The memory trade splits in two. One number decides which SanDisk you own

SanDisk is the cleanest test case for a question the whole memory complex is now asking: are these contracted utilities or cyclical commodities dressed up as growth?

The bull number is real and it is the whole thesis. As a pure-play NAND and enterprise-SSD company post-split from Western Digital, SanDisk carries a fortress balance sheet: $3.74B cash against just $207M of total debt, ROE of 39.30%, and a spot in the Nasdaq-100 since April 2026. Consensus puts FY2026 revenue at $19.84B, up 169.70% year over year, with EPS of $66.10, and the sell-side models FY2027 at $49.87B revenue and $212.95 EPS. On those forwards the stock trades near 6.7x FY27 earnings. If those numbers hold, that is a utility priced like a cyclical.

The "if" is doing a lot of work, and the market knows it. The stock peaked at $2,335 in June, fell nearly 50% to about $1,016 in late July, and rebounded above $1,400 into earnings. The options market priced a single-session move of plus or minus 13.1%, and the analyst target spread runs from $1,000 on the bear side to $3,250 on the bull side. That is not disagreement about a quarter, it is disagreement about the business model.

The bridge between the two cases is the contract, not the chart. The thing that would turn SanDisk from commodity NAND into a utility is the shift to multi-year NBMs and LTAs with tier-1 cloud buyers, backed by upfront prepayments that lock volume through 2027 and 2028. That is the same take-or-pay mechanism that de-risked the earlier Micron story. The bear case attacks exactly that seam: non-binding agreements can be renegotiated if spot NAND collapses, and Chinese fabs building capacity could break the pricing.

Why it matters (allocator lens): tie this to yesterday's tape, where AMD printed a record beat and still fell 5%. The market is no longer paying for memory growth on faith. It is paying for the durability of the revenue. SanDisk's valuation gap versus its own forwards is not a free lunch, it is the price of the contract-durability question being unresolved.

So-what for ABC. This is an Alpha-sleeve problem, not a Beta one, given a plus-or-minus 13.1% implied move and a 50% round trip in six weeks. If you are underwriting the name, the variable to track is not the stock, it is contract quality: the mix of prepaid, locked-volume NBMs and whether hyperscalers honor them into a softer spot market. Size it as the high-volatility single-name it is, and let the contract disclosures, not the price target range, move your conviction. Neutral, because the same fact set genuinely supports a $1,000 and a $3,250 view.

Takeaway: SanDisk is either a mispriced memory utility or a cyclical at the top, and the tell is the contracts, not the candles. Underwrite the take-or-pay, not the target.

Analytics & education, not advice. DYOR.