the probability of downstream margin pressure at hyperscalers and server OEMs if storage remains scarce longer than planned. On a relative basis, the more durable beneficiaries may be the companies that can lock in multi-quarter allocations and pass through cost inflation; the losers are the buyers forced into spot exposure or inventory pre-buys.

The next 1-3 months are about guidance quality, not the stock chart. If management signals that contract pricing is still rising and lead times remain extended, momentum can persist despite the multiple; if sequential pricing flattens, the stock’s very high expectations make it vulnerable to a sharp de-rating. The 6-18 month risk is classic memory-cycle supply response: new fabs and line conversions can arrive just as customers have already stocked up, creating an earnings air pocket well before the market notices it in headline revenue.

Consensus seems to be underestimating how quickly “scarcity premium” can disappear once buyers have enough inventory visibility. The article’s revenue path implicitly assumes the current shortage is structurally durable; that is possible, but history says the first sign of normalization is usually in spot pricing and order cadence, not in reported revenue. A second-order contrarian view: AI storage intensity may be less elastic than GPU demand, so if inference architectures shift toward more compression/caching, NAND could be the easiest part of the AI stack to over-earn on and then mean-revert fastest.