SanDisk Just Went From $37 to $593 in One Year. Analysts Think It Could Hit $1,000

SanDisk just went from $36 to $593 in one year. Analysts think it could hit $1,000.

$SNDK ( ▼ 2.54% ) traded at $36 exactly one year ago. 

Yesterday it's sitting around $593. That's a 1,500% gain in 12 months, not in some penny stock or crypto gamble, but in a NAND flash memory company that most people forgot existed.

And here's the thing: Wall Street analysts aren't calling it overpriced. They're raising price targets to $700, $800, even $1,000.

AI doesn't just need GPUs. It needs somewhere to store massive amounts of data—model weights, training datasets, inference logs. SanDisk makes enterprise SSDs that handle that storage.

The memory wall problem is creating significant challenges in the datacenter and for edge AI applications.

HBF is our answer to this problem. This NAND-based architecture offers 8x to 16x the capacity of High Bandwidth Memory (HBM), while delivering the same read bandwidth at the same price points.

In the datacenter, SanDisk sees HBF augmenting HBM with the ability to attach terabytes of memory to GPUs.

Q2 fiscal 2026 revenue hit $3.03 billion, up 61% YoY. 

Datacenter revenue alone jumped 64% QoQ. Gross margins crossed 50%. The company guided Q3 revenue between $4.4 and $4.8 billion when analysts expected maybe $2.9 billion.

That guidance gap matters more than you might think.

When a company beats expectations by 10%, the Street takes notice. When they double the consensus forecast, something structural is happening. 

In this case, it's hyperscalers and AI infrastructure builders buying far more enterprise SSDs than anyone predicted six months ago.

$SNDK went from $539 to an intraday high of $676 in a single trading session.

Volume hit over 20 million shares. That's not normal price discovery.

Real on-the-ground question: Is this an early-stage AI infrastructure play that could run for years, or is it a memory cycle spike dressed up in AI hype?

Management says demand will exceed supply through at least 2026.

They're talking about moving customers to longer-term contracts with fixed pricing, something memory vendors almost never do. That would lock in economics over multi-year AI buildouts instead of riding the usual boom-bust cycle.

Bernstein raised their price target to $1,000 and projected fiscal 2027 earnings per share at $90.96. At today's $593 price, that's only 6.3x forward earnings.

Goldman, Jefferies, and RBC all moved targets to the $650-$700 range.

Here's where it gets messy.

If you believe SanDisk is now an AI infrastructure company, not just another memory chipmaker, then a mid-teens P/E on 2027 earnings makes sense. That puts fair value somewhere in the $550-$650 band, which is basically where it's trading now.

But if you haircut those bullish earnings estimates by 30-40% to account for competition and cyclicality, then $SNDK is already pricing in a lot of optimism.

A more conservative entry price would wait for a pullback toward the $400-$500 range.

The technical picture doesn't help. When a stock moves 25% in one day on massive volume, it's signaling either a major regime change or an overshoot that will mean-revert.

Nobody knows which until after the fact.

CEO David Goeckeler keeps emphasizing that this isn't a one-quarter fluke. He's talking about "industry-leading financial performance" and structural resets that position the company for sustained high margins.

The CFO keeps repeating that demand will outstrip supply. If true, that's pricing power. If it's just talking up the stock, well, that's also what CFOs do.

Cantor Fitzgerald's analyst called SanDisk's enterprise SSDs "critical enablers" of hyperscale AI infrastructure, putting them in the same category as GPUs and networking gear as a core bottleneck.

Several research notes frame this as early innings of a multi-year AI capex cycle. They also explicitly warn about volatility given how far the stock has already run.

SanDisk extended its Yokkaichi joint venture with Kioxia through 2034 to lock in NAND supply. They're also evaluating a shift from short-term contracts (over 90% of sales today) toward longer-term agreements.

That would be a structural change in how memory vendors capture value in the AI era. It would also reduce cyclical volatility, which is the main reason memory stocks historically trade at low multiples.

If they pull that off, the AI premium baked into the current valuation starts to make more sense.

At $593, you're somewhere in the middle. Not cheap, not absurdly expensive, but definitely not undiscovered.

For a multi-quarter position focused on AI infrastructure rather than trading momentum, staged entry makes more sense. Start small, add on 20-30% pullbacks toward support levels.

It's not competition from Samsung or SK Hynix. It's not even cyclic.

It's that the stock has gone parabolic in a year, options are pricing in wild swings, and retail traders are piling in after the move is mostly done. When a stock becomes a meme, even one backed by legitimate fundamentals, it can overshoot in both directions.

The Q3 guidance was spectacular. If the company meets or beats that, $SNDK probably holds. 

If they miss, or if macro sentiment wobbles, or if AI infrastructure spending shows any sign of slowing, this could give back 30-40% in a hurry.

SanDisk is riding real AI demand, not just hype. 

Datacenter revenue is up 64% QoQ. 

Management is guiding to numbers that imply a step-change in earnings power. Analysts who actually cover semiconductors are raising targets, not calling it overvalued.

But $SNDK has also gone from $36 to $593 in a year. 

Momentum that extreme doesn't reverse gently. It either consolidates for months or it crashes back to reality.

If you're looking at this for the long term and believe the AI storage thesis, wait for a pullback. 

If you're already in, maybe take some profits and let the rest ride with a tighter stop.

And if you're thinking about jumping in right now at $593 because you're afraid of missing out?

That's exactly when you should wait.