Conventional wisdom says when you are up 60% on a position you trim. Lock in some gains. Take money off the table. Sleep better.

I am not doing that. And I want to walk through exactly why, including the things that would change my mind, because a thesis without an exit is just a feeling.

This Is a Re-Rating, Not a Rally

Here is the most important thing I have realized watching this sector over the past year.

This is not memory stocks having a good run. This is the entire memory sector being re-rated by the market because of AI infrastructure. Those are completely different things and they call for completely different decisions.

A rally you trim. A re-rating you hold.

Look at SanDisk over the past year. Look at what the market is paying for AI infrastructure names across the board. Bloom Energy trading at absurd multiples on a power infrastructure thesis. The whole category is being repriced from boring commodity businesses into critical AI infrastructure. Memory is going through the same repricing, except memory has the tightest supply situation of any of them and only three companies that can make the high end.

When a sector re-rates, the old price anchors become meaningless. People who anchored to Micron as a cyclical commodity stock trading at 7 times earnings are using the wrong map. The market is in the process of deciding these are irreplaceable infrastructure companies. That process is not finished. Trimming in the middle of a re-rating is selling something for yesterday's valuation while the market is still figuring out today's.

What History Taught Me About Selling Too Early

I have been here before. I held Tencent from around 700. I held Nvidia from around 300. ( Both cost prices are before split price ) Both became enormous winners.

And I sold both too early.

In fairness to myself, I was up three to four times on each when I sold, so it was not unreasonable. Taking a 3x or 4x off the table is a defensible decision and I do not beat myself up over it. But I watched both run much further after I was out. I left a lot on the table because I treated a structural shift like a trade.

The lesson I took is specific. When you have correctly identified a structural shift before the consensus, the mistake is not holding too long. The mistake is selling a structural winner as if it were a swing trade. The conviction that got you in early is the same conviction that should keep you in through the noise.

So this time I want to let it ride longer. Not forever. Longer. Up 60% is not where structural winners are sold. It is barely where they get started.

Why $80

My first target is $80. That is roughly a double from my average cost.

It is not a precise valuation model. I will be honest about that. It is a marker. My genuine belief is that this cycle runs at least another 1.5 to 2 years given the supply timelines, the fab construction lead times, and the demand visibility the chipmakers themselves are describing. If we get to $80 and there is no negative structural news, I will probably not sell even there. I will reassess based on what the world looks like at that point, not based on a number I picked today.

The number is a checkpoint, not a trapdoor.

What Would Actually Make Me Sell

This is the part most people skip, so here it is plainly.

My real tripwire is hyperscaler guidance. If Microsoft, Google, Amazon, and Meta start scaling back their AI infrastructure spending in their forward guidance, that is the signal that matters. Those four companies are the demand engine. Everything downstream, including every memory order, traces back to their capex. The day their guidance turns down is the day the demand side of my thesis cracks. That is what I watch, every earnings season, more closely than the stock price.

Not the stock dropping. Not a scary headline. Not a China announcement. Hyperscaler capex guidance turning down. That is the one.

There are slower-moving things I keep an eye on too. CXMT eventually qualifying for high-end HBM. VC money flooding into memory-efficiency startups. But those are years out. The fast tripwire, the one that would make me act, is hyperscaler spending intentions.

The Macro Question I Cannot Answer or Predict

Someone will say what about a market crash. What about a recession. What about the vibe turning risk-off.

Here is my honest answer. Macro is always there. It is like taxes. It is a permanent background condition, not a reason to avoid owning something good. I cannot predict it and I have stopped trying.

But here is why I can hold through it. Because this is a re-rating backed by real earnings. In a market downturn this position will drop, maybe hard. But when the dust settles, the earnings will still be there. The supply will still be tight. The AI buildout will still need memory. A re-rating supported by actual cash flows recovers after a macro shock in a way that a hype bubble does not. The earnings are the floor under the re-rating.

If this were a story stock with no profits I would be terrified of a downturn. But Micron is printing cash. SK Hynix is printing cash. The strength shows up in the financials, not just the price. That is what lets me hold through the macro I cannot control.

Not Trimming and Preparing My Diamond Hands

Up 60% and holding. First marker at $80. Real exit signal is hyperscaler guidance turning down. Macro will come and go and I will hold through it because earnings are the floor.

I sold Tencent too early. I sold Nvidia too early. I am not going to make the same mistake a third time on the clearest structural thesis I have seen since either of them.

I hold a long position in DRAM ETF (ticker: DRAM). Not investment advice. Do your own research.

My current positions as of June 1, 2026. All MU and SNDK puts expired worthless and only left holding $DRAM ( ▼ 0.17% )