To The Moon

Monday’s market looked completely contradictory.

The United States paused its strikes on Iran, Brent crude briefly fell below $88, and oil dropped more than 8% in a single session. Normally, falling oil prices should ease inflation fears and support growth stocks.

Instead, AI hardware was crushed.

$SanDisk Corp.(SNDK)$ closed down about 11% after falling more than 13% intraday. $NVIDIA(NVDA)$ lost roughly 5%, while $Western Digital(WDC)$ and $Seagate Technology(STX)$ declined around 4%. $Micron Technology(MU)$ was down more than 7% at one point before recovering part of the loss.

Meanwhile, software ETF IGV gained more than 3%, and $Apple(AAPL)$remained strong.

This was not a broad risk-off move. It was a clear rotation out of the most crowded AI hardware, semiconductor and memory trades—and into software and lower-volatility names.

My conclusion is simple: the TACO trade removed part of the oil risk, but it did nothing to solve the positioning, valuation and AI spending concerns inside the hardware trade.

TACO Worked Again—but It Only Solved the Oil Problem

The so-called “TACO Index” is not a real tradable index. It is a framework for estimating when Trump may change direction under pressure from oil prices, equities, Treasury yields and disruptions around the Strait of Hormuz.

Brent had climbed above $100 last week while Treasury yields continued rising. If the conflict had escalated further, Trump would have faced higher oil prices, stronger inflation, tighter monetary policy and growing pressure ahead of the midterm elections.

Under those conditions, the pause was not surprising.

But a pause does not mean the Middle East risk has disappeared. Military action could still resume, while shipping through the Strait of Hormuz and the Red Sea has not fully normalized.

Monday’s oil collapse therefore looked more like a rapid removal of the war premium than the beginning of a long-term bear market in crude. If oil continues falling, the EMA20 area near $80 will be the first level I watch.

More importantly, cheaper oil cannot fix the internal problems facing AI hardware.

CXMT Was the Trigger, Not the Whole Reason

The most obvious trigger for the memory selloff was CXMT’s market debut.

The stock surged 466% on its first day, briefly reaching a valuation close to $490 billion while raising substantial new capital. Overseas investors are not worried about how much CXMT gained on day one. They are worried about how much new equipment, capacity and supply that funding could eventually create.

The market also reacted to reports that domestically developed immersion DUV equipment had entered small-scale production. Investors had assumed that China’s memory expansion would remain constrained by semiconductor equipment restrictions. That assumption is now being questioned.

But there is an important distinction.

CXMT primarily produces DRAM, while $SanDisk(SNDK)$ is mainly exposed to NAND. If Monday’s move had only been about new Chinese DRAM capacity, SanDisk should not have fallen more than Micron.

That tells us CXMT was only the trigger. The real source of the damage was crowded positioning.

SanDisk, Micron and SK Hynix were among the strongest momentum trades of the year. Once investors started questioning AI capital-spending returns, NVIDIA’s circular financing exposure to OpenAI and the possibility of future Chinese supply growth, the highest-beta winners became the first positions to be cut.

The market was not selling one specific SanDisk business. It was unwinding the entire AI hardware and memory trade.

SanDisk: The Real Test May Come at the EMA200[Spurting][Cry]

$SanDisk Corp.(SNDK)$ has recently followed a surprisingly consistent rhythm:

Roughly five days down, four days up—followed by another five-day decline and another four-day rebound.

The stock has now entered a new downswing and broken below its daily EMA100.

The long-term descending trendline near 1,700 has already produced three false breakouts. Each attempt above it was quickly rejected, confirming that overhead supply remains heavy.

At the same time, repeated tests gradually absorb that supply. A fourth breakout attempt may have a better chance of succeeding, but the stock probably needs to complete this correction first.

The next major level is the EMA200 near 1,000.[Cry][Cry][Cry][Cry][Cry]

A decline to that level would still represent more than 20% downside, which sounds extreme. But SanDisk is a high-volatility momentum stock, and its recent trading rhythm suggests that it could approach this area between this week and next.

Reaching the EMA200 would not automatically confirm a bottom. I would want to see three signals:

• Volume expands, but the stock stops making new lows.

• SanDisk, Micron and SK Hynix stabilize together.

• Earnings, orders and memory pricing show no material deterioration.

If all three appear at the same time, the EMA200 could become the starting point for a major rebound—or even the next trend reversal.

SK Hynix: One More Capitulation Before Earnings?

$SK hynix(SKHY)$ has shown a similar but even more regular pattern:

Two days of selling, roughly seven days of bear-flag consolidation, and then another move lower.

Its daily EMA100 has now been tested seven times. Repeated tests do not make support stronger. Each test consumes another layer of demand, increasing the risk that the level eventually breaks.

I am watching two downside areas:

• The gap near KRW 1.46 million.

• The EMA200 near KRW 1.33 million.

SK Hynix is about to report earnings. If the stock continues falling before the announcement and forces leveraged and panic-driven holders out, results may only need to avoid a major disappointment to produce a “sell the rumor, buy the news” rebound.

The massive Korea–U.S. AI semiconductor cooperation framework announced over the weekend is positive for long-term demand. But headline numbers should not be treated as guaranteed SK Hynix revenue.

What matters is how much purchasing volume is actually committed, how pricing will be determined, who will fund the required capacity and how the risks will be divided if demand falls short.

A more immediate positive signal is that Korean pension funds became net buyers of domestic equities in July and increased their exposure to SK Hynix. At the very least, the feared wave of pension-fund selling has not continued to worsen.

Software Is Taking the Lead—for Now

IGV gained more than 3% while memory and semiconductor stocks fell sharply. The hardware-to-software rotation is already visible.

But it is too early to conclude that software has started a sustainable new bull market.

IGV previously failed to reclaim 108 and has returned to its long-term trendline. The first important short-term resistance level is the daily EMA200 near 94.

If $Microsoft(MSFT)$ proves that Azure growth can justify its massive capital spending—and IGV successfully clears 94—the software rotation could continue. That would keep pressure on memory and semiconductor stocks.

But if Microsoft repeats the same pattern as Alphabet—higher CapEx without enough near-term cash flow or AI revenue to justify it—IGV could be rejected near 94.

Microsoft’s earnings may therefore mark the end of the software rebound rather than its beginning.

If SanDisk, Micron and SK Hynix have already completed their capitulation by then, capital could begin rotating back from software into hardware.

Three Variables Matter Next

The first is the Federal Reserve.

Oil has fallen sharply, but the inflationary pressure created by its earlier rally has not disappeared completely. The key issue is not simply whether the Fed changes rates at one meeting. It is how policymakers describe the future path of inflation and interest rates.

As long as Treasury yields remain elevated, expensive technology stocks will struggle to escape valuation pressure.

The second is the Bank of Japan.

If the yen continues weakening, Japanese investors may reduce their exposure to U.S. Treasuries, placing additional upward pressure on yields. That may not be fully reflected in a single trading session, but it could continue limiting the valuation ceiling for growth stocks.

The third is the upcoming earnings wave.

$Microsoft(MSFT)$ $Meta Platforms, Inc.(META)$, $Apple(AAPL)$$Amazon.com(AMZN)$ SK Hynix and Samsung will all report results.

This earnings season, beating revenue and EPS estimates is no longer enough. $Alphabet(GOOG)$ and $Tesla(TSLA)$ have already shown that strong headline numbers can still be punished if capital spending rises too quickly, free cash flow deteriorates or management cannot clearly explain the return on AI investment.

The key questions are:

• Can Azure growth justify Microsoft’s Capex?

• Will Meta raise its spending guidance again?

• Can AWS orders and margins keep pace with Amazon’s AI investment?

• Can Apple maintain hardware demand while showing real AI progress?

• Can SK Hynix and Samsung protect HBM shipments, margins and pricing?

Final Take

The TACO trade and the collapse in oil prices temporarily reduced the risk of another inflation shock. But they did not solve the positioning, valuation and capital-spending anxiety inside AI hardware.

CXMT’s debut was the trigger for Monday’s memory selloff. The deeper problem was an overcrowded trade being forced to unwind.

For SanDisk, the key level is the EMA200 near 1,000. For SK Hynix, I am watching the gap near KRW 1.46 million and the EMA200 near KRW 1.33 million.

Memory fundamentals have not collapsed. What has broken is the technical structure—and the market is now clearing out excessive positioning.

There is no reason to treat every rebound as a reversal. But there is also no reason to abandon the long-term memory thesis at the moment of maximum fear.

If SK Hynix avoids a major earnings disappointment, memory prices and AI orders remain firm, and SanDisk, Micron and SK Hynix stabilize together at major support, this week or next could mark the final stage of the current memory correction.

The fundamentals are still alive. The crowded trade is what is being destroyed.only. Oh my sndk[Cry]

Poll: Is the memory-stock selloff close to ending?

• Yes—capitulation is close

• No—more downside ahead

• Only after earnings

• Watching the EMA200

Not financial advice. For discussion