To The Moon

This memory correction has gone from uncomfortable to brutal.

Micron has fallen from above $1,000 to roughly $874 in the screenshot. SanDisk is around $1,336. SK Hynix has been hammered, and leveraged memory products have suffered even more.

The immediate culprit is China's CXMT.

Its new 4F² DRAM architecture potentially allows more memory cells to fit into the same area, and CXMT says it intends to launch server-grade DDR5 products using the technology before year-end.

That matters.

But there is one enormous distinction investors should not overlook:

CXMT has demonstrated technological progress. It has not yet demonstrated high-volume, high-yield, economically competitive production.

The market is currently pricing those two things almost as if they were the same.

I think that is premature.

Why This News Hit MU So Hard

The market is not really worried about a few CXMT server-memory modules.

It is worried about what happens next.

The memory bull market has been built partly on scarcity.

AI has absorbed enormous amounts of HBM. Capacity has migrated toward higher-value products. Conventional memory tightened. Prices rose. Margins exploded.

That created extraordinary economics for Micron, SK Hynix and Samsung.

Now introduce another potentially credible DRAM supplier.

The market immediately starts modelling:

More supply → weaker pricing → lower margins → lower peak earnings → lower valuation.

That is why MU can lose billions of dollars in market value before CXMT ships meaningful commercial volume.

Stocks price the future.

But sometimes they price the future too aggressively.

The Most Important Missing Number: Yield

This is where I think the market narrative has run ahead of the evidence.

A denser memory architecture looks impressive.

But semiconductor manufacturing does not reward PowerPoint slides.

It rewards yield.

Can CXMT manufacture the new design consistently?

At scale?

At acceptable cost?

With server-grade reliability?

With sufficient customer qualification?

And in enough volume to influence global pricing?

Those questions remain much more important than the architecture announcement itself.

Even bullish analysts are highlighting that commercialisation and manufacturing scale remain unproven.

That does not mean CXMT should be ignored.

It means I would treat CXMT as a credible emerging risk, not yet a proven destroyer of Micron economics.

And CXMT Is Not Yet the Same Thing as HBM Competition

This distinction matters enormously.

CXMT is targeting server DRAM.

The crown jewel of the current AI memory cycle remains HBM.

HBM is technologically difficult, packaging intensive and deeply tied to qualification with leading AI accelerator platforms.

That is still an area where SK Hynix, Micron and Samsung possess major advantages.

China is progressing quickly in memory.

But conventional/server DRAM competition and leading-edge HBM competition are not interchangeable.

So the question I am asking is not:

Can CXMT make DRAM?

Clearly it can.

The question is:

Can CXMT disrupt the profit pool that investors are currently paying Micron and SK Hynix for?

Today, I think the answer is:

Eventually possible. Immediately proven? No.

Something Else Has Changed

There is another reason MU is falling harder than the CXMT announcement alone would justify.

The stock had become priced for extraordinary conditions.

Micron's margins surged.

Memory prices surged.

AI demand surged.

Investors extrapolated.

When everything is excellent, the market becomes vulnerable to anything suggesting that excellence might eventually normalise.

So CXMT did not need to destroy the memory thesis.

It merely needed to challenge the assumption that today's scarcity lasts indefinitely.

That is enough to compress the multiple.

This is why I think the current decline contains two corrections at once:

A fundamental repricing of future competition.

And

a valuation reset after an extraordinary rally.

Those should not be confused with an immediate collapse in current demand.

Musk's Comment Is More Important Than It Looks

There is also an interesting counter-signal.

Elon Musk has publicly praised Micron's approach to memory pricing, arguing that Micron has been comparatively reasonable with customers.

That might sound like a throwaway comment.

I do not think it is.

Memory buyers dislike extreme supplier pricing power.

If customers believe suppliers are exploiting shortages, they have every incentive to qualify alternatives.

A supplier that offers reasonable long-term pricing can potentially build deeper customer relationships before additional competition arrives.

Micron has also been securing substantial long-term supply commitments.

That gives me greater confidence than simply looking at today's spot memory price.

The bull thesis increasingly needs to become:

HBM leadership + long-term contracts + customer relationships + technology execution.

Not simply:

Memory is scarce, therefore MU goes up.

That old thesis is no longer enough.

MU Has Now Reached My Previous Strong Pick Zone

This is where things get interesting.

My previous framework had:

$990-$1,020: First Pick

$940-$975: Preferred Pick

$880-$925: Strong Pick

MU has now fallen straight through the first two zones and reached the third.

So I am not going to move the goalposts simply because the candles turned red.

But I am going to adjust for the new information.

My Updated MU Levels

$860-$890: First Pick Now

At roughly $874, MU has reached a level where I would begin accumulating again.

But I would not deploy everything.

$800-$840: Preferred Pick

This becomes my favourite risk/reward zone if HBM demand remains strong and CXMT still lacks evidence of large-scale competitive production.

$740-$790: Strong Pick

This would represent a very substantial reset.

If MU reaches here while earnings, HBM commitments and AI capex remain intact, I would become considerably more aggressive.

$930-$950: First Recovery

Reclaiming this area would suggest forced selling is beginning to exhaust itself.

$990-$1,020: Bullish Confirmation

$1,000 has changed from support into resistance.

MU needs to earn it back.

$1,080-$1,120: Next Target

$1,180-$1,220: Bull Extension

Below $740: Thesis Check

If MU reaches this level, I would stop assuming the market is merely overreacting and investigate whether forward memory economics have changed substantially.

Why I Would Not Use Leverage Here

This is important.

MU itself has fallen hard.

A leveraged MU product has fallen much harder.

When volatility expands, leveraged products amplify not only direction but also the damage caused by repeated swings.

If MU drops 8%, rebounds 6%, falls 7%, then rebounds again, a leveraged product can suffer substantial path-dependent erosion even if MU eventually recovers.

For this phase, I prefer owning the underlying stock.

I want time working for me.

Not leverage working against me.

SanDisk at $1,336 Is Becoming Interesting Too

SNDK has been dragged into the selloff despite CXMT being primarily a DRAM threat.

That tells me part of this move is indiscriminate memory-sector liquidation.

And that creates potential opportunity.

SanDisk is primarily a NAND story.

CXMT server DRAM does not directly attack SanDisk's core business.

Yet SNDK has been punished almost alongside the DRAM names.

That is precisely the kind of dislocation I look for.

Updated SNDK Levels

$1,300-$1,350: First Pick

We are essentially here.

$1,200-$1,275: Preferred Pick

$1,100-$1,175: Strong Pick

$1,425-$1,475: First Recovery

$1,525-$1,575: Bullish Confirmation

$1,650-$1,700: Next Target

Below $1,100: Thesis Check

Between MU and SNDK, I still prefer MU fundamentally.

But SNDK may actually offer the cleaner CXMT overreaction trade, because CXMT's announcement does not directly attack NAND.

What About SK Hynix?

SK Hynix has the strongest HBM positioning of the group.

That is simultaneously its strength and its vulnerability.

If HBM remains structurally scarce, SK Hynix deserves a premium.

If the market begins questioning AI infrastructure spending itself, that premium can compress rapidly.

CXMT's server DRAM announcement alone does not erase SK Hynix's HBM technology advantage.

So I would be watching HBM orders far more closely than the stock's daily volatility.

My 2-4 Week MU Scenarios

I have become more cautious because technical damage is now significant.

45% Base Case: Bottoming Process

MU trades roughly $820-$930.

There are violent rebounds and equally uncomfortable selloffs as investors determine whether CXMT represents an immediate earnings threat or primarily a future competitive threat.

I think this is the most likely outcome.

30% Bull Case: Market Decides It Overreacted

Evidence emerges that CXMT still faces meaningful yield, scale or qualification barriers.

HBM demand remains extremely strong.

AI capex stays intact.

MU reclaims $930-$950, then attacks $990-$1,000.

That would be my first meaningful evidence that the correction is ending.

25% Bear Case: Scarcity Premium Keeps Unwinding

MU loses $850 decisively.

Investors begin pricing lower conventional DRAM prices and structurally greater Chinese supply.

MU falls toward $780-$820, with an overshoot toward $740-$760 possible if semiconductor sentiment deteriorates simultaneously.

This scenario cannot be dismissed anymore.

What Would Tell Me CXMT Really Changes the Thesis?

Not another headline.

I want evidence.

High manufacturing yields.

Meaningful production volume.

Major server customer qualification.

Competitive pricing.

Rapid market-share gains.

Evidence that incumbent DRAM contract prices are falling because of CXMT supply.

And eventually:

credible competition in HBM.

If those start happening together, I reduce my valuation assumptions for the incumbent memory companies.

Until then, I refuse to treat an architectural breakthrough as identical to commercially scaled production.

The Real Signal I Am Watching

Watch memory pricing.

If CXMT headlines keep coming but DRAM contract pricing remains resilient, HBM stays tight and Micron maintains strong forward commitments, the stock market may eventually conclude it moved too quickly.

But if DRAM pricing begins weakening sharply while supply estimates rise, the bears will have evidence rather than simply a narrative.

That distinction matters.

My Ranking After the Selloff

1. MU: My highest-conviction fundamental pick

At $860-$890, I start accumulating.

At $800-$840, I become more interested.

Below $790, I check the thesis before buying aggressively.

2. SNDK: My favourite overreaction setup

It is being punished for a DRAM development despite being primarily exposed to NAND.

That mismatch interests me.

3. SK Hynix: Best HBM technology positioning

But after the enormous run, its valuation remains sensitive to any change in AI capex expectations.

My Pick

I am choosing MU.

But this is no longer the easy memory trade we had earlier in the year.

CXMT matters.

China matters.

Conventional DRAM supply matters.

And memory companies will eventually have to prove that today's extraordinary margins can survive greater competition.

But at approximately $874, the market has already erased a substantial amount of value from MU.

The question is therefore no longer:

Is CXMT a threat?

Yes, it is.

The better question is:

How much of that threat is already reflected in MU's price before CXMT has proven yield, scale and customer adoption?

At $1,100+, I wanted caution.

At $1,000, I wanted staged accumulation.

At $860-$890, I am interested.

At $800-$840, I would become considerably more interested if the HBM thesis remains intact.

And that is the discipline I want during corrections.

Do not pretend new risks do not exist.

But do not wait until everyone feels safe again before deciding the price is attractive.

By then, the opportunity usually looks very different.

CXMT has changed the risk premium. It has not yet proven that it can change the HBM hierarchy.

That distinction is why I am not abandoning MU here.

I am repricing it.

I am not a financial advisor. Trade wisely, Comrades!

Is MU at $874 becoming attractive to you, or are you waiting for $800?

If you find this post interesting, give it wings! 🪽 Repost and share the insights.

Do consider ‘Follow me’ and get firsthand read of my daily new post. Thank you.

@Tiger_SG  @TigerEvents  @TigerStars  @TigerPM  @Daily_Discussion