To The Moon

🔥 Two-Thirds of SNDK’s FY2028 Capacity Is Already Sold — This Memory Cycle Is Starting to Look Different

The most important number in the memory market right now may not be NAND pricing.

It may be two-thirds.

SanDisk has already signed New Business Model agreements covering approximately 50% of its FY2027 bits and ~two-thirds of FY2028 bits. These aren’t simple purchase intentions — the agreements are built around committed volumes, minimum financial guarantees and structured pricing mechanisms.

That matters because traditional NAND has always been brutally cyclical:

Demand rises → manufacturers add capacity → supply catches up → prices collapse → margins compress.

But what happens when a substantial portion of future production is already spoken for?

💡 The cycle becomes much more predictable.

And this is where the AI story gets interesting.

AI inference isn’t just creating demand for GPUs. Larger models, growing token volumes and massive KV-cache requirements are increasing the amount of storage required inside data centers. SanDisk itself expects enterprise data-center flash demand to expand substantially as AI inference develops.

So NAND is increasingly moving from “cheap storage” → “critical AI infrastructure.”

That distinction is huge.

If customers need higher density, performance, endurance and reliable supply, the purchasing decision becomes less about finding the absolute cheapest NAND and more about securing supply and capacity ahead of demand.

And SNDK is already showing the evidence.

📌 FY2026 revenue: $20.3B, +175% YoY

📌 Data-center revenue: +437% YoY

📌 ~50% of FY2027 bits contracted

📌 ~⅔ of FY2028 bits contracted

📌 New BiCS10 technology delivering a 60% increase in bit density vs BiCS8

📌 Management’s FY2028–FY2030 model targets mid-to-high-teens revenue growth and ~80% non-GAAP gross margins.

Then Rosenblatt arrives with a $2,400 target, arguing that AI is fundamentally changing the economics and strategic importance of NAND. The market responded immediately: SNDK gained 6.82% to $1,887.04 on Tuesday.

But here’s the part I find most interesting:

The $2,400 target isn’t the thesis.

The thesis is that SNDK is gaining something memory companies historically struggle to obtain:

VISIBILITY.

When two-thirds of FY2028 capacity is already contracted, SNDK isn’t waiting for customers to appear after building supply.

Customers are effectively saying:

“We want the capacity. Reserve it for us.”

That gives SNDK better planning visibility, stronger cash-flow visibility and potentially less exposure to the classic NAND boom-bust cycle.

🚀 This is why I think the memory story has evolved beyond simply “NAND prices are going up.”

The bigger story is:

AI demand → storage intensity → capacity scarcity → customers lock in supply → SNDK gains visibility → stronger earnings visibility → potentially higher valuation.

And if AI inference continues increasing storage requirements, the market may eventually have to rethink how it values NAND companies.

Maybe the question isn’t:

“How high can NAND prices go?”

Maybe it’s:

“How valuable is guaranteed access to NAND capacity when AI infrastructure demand keeps expanding?” 👀

That’s a very different memory-cycle story.

SNDK isn’t just selling memory anymore. It’s increasingly selling something the AI infrastructure market desperately wants: capacity certainty. 🔥