To The Moon

Hi everyone. Today I want to share a special analysis of the memory sector.

Let me start with the conclusion: the most violent phase of forced deleveraging may be close to ending, but that does not mean the correction is over. Memory fundamentals remain strong, while price action and positioning have yet to confirm a genuine reversal.

1. Fundamentals: The thesis remains intact, but expectations got too far ahead

The latest rally began after SanDisk’s earnings-driven rerating in late April. Kioxia, SK hynix, Micron and SanDisk all moved sharply higher. Fundamentals started the rally, but Momentum-driven capital eventually took over.

This was not a purely speculative move. $SanDisk Corp.(SNDK)$ reported quarterly revenue of $5.95 billion, up 97% sequentially. Data-center revenue increased 233%, while non-GAAP gross margin reached 78.4%. Management guided next-quarter revenue to between $7.75 billion and $8.25 billion, showing that earnings were still growing rapidly. In other words, the rally did have real earnings support.

Memory pricing has not suddenly turned negative either. TrendForce expects third-quarter conventional DRAM contract prices to rise 13%–18% quarter over quarter, with NAND Flash prices increasing another 10%–15%. The problem is that prices are still rising, but the pace of those increases is beginning to moderate, while consumer-electronics customers are approaching the limits of what they can afford.

So the issue is not that memory fundamentals suddenly collapsed.

The real issue is that share prices had already priced in higher memory prices, a longer upcycle and a much more optimistic 2027 earnings outlook.

Stocks do not trade simply on whether the results are “good” or “bad.” They trade on whether companies can continue beating increasingly aggressive expectations. Once earnings remain strong but stop accelerating, crowded investors begin locking in profits.

$Alphabet(GOOG)$Alphabet’s upcoming earnings will therefore be the first major test. The market will focus less on quarterly profit and more on AI capital expenditure, data-center investment and Google Cloud growth.

If Alphabet maintains or raises its AI spending plans, it would confirm that infrastructure demand remains intact. However, that alone may only remove part of the fundamental concern—it may not be enough to produce an immediate V-shaped recovery in memory stocks.

Conversely, if hyperscalers begin cutting CapEx guidance or become vague about future spending, memory stocks could quickly enter a second leg lower.

$Seagate Technology (STX)${{e3f9e53c189e4c89bb98c25b11df66e8}} will also report earnings around the end of the month. Although HDDs, NAND and DRAM are different markets, their stocks are closely linked through AI infrastructure spending and data-center demand. In the current environment, strength or weakness in one part of the storage complex can easily affect sentiment across the entire sector.

For SanDisk specifically, the events that could truly determine its medium-term direction are the August 5 earnings report and the August 13 Investor Day. The fundamentals may not regain control over fund flows unless SanDisk beats the upper end of its guidance and raises its outlook for pricing, margins or future earnings.

2. Technicals and positioning: The fastest part of the selloff may be over, but the correction is not

SNDK fell from approximately $2,354 to $1,354 by last Friday, representing a drawdown of more than 40%.

The stock touched $1,325 during the week and closed near its weekly low. Several intraday rebounds were also sold aggressively. That is not the price action of a stock that has already confirmed a durable bottom.

This decline looks more like a classic Momentum unwind.

The first phase is driven by leveraged funds, quantitative strategies and stop-loss orders exiting simultaneously. It is usually the fastest and most violent phase of the correction.

Once those forced liquidations begin to fade, the stock can stage a sharp rebound because mechanical selling pressure has temporarily eased and short sellers start taking profits.

But a second phase often follows.

During that phase, unleveraged long-term holders use rebounds to reduce their positions. The stock repeatedly tests resistance, fails to break through and then enters a period of sideways consolidation or gradual decline.

The second phase may be less violent, but it usually lasts longer. It is also the phase most likely to create the illusion that investors are “about to make their money back.”

On Monday, SNDK and $Micron Technology(MU)$ rebounded alongside the broader semiconductor sector, with SNDK briefly reaching approximately $1,450. For now, however, this should still be treated as an oversold bounce rather than a confirmed trend reversal. Monday’s move did not show enough buying conviction to change the broader technical structure.

Here are the key levels I am watching:

• $1,325–$1,400: The first support zone. Holding this area would allow the rebound to continue.

• $1,450–$1,500: The first resistance zone. Failure to hold above it would show that selling pressure remains.

• $1,600–$1,650: SNDK needs to reclaim this area before the short-term structure can be considered meaningfully repaired.

• $1,740–$1,800: Only a sustained breakout above this zone would justify discussing a medium-term reversal.

• Around $1,200: The next major support if $1,325 breaks.

• Around $1,000: This would require broader equity-market or credit-market stress. It is an extreme scenario rather than my base case—（and hopefully we never get there. I am already close to crying after this decline. 😭[Cry]）

My view

Strong fundamentals are a reason not to panic and declare the entire memory cycle dead. Weak technicals, however, are a reason not to rush into leveraged dip-buying.

Investing is a personal decision. Do not blindly follow anyone. Nobody can predict the market with certainty, and no investor or commentator should be treated as infallible.

The more likely short-term path remains:

an oversold rebound, rejection near resistance, followed by a period of consolidation.

Until SNDK can reclaim and hold the $1,600–$1,650 area, I will treat any rally as a rebound rather than the beginning of another major uptrend.

Wanting to recover your losses is understandable, but your cost basis is not a support level, and the desire to break even is not a trading signal.

What is worth waiting for is clear evidence that capital outflows have stopped, resistance has been broken and earnings expectations are being revised higher again.

One encouraging development this week came from Kimi K3. After the model went viral, user demand pushed Moonshot AI’s GPU capacity close to its limit within 48 hours. The company temporarily paused new subscriptions to protect the experience of existing users and said it was expanding capacity as quickly as possible. (Kimi’s official announcement)

GPUs were the immediate bottleneck, but AI servers also require enormous amounts of HBM, DRAM and storage. This episode suggests that more efficient and affordable models do not automatically reduce hardware demand. Better models can attract more users, generate more inference requests and ultimately increase total compute requirements.

Model-level advantages may also prove less durable than many people assume, while hardware bottlenecks remain far more concentrated. At scale, HBM supply is still dominated by only three companies: $SK hynix(SKHY)$, Samsung Electronics and $Micron Technology

That is another reason I believe the long-term memory fundamentals remain strong—even if the stocks still need more time to repair the damage caused by this Momentum unwind.

This is not financial advice and is provided for discussion purposes only.