Key points
- My theory: SKHY pops Friday. Micron gets sympathy love, the same way it just moved on zero news of its own tonight. Then memory goes flat or red until real earnings hit.
- Leopold Aschenbrenner's fund just anchored the SKHY deal even though its single biggest disclosed position is a $2 billion put against the whole chip sector.
- Tonight's Samsung earnings reaction already knocked Korean memory stocks down again. I think that plus rotation into Friday's ADR keeps Kospi soft into Thursday.
- Update, July 16: I sold near $190 on the 14th and bought back today at $154, same position at a lower cost basis heading into SK Hynix's July 29 earnings.
Here's the theory I posted on July 7 for the next few days in chips. I told you to write it down and hold me to it. A week later it mostly played out, so I've left it exactly as written below and timestamped my trades at the bottom. Here's what I said would happen.
Kospi stays soft into Thursday as Samsung and SK Hynix's Korean shares keep bleeding, some of it rotation into Friday's new ADR. Then $SKHY lists Friday. It pops. Micron (MU) rides the wave too even though nothing actually changed at Micron. That's just how sympathy trades work, and we already saw the mirror image of it tonight. Then the excitement drains out and memory stocks go flat or slide for a bit while everyone waits on the next real catalyst. After that it's a whole new story. Good or bad, it comes down to what the actual earnings say.
I'm buying SKHY the moment it's tradable because I think stage one of that theory is the easy part to get right.
Tonight's proof of the sympathy trade
Samsung just reported preliminary Q2 profit up about 18-fold year over year, its best quarter ever, and the stock opened down roughly 5% and kept sliding, closing the Seoul session down almost 7%. A surprise employee bonus provision ate into the beat, on top of a stock that had already run hard into the print. Micron didn't report anything tonight. It fell in sympathy anyway, down as much as 3.7% in the initial after-hours reaction and running closer to 5% once the regular session got going, right alongside SanDisk and Western Digital, while Nvidia, AMD and the Nasdaq 100 barely moved. We broke down the full board here. That's the exact mechanic I'm betting runs in reverse Friday. No news at Micron, just a ride on whatever SK Hynix does.
Why Kospi keeps bleeding into Friday
Korea's market has now gone red two sessions running. Kospi crashed almost 8% on July 2, then ripped back 5.76% on July 3 to 8,088.34. It slipped again Monday, down 0.46% to 8,051, then fell hard again today, down 4.91% to 7,656.31, essentially erasing the entire Friday rebound. My theory from here: some of that ongoing weakness is money leaving the Korea-listed shares early to buy the more liquid ADR once it opens, so I expect Kospi to stay soft into Thursday's final pricing, then steady out once the ADR itself is live and can absorb that demand directly.
The Leopold contradiction
Leopold Aschenbrenner ran research at OpenAI before starting his own fund, Situational Awareness. We dug into this fund before. His fund is one of three cornerstone investors in the SKHY deal, together good for up to $7 billion. Here's the twist. His own single biggest disclosed position is a $2 billion put against the entire semiconductor ETF, with more puts stacked against Nvidia, Broadcom, AMD and Micron specifically. So he's writing a huge anchor check for SK Hynix while betting against basically everyone else in the space.
I don't think that's him flip flopping. I think it's him separating "chips in general" from "the one company that makes the memory Nvidia's chips actually need." Estimates vary by source, but SK Hynix builds somewhere in the high 40s to upper 50s percent of the world's HBM supply, more than anyone else in the business. If you want AI memory exposure without the rest of the sector, SK Hynix is close to the only way to make that trade.
Can it actually save the industry
Probably not by itself. But tonight's memory-specific selloff, on genuinely record numbers, tells you this market is nervous, not convinced. SKHY debuting into that same stretch is a real pressure test. If it holds up, that's a signal the memory story still has legs into SK Hynix's own earnings on July 29. If it doesn't, that tells you the correction has more room to run than the bulls admit.
What I'm doing about it
I'm watching Kospi the next few days. If Samsung and SK Hynix keep sliding into Thursday, that confirms the rotation story and makes me more confident in Friday's pop.
I can't buy SKHY today. It isn't listed yet. Friday I plan to buy the pop. Not chase it too high. Not marry this position either. If my theory's right the easy money is stage one. Stage two I'll sit through or trim into. Stage three, the earnings reaction on July 29, is its own bet for later.
Update, July 10: The order filled, just not the way I planned. Had a limit in at $160 that never came close, the stock never traded that low today. Waited for a small dip off the opening spike and paid around $170 instead, more than I wanted to pay. Stage one is done. Now I sit through stage two like I said, or trim into strength if it runs too hot too fast. Stage three, the earnings reaction on July 29, is still its own bet for later.
Update, July 16: Stage two played out and I traded it instead of just sitting through it. SKHY spiked toward $190 on the 14th, which was too far too fast for a stock this new, so I sold the whole position there. Then the broad memory selloff did its thing and dragged SKHY back down with everything else, and I bought my shares back today at $154. Same position, cost basis well below where I sold, and about $20 a share banked in between. Stage three, the July 29 earnings, is still the real bet, and now I am sitting in it a lot cheaper than I started.
This is general market commentary and opinion, not investment advice. Markets can go down as well as up, and you can lose money. Always do your own research and consider speaking with a licensed financial professional before making any investment decision.
