SNDK
SanDisk
Summary
What they do:
Develop, manufacture, and sell NAND flash storage devices and solutions — SSDs for data centers, edge computing, and consumer markets — spun off from Western Digital in February 2025 as a pure-play NAND flash company with ~30% global market share in flash storage.
Why they matter:
NAND flash is the storage backbone of AI infrastructure. Every AI training cluster needs terabytes of fast storage for model checkpoints, training data, and inference caching. SanDisk is the second-largest NAND player globally (behind Samsung), riding an extraordinary NAND supercycle — Q2 FY2026 revenue of $3.03B was up 61% YoY with 51% gross margins, and Q3 is guided at $4.4–4.8B. Joined the Nasdaq-100 on April 20, 2026.
Recent performance:
Q2 FY2026 (Jan quarter) revenue $3.03B, up 61% YoY, 31% QoQ. EPS $6.20. Gross margin 51.1% (51.9% ex $24M startup costs). Q3 guided $4.4–4.8B revenue, gross margin 65–67%, EPS $12–14. Management expects market "more undersupplied" in Q3 than Q2. FY2026 revenue expected ~$16B (+118% YoY). Stock ~$906, market cap ~$126B. Stock up 30x in one year. Reports Q3 April 30.
Our Verdict
The pure-play NAND flash storage company riding an extraordinary supercycle with 61% revenue growth, 51% gross margins, and AI data center demand — but at ~8x forward revenue and ~$126B market cap after a 30x one-year run, NAND cycle peak is the central risk.
Structural trends
Structural
73
/ 100
Moat
6/10
One of 5 global NAND manufacturers, Kioxia JV access, BiCS8 technology — but NAND is cyclical commodity
AI Exp.AI Exposure
Embedded~25% AI
Play Type
EstablishedAI Growth
~40%
Rel. Value
24
PREMIUMSanDisk was spun off from Western Digital on February 21, 2025, becoming an independent pure-play NAND flash company. WDC distributed 80.1% of its SanDisk stake to shareholders, and WDC subsequently divested most of its remaining 19.9% through a secondary offering exchanged for WDC debt. SanDisk is now fully independent with its own balance sheet, management team, and strategic direction.
The company operates across three end markets. Edge ($1.678B, ~55% of Q2 revenue, up 21% QoQ) serves enterprise and consumer SSDs, flash memory modules, and embedded storage — demand "meaningfully exceeded supply" as AI PC adoption and replacement cycles drove richer configurations. Consumer ($907M, ~30%, up 39% QoQ, 50% YoY) sells retail flash drives, memory cards, and portable SSDs — mix shifted toward premium products; new Extreme Fit USB-C and Optimus SSD rebrand launched. Datacenter ($440M, ~15%, up 64% QoQ) provides enterprise SSDs for cloud and hyperscaler storage — management expects data center to become the largest NAND market in 2026 for the first time, and sees "meaningful" near- and long-term growth ahead. Enterprise SSD bits are in the "high teens" percentage of total bits shipped.
The NAND supercycle is the story — but management is now framing it as a "structural evolution" rather than a cycle. After a brutal downturn in 2023–2024 that drove NAND prices to record lows and forced production cuts, the industry has entered an extraordinary upcycle. AI workloads demand massive storage — inference in particular is "driving a meaningful increase in NAND content per deployment." Data center exabyte growth forecasts have been revised up three times: from mid-20s%, to mid-40s%, to now high-60% for 2026. Additionally, NVIDIA's KV cache architecture represents potential incremental demand of ~75–100 exabytes in 2027, doubling in 2028 — none of which is in current demand forecasts. Management sees customer demand "well above supply beyond calendar year 2026." SanDisk's Q2 FY2026 revenue of $3.03B was up 61% YoY with gross margins surging to 51.1%, and Q3 guidance implies a further step to 65–67% gross margins.
Management is actively pursuing multiyear long-term agreements (LTAs) with prepayment components to replace the traditional quarterly pricing model. One LTA has been signed and closed (terms undisclosed); several more are in the pipeline. CEO Goeckeler frames this as customers seeking "supply certainty" years out (conversations extend to 2029–2030) and argues the shift toward data center as the primary NAND market fundamentally changes industry dynamics — data center NAND is "not commodity" but a "highly strategic product" requiring specific performance characteristics.
The BiCS8 technology transition is the product driver. CEO called BiCS8 a "fantastic node" — featuring a 2-terabit die with strong QLC performance. BiCS8 is expected to reach majority of production by end of FY2026. The PCIe Gen5 high-performance TLC drive has been qualified at a second hyperscaler, with additional qualifications on track. BiCS8 TLC solutions to follow. The BiCS8 QLC storage-class product, code-named "Stargate," is advancing with two major hyperscalers and expected to ship for revenue within the next several quarters. Management is also developing "high bandwidth flash" (HBF) — a rearchitected NAND die plus custom controller designed for AI compute attachment — with active customer engagement and design work underway.
Supply Chain Dependencies
Upstream Suppliers
The Catch
NAND flash is the most cyclical subsector in semiconductors — and management is making the boldest counter-argument in NAND history. SanDisk's 65–67% guided gross margins are not just peak; they are unprecedented for any NAND company in any cycle. Management's structural thesis rests on: (1) data center becoming the largest NAND market, (2) LTAs with prepayments replacing quarterly auctions, (3) supply discipline at mid-to-high-teens bit growth, and (4) customer demand extending to 2029–2030. The thesis is internally consistent and the demand evidence (high-60% DC exabyte growth, KV cache, supply shortage through CY2026+) is real. But every prior NAND "this time is different" argument (2017–2018, 2021–2022) proved wrong. One signed LTA does not validate a structural shift. At ~$126B market cap, SanDisk is priced for the structural thesis to be correct. If it is just the best cycle ever — and cycles end — the correction will be proportional to the unprecedented margins.
If They Win
If the structural thesis holds — LTAs scale across multiple hyperscalers with prepayments, KV cache adds 75–100+ exabytes of incremental demand in 2027, HBF opens a new product category, and data center sustains as 25%+ of revenue — SanDisk grows into a $22B+ revenue company with 50%+ sustainable margins and a mid-cycle floor well above 35%. At that scale, with reduced cyclicality and predictable multi-year revenue, SanDisk deserves a $150B+ valuation as the pure-play NAND leader in the AI era. But this requires NAND to break its historical cyclical pattern at the exact moment margins are at all-time highs — which is when every prior "this time is different" argument has been made.
Others in Memory for the Server
Not financial advice. All scores generated via AI algorithms using public data.