Micron $MU and SanDisk $SNDK Fell on the Apple-CXMT Report: The Market Priced Volume in a Price-Driven Cycle
The memory complex lost several billion dollars of market value on Monday on the strength of a Weibo account. Reports circulating over the weekend, traced back to a pseudonymous poster called Mobile Chip Expert and relayed through Wccftech and AppleInsider, said the Trump administration may permit Apple to source DRAM from ChangXin Memory Technologies and NAND flash from Yangtze Memory Technologies, with a decision communicated after a Trump-Xi meeting expected around September 24. SanDisk traded down as much as 9% to $1,458.29. Micron and Western Digital fell around 7%. Eight days earlier, Commerce Secretary Howard Lutnick had told the Wall Street Journal that it is not great for American companies to be using Chinese memory, and the same names rallied hard on the session. A cabinet secretary on the record moved the group less than an anonymous leaker did. That is worth sitting with before anything else, because it says the complex is currently being priced on the policy narrative rather than on bits.
The trade being expressed is a volume trade. Apple diverts DRAM and NAND purchases to Chinese suppliers, Micron and SanDisk lose Apple’s business, revenue falls. That arithmetic works in a market with slack. It does not work in this one. DRAM average selling prices this year have risen by a multiple of the increase in bits shipped, which is the entire structure of the current cycle: the shortage is a supply-exit story, not a demand story, and the industry is being paid on price rather than on units. In that configuration a lost customer is not lost revenue. It is reallocated supply. Every bit Apple declines to buy from Micron gets sold to somebody else at a spot price that has gone vertical, and the buyer on the other side is bidding into the same shortage Apple is trying to escape. Customer defection costs a supplier money when the supplier has capacity it cannot fill. Nobody in DRAM or NAND has capacity it cannot fill.
The specifics make the point sharper, because the supply being discussed does not exist in the form the sell-off assumes. CXMT maxed out its production this year and is prioritising domestic Chinese customers, with capacity expansion targeted at 2028. YMTC has allocated its newest-generation NAND to domestic consumption, directing it toward Android handsets, electric vehicles and Lenovo notebooks. Neither company has spare output waiting for an American buyer to be granted permission. So a Washington approval does not add bits to the world. It changes which buyer stands in front of a fixed queue. Bits routed to iPhones assembled for the China market come out of the pool currently feeding Chinese Android OEMs and EV makers, and those buyers then go bid for Samsung, SK Hynix and Micron supply instead. The approval scenario is close to globally demand-neutral. It shuffles counterparties.
Then there is the qualification gap, which is where the near-term threat mostly evaporates. Lynx Equity Research’s KC Rajkumar, calling the move an overreaction, found CXMT qualified for exactly one low-volume Mac SKU and not qualified for iPhones at all, with lpDDR5x yields poor enough to make meaningful Apple-scale supply structurally implausible regardless of political cover. On the NAND side he found Apple has not qualified YMTC’s flash for any product. Qualification is not a formality that follows a diplomatic gesture by a few weeks. It is a multi-quarter engineering process with reliability data behind it, and the export rules make it harder rather than easier: American firms may buy off-the-shelf parts from CXMT but may not share specifications or transfer technology, which is precisely what a custom memory configuration requires. Apple’s memory is tuned to its silicon. Using standard CXMT parts means redesigning portions of the China-market product line around a commodity component. That is a real engineering commitment, undertaken in exchange for a concession that arrives as a summit gift and can be withdrawn as a summit punishment at the next meeting. Both companies also remain on the Pentagon’s Section 1260H list of firms with alleged ties to China’s military-industrial base, which does not bar the purchase but does mean the political exposure travels with every unit shipped. You cannot build a bill of materials on that.
The single most informative detail in the reporting got almost no attention: CXMT has reportedly declined to offer Apple any price concession. Consider what that requires. A state-subsidised producer, fourth in the world in DRAM at roughly a tenth of global output, sitting on a designation that makes it commercially radioactive to its target customer, is being approached by the largest consumer-electronics buyer on the planet, and is not discounting. That is not the behaviour of a challenger buying share. It is the behaviour of a supplier who knows the clearing price and knows it does not need this account. It also disposes of the second leg of the bear case, the one arguing that a China option strengthens Apple’s hand in negotiations with Micron. There is no leverage to extract from an alternative supplier who is sold out and charging market. The alternative has to be cheaper or spare to be leverage, and CXMT is neither.
The dispersion within the sell-off is where the mispricing is easiest to see. SanDisk fell hardest, roughly two points more than Micron, despite being the name whose revenue is most insulated from any single customer’s sourcing decision. Its NBM agreements cover something on the order of half of fiscal 2027 bits and around two-thirds of fiscal 2028 across eight customers at floor pricing near $0.29 per gigabyte. Contracted floor-price revenue is by construction the exposure that reprices least when one buyer changes its mind. Micron, as the dominant supplier of the high-density lpDDR5x Apple actually uses, carries the more direct read-through, and fell less. The group traded as a single instrument on a China headline, which is what happens when positioning is crowded and the marginal holder is there for beta rather than for the contracts. It also happened into Nvidia’s print, with broad semiconductor profit-taking running underneath, so some portion of the move is not about Apple at all.
What would change the analysis is not a September announcement. It is CXMT discounting.