Sandisk Completes Eight-for-One Split as NAND Outlook Darkens

Eight for one. Sandisk’s stock split took effect on Nov. 19, and the shares began trading on the adjusted basis on Nov. 20, a change that divides the price by eight, multiplies the share count by the same factor and leaves the company’s market value untouched.

The split itself is arithmetic. What arrived beside it was not.

TrendForce cut its forecast for 2027 NAND flash prices the same day, citing supply additions running ahead of demand. The revision landed hours after the split took effect, which is awkward timing for a company whose pitch to new shareholders is that memory pricing has turned. A split invites retail buyers to look at the chart. The forecast tells them what the chart is likely to do next year.

Contract prices are the whole story in this business. NAND is sold on quarterly agreements, the spread between cost and price is narrow at the volumes Sandisk ships, and a move of a few percentage points in the contract price swings gross margin by more than most hardware investors expect. Sandisk was spun out of Western Digital in February 2025 and sells NAND flash and the solid-state drives built from it. Its shares re-rated through the year as AI servers absorbed high-capacity drives, and a split is the kind of step a board takes after a run rather than before one.

Much of the output comes from joint ventures with Kioxia in Japan, an arrangement that spreads the capital cost of a fabrication plant and also spreads the decision to add capacity across two boards. When both partners want volume in the same year, the market gets it.

Supply is where the forecast changed. Every major maker has been adding deposition layers and wafer capacity, and the capital spending announced through 2025 arrives as output in 2026 and 2027 rather than immediately. Capacity ordered in a shortage usually ships into a surplus.

Demand is narrower than the headline suggests. Enterprise drives for AI clusters are bought by a handful of customers whose orders move in blocks, while the personal computer and handset markets that once set the baseline have run through several weak replacement cycles.

That mix makes the price forecast unusually sensitive. A single hyperscaler pausing orders can move the aggregate, and a single supplier refusing to cut wafer starts can hold it up. Neither is visible in a quarterly report until it has already happened.

The mechanics of the split are worth stating plainly. Dividing the price by eight does not change earnings per share, the ratio or the cash flow, and it does not create value. What it changes is the size of the minimum ticket and the number of contracts an options market can support.

Companies split when the price has run far enough that retail buyers and employee plans start to notice the unit. The move widens the holder base, and a wider base is useful when a stock needs defending. It is not a defence by itself, and it does nothing for the contract price of a memory chip.

Analysts said the split will do what splits usually do, which is expand participation for a quarter or two. The forecast revision is the item that changes models, because 2027 is the year the current capacity build is scheduled to reach the market.

Memory has a pattern here. The upcycles of 2017 and 2021 both ended the same way, with suppliers adding capacity into demand that had already peaked and contract prices falling faster than anyone had modelled. Each cycle was described at the time as structurally different, and the suppliers that said so were the ones adding the most capacity.

There is a genuine argument that this one is. AI training and inference consume storage at a rate the previous cycles did not, and the density of the drives is rising faster than cost per bit is falling, which supports price. The counterargument is that every supplier can see the same demand curve and is building for it.

What to watch is unglamorous. Contract negotiations for the first quarter will show whether the 2027 revision is a forecast or a warning. Bit shipment guidance will show whether customers are still taking volume. Capital spending plans will show whether anyone intends to slow down.

Inventory is the other tell. Module makers and distributors carry the buffer that absorbs a mismatch, and when that buffer stops being replenished the price falls first at the spot end and later in contracts. Distributor days of supply have historically turned before the contract price does.

Sandisk’s next report will be read against a share price that has already been divided by eight and an outlook that has already been cut once. The split changes how the stock is quoted. The forecast changes how it is priced, and only one of those two can be fixed with a press release.

Related Posts

  • November 20, 2026
  • 10 views
A Lawmaker’s Privacy Suit Puts Model Design on Trial

For three years British courts have sidestepped the question of whether the company that builds an AI model answers for what its users produce with it. A written defence filed…

  • November 20, 2026
  • 6 views
Salesforce Opens Its Customer Data to Microsoft’s AI Tools

Salesforce has spent a decade telling buyers that the customer record belongs in one place, and that its software is that place. On Nov. 20 the company began wiring part…