[Part 2] Is Sandisk is a buy?

The spinoff alone can’t explain the 600%+ gain

Let’s dig deeper into the best performing S&P 500 stock in 2025 (and the one up 64% already in 2026)...

Of course, I’m talking about Sandisk (SNDK).

If you missed the first part of my breakdown, you can catch up here: Why Sandisk stock is going gangbusters.

Considering the stock’s enormous gain in such short order, it may be easy to assume Sandisk is just another meme stock or part of a speculative bubble.

But there’s a lot more going on—including significant real business improvement.

Let me explain...

As I mentioned before, Sandisk spun out of Western Digital in late February 2025. Westen Digital is a storage company that makes both NAND flash products (like SSDs) and old school hard disk drives (HDDs)—the spinning platters that store data magnetically.

The problem with being combined was that the two businesses have very different characteristics. Hard drives are a mature, slow-growth business. NAND flash is faster-growing but more volatile. It was difficult to value the combined company, so investors just slapped a low multiple on the whole thing.

The spinoff unlocked hidden value. SanDisk emerged as a pure-play NAND company—one that investors could evaluate on its own merits.

But the spinoff alone can’t explain the 600% plus gain.

That’s where the dramatic improvement in Sandisk’s business fundamentals come in…

Let’s look at the most recent quarter (fiscal Q1 2026, reported in November 2025)

Revenue came in at $2.31 billion, up 21% from the prior quarter and 23% from a year ago. And non-GAAP EPS was $1.22, well above Wall Street’s expectations.

What’s more, management expects gross profit margin to jump from 30% in fiscal Q1 to 41%-43% in fiscal Q2 (this quarter). At the midpoint, that’s a 12-percentage-point improvement in a single quarter, which is huge. And it’s coming from pricing power and cost efficiencies.

On the pricing side—thanks to the memory supercycle I mentioned last time—Sandisk can continue to raise prices without losing customers because demand far exceeds supply. Management guided for double-digit price increases this quarter—meaning prices are going up more than 10%.

On the cost side, Sandisk is ramping a new generation of technology called BiCS8, which comes with lower manufacturing costs per unit of storage. As BiCS8 becomes a bigger portion of production, costs per gigabyte fall.

The company also eliminated some transitional costs that were weighing on margins—things like factory startup expenses and underutilization charges from when it wasn’t running facilities at full capacity.

How Sandisk stacks up against the competition

The NAND flash market is extremely competitive. And it’s dominated by five suppliers who capture most of the revenue.

Here’s a breakdown of the space from TrendForce:

As you can see in the table above, Samsung is the biggest NAND flash supplier with about a third of the market. South Korean rival SK Group is second, followed by Japan’s Kioxia. Then comes US-based Micron, and finally Sandisk with about a 12% market share.

But Sandisk isn’t really the smallest of the “big 5” NAND flash providers because it has a unique arrangement with Kioxia. The two companies jointly operate manufacturing facilities in Japan and share the enormous R&D costs required to develop new memory technology.

This joint venture has existed for more than two decades, and it gives Sandisk access to world-class manufacturing without having to fund it entirely on its own.

Together, Sandisk and Kioxia are the second largest NAND supplier—and they were the largest until recently.

When it comes to the technology itself, NAND flash makers compete on something called layer count. More layers mean you can store more data in the same physical space.

SK Hynix leads here with 321 layers. Samsung’s has 286. Micron has 276. And Sandisk’s current BiCS8 tech has 218 layers.

So it looks like Sandisk is behind here—and technically it is. But the company argues that raw layer count isn’t the only metric that matters. What really matters is capital efficiency—how much storage you can produce for each dollar of investment.

Sandisk uses a different approach called memory hole density—basically packing memory cells closer together horizontally instead of just stacking them vertically like standard NAND flash. Think of it as building apartments with smaller hallways and more rooms per floor, rather than just adding more floors.

Management claims this approach lets them achieve competitive storage density with fewer layers, which means less capital investment per gigabyte of output.

The fact that Sandisk’s margins are expanding lends credence to this claim. The tech is competitive enough to participate in the current upcycle.

That said, I should acknowledge a legitimate weakness: Sandisk has been a laggard in the enterprise SSD market. Enterprise SSDs are the storage drives that go into data center servers—exactly where AI demand is hottest. Samsung, SK Hynix’s Solidigm unit, and Micron have all been ahead of Sandisk in winning business from the big cloud providers.

Sandisk is working to close this gap. It’s developed a new line of enterprise SSDs called Stargate that’s currently being tested by major hyperscalers like Amazon, Microsoft, and Google.

But it’s important to recognize that Sandisk is currently behind here.

One more thing worth mentioning...

Sandisk is developing a potentially game-changing technology called High Bandwidth Flash, or HBF

Remember I mentioned HBM (High Bandwidth Memory) last time. It’s the premium type of DRAM that sits directly on top of the AI processors, allowing data to flow much faster—and commanding fat margins.

The problem is that HBM is expensive and can only provide limited storage capacity.

HBF is Sandisk’s attempt to bring some of HBM’s speed advantages to NAND flash. Think of it as a hybrid—not as fast as HBM, but way faster than regular flash, and with much higher capacity.

For AI inference (when the model is actually answering questions, rather than being trained), HBF could be a big deal. It could allow systems to store much larger AI models and retrieve them faster than current technology allows.

SK Hynix has agreed to collaborate with Sandisk on HBF standards—a validation that the technology has merit. But even the companies developing HBF say it won’t be commercially available until next year—meaning that’s the best case scenario.

This brief discussion on HBM and HBF is a good segue to my conclusion about Sandisk’s stock today…

See, HBM turned DRAM from a commodity-like product into a high-tech, customized core component of AI that’s tightly integrated with AI processors and crucial to consider in the earliest stages of AI system design.

This tight coupling flipped the script on how memory is procured and priced. With HBM, long-term contracts and sold-out supply are the norm, keeping prices much higher and much steadier, and reflecting a structural shift in how this part of the memory market works (as opposed to the traditional boom-bust cycle).

If HBF can essentially do the same thing for NAND flash, then Sandisk will become a strong long-term buy in my view. But that’s a big “if”...

Meanwhile, I think NAND flash remains a commodity-like technology subject to the memory market’s traditional boom-bust cycle. So even though the current supercycle is exceptionally strong, I expect it to eventually crash like all the others, taking Sandisk down with it, unless we see a tech transformation like we did with HBM.

Long story short: I think Sandisk is a good company doing good things. But I wouldn’t buy the stock personally unless and until we see HBF transform the market like HBM has.

Thanks for reading,

Chris Wood
Grow or Die