Can Sandisk Become a $1 Trillion Company in 5 Years?

Sandisk reported 97% sequential revenue growth to $5.95 billion in fiscal Q3 and guided to $8 billion at the midpoint for fiscal Q4, underscoring extraordinary AI-driven demand. The company says it is sold out of memory products through 2026 and is already seeing strong demand for 2027 output, while net profit margin expanded to more than 60%. The article cites Melius Research as expecting memory demand to stay elevated through the end of the decade, supporting the bull case for continued upside.

Analysis

The market is increasingly treating NAND as an AI bottleneck rather than a commodity input, which changes the earnings durability lens for SNDK and, by extension, MU. The second-order effect is that memory vendors gain pricing power precisely because hyperscalers cannot defer purchases without jeopardizing near-term deployment schedules; that should support margins for multiple quarters even if unit demand moderates. The bigger winner may be the entire AI infrastructure supply chain: storage, networking, and server OEMs all benefit from faster capex conversion, but memory gets the cleanest direct uplift.

What matters most is not the headline demand curve but the duration of pre-buys. If customers are already locking 2027 output, the cycle extends from a normal 12-18 month inventory spike into a multi-year capacity rationing event, which historically allows producers to print outsized free cash flow before new supply arrives. That said, the marginal risk is self-inflicted: if management teams overcommit capex too aggressively, the current scarcity premium can unwind fast once wafer starts and packaging bottlenecks normalize in 2027-2028.

The contrarian view is that the move may be too linear from here. At this size, SNDK is priced like a quasi-monopoly on a peak-earnings run-rate, but memory has a long history of mean reversion once customers finish front-loading. The key tell over the next 1-3 quarters will be whether order visibility converts into sustained ASP discipline, or whether the market is simply pulling forward a few years of returns into one tape-driven rerating. NVDA is a secondary beneficiary, but its upside from memory tightness is smaller and more diffuse; the cleaner expression remains the component supplier with direct allocation power.

Key Decisions for Investors

• Stay long SNDK on pullbacks over the next 4-8 weeks, but use a disciplined trailing stop: the trade works as long as 2027 allocation tightens; if commentary shifts to capacity expansion or customer pushback, de-risk quickly because the multiple will compress before fundamentals do.
• Pair long SNDK / short a basket of memory-adjacent hardware names with less pricing power over the next 1-2 quarters; the relative trade should benefit from continued scarcity premium, but cap upside at 20-30% and reassess if capex announcements accelerate.
• Add MU as a lower-beta beneficiary for 6-12 months, but favor options over common stock if chasing the rerate: long-dated calls provide exposure to continued AI memory scarcity while limiting damage if the market starts discounting a 2027 supply response.
• Avoid shorting the group outright; instead, if you want a contrarian hedge, express it via a 9-12 month call spread fade in SNDK after another strong leg higher, since the most likely reversal trigger is not demand collapse but investor realization that peak margins are being pulled forward.
• Monitor NVDA and hyperscaler capex prints as the leading catalyst set: if data center spending remains resilient into the next 2 quarters, memory tightness can persist; if capex pauses, memory stocks will likely de-rate first and fastest.