One retail account, one frontier-tech thesis, documented in public. This week I nearly trimmed my biggest winner days before it printed the blowout of the year. The trade that didn't happen taught me more than the ones that did.
Last Friday, Micron popped 5% to a fresh record high and I came within a click of selling my entire position. The logic was real: it was my largest holding, it had run from the low-$300s to over $1,100, and a concentrated position that size makes me twitchy. I'd talked myself into it. Then the order didn't complete, the weekend came, and Monday a global memory-chip rout hammered MU 13% in a single session. For about 48 hours, not selling looked like a mistake.
Then Wednesday night happened. Micron reported revenue of $41.5 billion — up 346% year over year — beat EPS by 24%, posted a record 84.9% gross margin, and guided next quarter to $50 billion against a $43 billion estimate. It signed sixteen multi-year customer agreements and a memory-supply deal with Anthropic, and the CEO said the supply tightness is "locked in to persist beyond 2027." The stock ripped 12–15% after hours toward $1,180 — a new all-time high.
I held every share. My book crossed +58% and a new high tonight. And the lesson isn't "diamond hands always win" — it's the opposite of cocky. It's: I nearly let position-size anxiety override a thesis I had no fundamental reason to abandon. That's the trade I need to learn from, even though it worked out.
The map before the positions. ● firing ● mixed/watching ● next-wave seed ● headwind.
It was a two-act week. Act one: a global memory rout, sparked by SK Hynix slowing advanced-chip output and AI-spending skepticism, dragged the whole semi complex down — Micron fell 13% Tuesday alone. Act two: Micron's print blew the skepticism apart after the bell Wednesday and dragged the sector right back up. The fear and the answer arrived in the same five days.
Held (barely): the full MU position through a near-sale and a 13% rout — into a blowout. Trimmed: SPCX to one share near the highs; it's since fallen 26% off its peak, so the call holds. Next: right-size MU back toward ~15% of the book once the post-earnings euphoria settles — not at the gap-up. Same discipline that trimmed it at +77%, applied at +266%.
"The chip rout said AI spending would never pay off. Two days later Micron guided to fifty billion dollars a quarter. The bears didn't get a thesis — they got a tantrum."
For 48 hours this week the narrative was that the AI trade was over: SK Hynix slowing output, hyperscaler capex with no returns, a global memory selloff. Then the largest pure-play memory company on earth posted 346% revenue growth, record 85% margins, sixteen binding multi-year contracts, and a guide that obliterated estimates by seven billion dollars. The CEO said tightness is locked in "beyond 2027."
That's the difference between a valuation wobble and a demand problem. The rout was the former dressed up as the latter. The earnings are real, the contracts are signed, the margins are at all-time highs — and a hawkish Fed doesn't change any of that, it just changes the multiple you pay for it. I'll keep saying it: I get cautious when the orders get cancelled. This week the orders got locked in for five years. The tantrum is not the thesis.
Here's the honest part, and it cuts against the win. I almost sold my whole Micron position out of pure position-size anxiety — not because anything in the thesis had broken. It hadn't completed, the stock dipped, then it blew the doors off and I look like a genius for holding. But I didn't hold on conviction. I held by accident. If that order had gone through Friday, I'd have written a very different issue this week, and I'd have been telling you why selling the top was disciplined.
That's the trap of a good outcome: it can launder a sloppy process into a "great call." So I'm marking it plainly — the result was lucky, the near-decision was emotional, and the actual lesson is that I let a 22%-of-book position get that big without a plan, then nearly panic-trimmed it at the worst possible moment. Now MU is even more oversized after the pop, and the grown-up move is to right-size it deliberately once the euphoria cools — not to high-five myself and let it ride to 30% of the account.
You get the win and the self-critique in the same breath. That's the only way this stays honest.
Eighty-five days. $5,270 deployed, ~$8,360 on the board, a new all-time high at +58%, and a Micron position that just printed the blowout of the year — $41.5B in revenue, an $50B guide, 85% margins, contracts locked beyond 2027.
But the issue isn't the win — it's the wobble behind it. I nearly sold the steak right before the feast, out of anxiety rather than analysis, and now that same position is a fifth of my whole book. The thesis has never been stronger; the sizing has never been more lopsided. So the plan is simple and unglamorous: hold the conviction, trim the concentration deliberately after the pop settles, watch tomorrow's PCE print, and keep the SPCX lesson close — froth deflates, fundamentals don't.
Issue #3 was the dip that got bought. Issue #4 is the trade I almost made and the discipline I almost skipped. The market rewarded me anyway. I'd rather tell you it was luck than pretend it was genius.
See you next Sunday, buddy. — Alex