Our SanDisk Call: $SNDK reports in a few hours. The revenue beat is expected. Margin decides the print. Street revenue consensus is $8.42B, already above management’s $7.75B - $8.25B guide. But consensus EPS of $34.67 still appears to assume gross margin near 80%, around the midpoint of management’s 79% to 81% range. Those assumptions do not sit comfortably together. The Street has moved above management on revenue while remaining anchored to management on margin. Our view is that NAND pricing changed too quickly for the April 30 guide to capture, and gross margin is where that change should appear. Our call: - SanDisk fiscal Q4 non-GAAP gross margin above 81%, exceeding the top of management’s guide. - Central case: approximately 84%. Management already underestimated the cycle once. Last quarter, SanDisk guided gross margin to 65% to 67%. It printed 78.4%. That was more than 11 points above the top of its own range, driven by stronger pricing and a faster shift toward higher-value customers. When asked why the next guide still assumed more modest pricing improvement, management said it “pays to be a bit conservative” early in a fast-moving quarter. We think it was conservative again. The strongest evidence comes from Kioxia. Kioxia jointly operates the same Japanese NAND manufacturing system as SanDisk and has already reported the same April-to-June period. Its blended selling prices rose approximately 70% sequentially. Gross margin reached 80%. Kioxia is not a direct margin comparison. Its mix, customers and accounting differ. But it tells us something important: The shared manufacturing system experienced far stronger pricing economics than SanDisk’s guide implied. SK Hynix provided a second confirmation, reporting NAND pricing up in the mid-50% range despite the dampening effect of longer-term contracts. The debate is therefore no longer whether NAND pricing improved. It is how much of that repricing reached SanDisk’s realised revenue and product mix during the quarter. How we made our call: We hold the prior cost structure broadly constant and vary one input: how much of the observed NAND repricing appears in realised revenue and mix. We do not assume full capture. Consumer and retail pricing resets more slowly than enterprise contracts. Inventory is recognised at historical cost. Realised pricing generally trails spot markets. Even limited capture gets gross margin above 81%. Stronger, but still incomplete, capture points toward the mid-80s. That is why 84% is our central case. The earnings sensitivity is meaningful. Using frozen Street revenue, share-count and tax assumptions, each gross-margin point is worth approximately $0.47 of EPS. Consensus EPS of $34.67 appears to sit near 80% margin. At 84%, EPS moves to roughly $36.60, around 5% above consensus. That is meaningful upside on the quarter’s decisive variable, although it is not by itself a rerating thesis. What could make us wrong? Three things: Slower repricing in consumer and retail channels. Inventory costs that have not yet caught up with current wafer economics. Manufacturing or node-transition costs absorbing more of the pricing benefit than we expect. If gross margin prints at 81% or below, the call is wrong. If it prints above 81%, management guided conservatively again, NAND pricing is reaching reported earnings faster than consensus expects, and fiscal 2027 estimates should move higher. Btw: I formed our call using Tessara's new research workspace (link in bio) Every material claim links back to its original source, including the Kioxia print, SanDisk management commentary and the assumptions behind our margin bridge. You can inspect the evidence, challenge the assumptions and build your own view rather than simply accept ours.