#SNDK

$SNDK is becoming one of the most interesting ways to play the AI infrastructure boom — and the numbers show that this story is no longer only about traditional memory demand.
Sandisk has just closed a remarkable fiscal 2026.
Full-year revenue reached $20.25 billion, representing a massive 175% year-over-year increase.
But the number that caught my attention is what happened inside that revenue.
Datacenter revenue increased 437% year over year.
That is the real story behind $SNDK.
AI infrastructure doesn’t only require GPUs.
It requires enormous amounts of storage, high-performance SSDs and memory infrastructure capable of handling the rapidly increasing volume of data generated by AI workloads.
And Sandisk is increasingly positioning itself directly inside that demand cycle.
The acceleration is visible quarter by quarter.
Fiscal Q3 datacenter revenue was approximately $1.47 billion.
One quarter later, it jumped to approximately $2.98 billion.
That’s a sequential increase of roughly 103%.
Total Q4 revenue also reached $8.97 billion, up 51% sequentially and 372% year over year.
This is not normal semiconductor growth.
It is a structural shift in the company’s business mix.
And Sandisk isn’t expecting the momentum to stop there.
Management’s fiscal Q1 2027 revenue guidance is $10.3–$10.8 billion.
If the company reaches the midpoint, that would represent another major step up from the previous quarter.
There is also another number I find particularly interesting:
$14 billion.
Sandisk’s board authorized an additional $14 billion share repurchase program, bringing the remaining authorization to approximately $15.5 billion.
That creates another potential source of shareholder support while the company continues expanding its datacenter business.
But the biggest question is obvious:
Can this growth continue?
The answer depends heavily on AI infrastructure spending.
NVIDIA’s latest earnings suggest that AI capex is still accelerating rather than slowing.
That is extremely important for Sandisk because the AI data-center ecosystem needs more than compute.
It needs to store, move and retrieve enormous quantities of data.
And that’s where NAND and enterprise storage become increasingly important.
There is also a supply-side story developing.
Sandisk and Kioxia recently announced plans for more than $31 billion of combined investment in Japan through 2032 to expand advanced memory production and address growing AI-driven demand.
That tells me the industry itself is preparing for much higher memory consumption.
But I wouldn’t ignore the risks.
Memory is a cyclical business.
Prices can rise extremely quickly when supply is tight, but the same cycle can reverse when manufacturers add too much capacity.
And $SNDK has already experienced an extraordinary re-rating.
So the question isn’t whether the company is growing.
It clearly is.
The question is whether the market has already priced in too much of that growth.
For me, the key metrics going forward are:
Datacenter revenue
NAND pricing
AI infrastructure spending
Gross margins
Free cash flow
and most importantly:
Whether datacenter growth remains strong enough to offset the cyclical nature of the memory market.
The bullish scenario is straightforward.
AI spending remains elevated → storage demand increases → NAND pricing stays strong → Sandisk maintains high margins → datacenter becomes an increasingly dominant part of the business.
The bearish scenario is equally clear.
AI capex slows → memory supply catches up → NAND prices weaken → margins compress → the market begins questioning the current valuation.
That’s why I don’t see $SNDK simply as a memory-stock trade anymore.
I see it as a second-order AI infrastructure play.
NVIDIA provides the compute.
Hyperscalers build the infrastructure.
Sandisk helps provide the storage layer underneath it.
And as AI models become larger and generate more data, that storage requirement doesn’t disappear.
It grows.
$20.25B annual revenue.
175% year-over-year growth.
437% datacenter growth.
$10.3–$10.8B next-quarter revenue guidance.
Those numbers are difficult to ignore.
For me, the next question isn’t whether AI needs more storage.
It clearly does.
The real question is whether $SNDK can continue converting that AI demand into sustainable margins and free cash flow.
If it can, the current AI infrastructure cycle may still have another chapter to write. 👀$SNDK