SanDisk joins the S&P 100 on Sept. 21. Its own July shows what forced flows do

On September 21, index funds tracking the S&P 100 will be obliged to buy SanDisk. They will not be consulted about the price.

S&P Dow Jones Indices confirmed the change on September 4. SanDisk joins the index alongside Dell Technologies, Palo Alto Networks and Arista Networks, replacing Honeywell Aerospace, Nike, Simon Property Group and Colgate-Palmolive. The stock closed that session at $1,740.00, up 11.9%, its high for the day.

It has since given all of it back. SanDisk closed Friday at $1,633.35, down 3.5% on the day and 6.1% below where it finished on the announcement. The mechanical buyers have not arrived yet, and the discretionary ones have been leaving.

Market Brief

Key Takeaways

5 Points30s Read

1. 01What happens–SanDisk joins the S&P 100 before trading opens on September 21, alongside Dell, Palo Alto Networks and Arista Networks. Index funds must buy regardless of price.
2. 02The pop is gone–The stock closed at $1,740.00 on the September 4 announcement, up 11.9%. It closed Friday at $1,633.35, 6.1% below that.
3. 03The quarter–SanDisk is up 493% year-to-date and down 28.2% this quarter. Its highest close of the quarter was July 1, the first session.
4. 04The round trip–From $2,273.73 at the end of June it fell 55.3% to $1,015.89 by July 29, then rallied 60.8% off that low.
5. 05It has done this before–In July a leveraged fund was margin-called out of a book in which SanDisk was 28.0% of $20.2 billion. Forced selling moved the stock ~47% that month with no change in fundamentals.

What index inclusion actually does

Index membership is not an endorsement. It is a rule.

Funds that track the S&P 100 hold what the index holds, in the weights the index specifies, because that is what they promised investors. When a constituent changes, those funds transact regardless of whether the price looks sensible on the day. That is the defining feature of passive flow: it is price-insensitive by construction.

For most additions this is a modest, one-off event. For SanDisk it is worth more attention, because this is not a modest, one-off stock. It has moved more violently in 2026 than almost anything else of comparable size, and it is arriving in the index at a specific and awkward moment in that trajectory.

The composition change is itself a small piece of market history. A household-products company, a footwear brand, a mall operator and an aerospace business make way for a memory maker, a server builder, a network vendor and a security company. Four of the largest hundred US companies are being replaced by four suppliers to the AI buildout.

But the four arrivals are not in the same condition, and TECHi’s own numbers show it. In our ranking of 41 AI stocks this quarter, Dell placed sixth with a 31.5% gain and Arista placed eighth at 17.5%. Both are joining the index while working. SanDisk placed 36th.

That is the oddity at the centre of this. Index committees look at size, liquidity and sector representation, not momentum — and SanDisk qualifies on size precisely because of a first half it has spent the third quarter handing back. It is entering the S&P 100 on the strength of a valuation it no longer has.

The quarter that has already happened

Here is what the buyers are being marched into.

SanDisk is up 493% year-to-date. It is also down 28.2% this quarter. Both are true, and the distance between them is the story.

The path is worth tracing precisely, because it is unusual. SanDisk’s highest close of the entire quarter was July 1, the first session, at $2,032.22. It had ended June at $2,273.73. By July 29 it closed at $1,015.89 — a fall of 55.3% from the quarter’s starting price in four weeks.

Then it went back up. From that low the stock has rallied 60.8%, which is how a name can be down 28% for the quarter and still look, on a chart, like it is recovering. Both halves of that round trip happened inside twelve weeks, in a company about to be one of the hundred largest listed in America.

The name that led the S&P 500 in the first half has spent the third quarter near the bottom of the AI complex, and it is not alone in that. Across the 41 names we measured, the rank correlation between first-half and third-quarter returns was −0.53: the better a stock did in the first six months, the worse it tended to do in the next three. Micron fell 15.5% after rising 266%. Intel fell 26.3% after rising 255%. Arm fell 25.3% after rising 209%.

SanDisk is simply the most extreme case of a pattern that ran through the whole memory and semiconductor complex. It rose the most, and it fell the hardest.

July was the same mechanic, running the other way

The reason to take forced flows seriously in this particular stock is that SanDisk has already demonstrated what they do to it.

In July, a fund built entirely around the AI trade was margin-called out of its book. TECHi went through its regulatory filings and found that SanDisk was its single largest disclosed position — 28.0% of a $20.2 billion portfolio, 2,495,344 shares. When the lenders called, that position was not sold because anyone had changed their mind about NAND pricing. It was sold because it had to be.

SanDisk fell 46.6% that month.

That is the symmetry worth holding onto. Forced selling moved this stock nearly 50% in four weeks with no change in fundamentals. Forced buying is now scheduled, with a date on it, and the market has known about it since September 4. The difference is that everyone can see this one coming, which is usually enough to blunt it — the announcement pop has already been arbitraged away, and hedge fund ownership more than doubled before the change takes effect.

Anticipated flows tend to be priced. Unanticipated ones are what break things.

The more durable change is not the trade on the day but what comes after it. Index membership installs a permanent block of holders who never sell for valuation reasons and never buy for them either. That cuts both ways for a stock like this one. It adds a floor of demand that does not flinch during a 50% drawdown, which is stabilising. It also removes shares from the pool that actually sets the price, which means the remaining free float carries more of the volatility — and this is a company whose free float has just demonstrated what it can do when it is the only thing moving.

A smaller effective float in a commodity-cycle stock with a leveraged shareholder base is not obviously a calmer arrangement. It is a different one.

The only variable that actually matters

None of this is the investment case. The investment case is NAND contract pricing, and it always has been.

SanDisk’s earnings transformation came from AI-driven premium pricing on high-capacity storage, which is why the stock went up 30-fold before it went down by half. TECHi laid out the mechanics when the $42 billion AI backlog first started rewriting the NAND cycle, and the question then is the question now: what price does that contracted backlog convert at?

Reported terms put a floor near $0.29 per gigabyte on recent agreements, broadly in line with the quarter’s average selling prices. A floor is genuinely valuable in a commodity business — it caps the downside on contracted volume. It also tells you where the argument sits. Bulls are underwriting pricing above that floor for years; bears are underwriting reversion toward it as supply responds. The entire spread between the most optimistic and most pessimistic targets on the street comes from that single variable.

It is also worth being precise about what a contract floor protects. It sets a minimum price on volume that has already been contracted. It does not guarantee the volume, it does not cover capacity that comes online later, and it does not stop a customer from negotiating differently at renewal. In a business where the marginal bit of supply sets the marginal price, a floor is a useful shock absorber rather than a moat. TECHi has made this point through two cycles of SanDisk coverage, including when the market was treating a $2,300 target as conservative.

Index inclusion does not move any of it. Passive money does not have an opinion about gigabyte pricing.

What we wrote when this was going right

It is worth being straight about the trajectory of our own coverage, because it maps the cycle.

In July, when SanDisk had just finished as the best stock in the S&P 500, TECHi published a piece arguing that the hard part came next. The stock is down 28% since the quarter began. That is not a victory lap — the piece did not predict the size of the drawdown, and a 493% year-to-date gain means anyone who owned it through all of this is still comfortably ahead.

But the structure of the risk was visible then and is visible now. A stock that rose this far this fast, in a commodity industry, held by leveraged funds, trades on flows as much as on fundamentals. September 21 adds one more flow to the list.

What would change the read

Three things, in order of how much they matter.

The first is contract pricing. If NAND agreements continue to reprice above the reported floor, the earnings base supporting the current valuation is real and the third quarter’s decline was a de-rating rather than a warning. If pricing reverts, no index membership helps.

The second is who owns it. Hedge fund ownership more than doubled into the inclusion, which means a meaningful share of the register is positioned for an event that happens on a known date. Positioning built for a catalyst tends to unwind after the catalyst, whichever way the price went.

The third is supply. Every previous NAND upcycle ended the same way, with capacity arriving to meet the price. Nothing about AI demand suspends that, and the manufacturers that got rich on this cycle are the ones deciding when it ends.

On September 21 the index funds will buy what they are told to buy. Everything that determines whether that was a good price is decided somewhere else.