# My analysis of SNDK — retained record

> Retained reddit record from reddit:ValueInvesting.

Canonical: https://highsignal.app/data/records/65ef16ab7440097e

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ValueInvesting)  published 2026-10-24 13:53 ingested 2026-10-25 02:40

[original source ↗](https://www.reddit.com/r/ValueInvesting/comments/1xe9923/my_analysis_of_sndk/)

## retained text



```
Before you say anything, i know its up multiple times this year and that NAND is cyclical. However, I believe we are before a generational opportunity.

Let me explain:

Sandisk was spun off WDC in 2025. The idea was that WDC kept the HDD production, which is basically for the enterprise segment, and Sandisk retained the Flash/NAND/SSD one, which is (was, actually) more consumer oriented, and therefore more cyclical. Moreover, the NAND market was very capital intensive, because building more lines was very expensive. 

But AI has changed everything. Demand for NAND has exploded, and will keep exploding as inference becomes more intensive. Sandisk has signed 94B MINIMUM revenue by Long Term Agreements, which, if we assume their guidance of 50% FCF through 2030, gives us around 50B of guaranteee fcf, backed by 16.5B in financial guarantees (sandisk market cap is, today, 180B). Expectations are high for the next year, expecting 200+ EPS (a forward PE of 6). And what I like the most about all this: share buybacks. Sandisk does not give a dividend, and instead cannibalizes its own shares. In the last quarter, they repurchased 2% of their company at a 1500 average price. The currenr price is 1280 and their FCF has increased, so expect a 3% reduction in ONE QUARTER. Btw, their balance sheet is clean with zero debt, and their capex is kept very low because they have very few upcoming factories (only through their kioxia partnership they are opening a japan facility, but it is partly funded by kioxia and the Japanese Government.

And here is where, i believe, lies the most fundamental transformation for Sandisk, that the market is clearly overlooking: HBF. For non electronic-engineering guys, let me explain why this is so important: for many years, we have been using HBM for TRAINING AI. HMB is basically DRAM, which is volatile, and can be rewritten as many times you want, and is very fast (low latency). When AI is done training, it rewrites new info into the non volatile memory, which is mostly SSDs(this is what sandisk makes). SSDs cant be rewriten infinite times like DRAM, and loses capacity when it is rewriten, so DRAM (or HBM, which is DRAM stacked and connected) acts as the middle player between the GPU and the SSD. But here is where sandisk comes in. What if we used memory only for reading? This way, you would save HBM memory and the GPU could read traight fron the SSD. Of course, this would not be useful for training AI, because memory could not be rewritten, but very useful for cutting inference costs. 

In Summary, SNDK is to me a strong buy due to share buybacks, big backlog, HBF and strong earnings.
```



document fetched 2026-10-25 00:38 · canonical https://reddit.com/r/ValueInvesting/comments/1xe9923/my_analysis_of_sndk
