SanDisk (SNDK) Rallied 11.9% Into an S&P 100 Add That Forces Almost Nobody to Buy

SanDisk closed Friday at $1,740, up 11.9%, a third consecutive advance. The wire coverage split the credit between two things. Nvidia confirmed late Thursday that it will pay $12.9 billion for Hugging Face, which put the whole AI complex bid, and Dell’s COO had spent part of the earnings call earlier in the week saying the constraint on AI server supply is memory, DRAM first and NAND second. Underneath both, S&P Dow Jones Indices had announced in its quarterly rebalancing that SanDisk joins the S&P 100 on September 21, alongside Dell, Palo Alto Networks and Arista Networks.

The index line is the one being traded. It is also the one that does the least.

Every S&P 100 constituent is already an S&P 500 constituent. SanDisk has been in the S&P 500 since November 28 of last year, when it replaced Interpublic ahead of the Omnicom acquisition. Any fund tracking the broad index already owns the stock at its float-adjusted weight and has owned it for nine months. The September 21 event creates no obligation for those funds at all. The only vehicles that have to transact are the products benchmarked specifically to the S&P 100, and the assets sitting in those products are a small fraction of what tracks the 500. Against SanDisk’s float and its recent daily volume, the mechanical demand arriving on the effective date is close to a rounding error.

Compare it to the November add, which was a real flow event. The stock rose 13.3% in the regular session and another 9% after hours, and behind that move was compulsory buying from every S&P 500 tracker in existence. Friday’s 11.9% looks like the same trade and is not. What moved the stock is positioning ahead of a date, by participants who know that index adds have historically been front-run and who are pricing the anticipation rather than the flow. That works until the last marginal buyer has already bought, which is generally somewhere before the effective date rather than after it. Post-inclusion drift on names that have already run into the event tends to be unkind.

The deletion list is the more interesting disclosure. Honeywell Aerospace, Nike, Simon Property Group and Colgate-Palmolive come out. SanDisk, Dell, Palo Alto Networks and Arista go in. That is a mega-cap index swapping an aerospace supplier, an apparel brand, a mall REIT and a toothpaste company for two memory-and-systems names and two networking names. The S&P 100 is a slow index by construction, which is what makes the composition change worth reading. It is the concentration of the AI infrastructure trade being written into the benchmark that pension consultants and structured product desks use as a proxy for the American mega-cap. The rotation is no longer a market opinion at that point. It is the index definition.

SanDisk’s own ladder is the compressed version of the same thing. It entered the S&P SmallCap 600 in February 2025 when Western Digital spun it out, replacing Leslie’s. It entered the S&P 500 nine months later. It enters the S&P 100 nineteen months after listing as an independent company. Very few businesses traverse that in a single cycle, and the ones that do are usually being repriced by an external input rather than by anything they did operationally. Here the external input is NAND pricing.

That is the part worth separating from the index news. The Dell comment matters more than the Hugging Face deal, because Hugging Face is a model hub that buys no flash and the read-through to SanDisk is thematic. Dell’s remark is a buyer describing what it cannot get. The valuation reported around Friday’s move, roughly 23 times trailing earnings and about 8 times forward, is the market saying the same thing in a different notation. An 8x forward multiple on a NAND business does not mean cheap. It means the forward estimate embeds a very steep earnings ramp that the market does not fully believe will hold, which is the standard way a cyclical trades at the point of maximum reported strength.

The structural argument for why this cycle might behave differently rests on the asset-light position, roughly $674 million of net property, plant and equipment carrying a business with $20 billion of annual revenue, with the Kioxia joint venture holding the capital intensity. The contracted floor pricing signed under the new business model does the rest, capping the upside in exchange for holding the downside. That is a genuinely different NAND exposure than the one investors were burned by in prior cycles. It is not an index story and it will not be resolved on September 21.

The decision-relevant number over the next fortnight is not the inclusion. It is whether NAND contract pricing for the September quarter confirms what Dell described. If it does, the index date is noise inside a fundamental move. If it does not, the anticipation flow will have been the entire thing, and it expires on a calendar.