My position: I own SKHY. I bought back in at $154 after selling near $190 on July 14, and we laid out on Sunday why I'd take it over Nvidia and Micron. It was $126.78 Wednesday midday, so I'm down about 18%. I am an investor and not a financial advisor.
Key points
- Operating profit was 60.5426 trillion won ($41.6 billion USD) against a 64.09 trillion won estimate, a 5.5% miss on the best quarter the company has ever had.
- Margin landed at 76%, right where analysts had it, so the whole shortfall was revenue. Some HBM4 volume meant for the June quarter shipped into the second half instead.
- Long-term supply deals are signed with more than 10 customers, running 3 to 5 years and covering an estimated half of revenue.
- The company says supply requests keep piling up while it holds spending in check, the opposite of what the US hyperscalers keep saying.
- Seoul took the stock down 19.6% intraday and closed it off 9.61%. SKHY was down 2.6% at midday.
I was signed into the SK Hynix earnings call last night, watching SKHY drop in real time and trying to figure out what I was looking at. The stock ran up with futures, rolled over into the red, then did the whole thing again. I went in expecting to feel sick about it. I came out a little more comfortable than when I went in, which isn't how these usually go.
Here's what actually happened. SK Hynix made 60.5426 trillion won ($41.6 billion USD) in operating profit in three months. That's up 557% from a year ago. Seoul sold it anyway, because the estimates were 64.09 trillion won ($44.1 billion USD). Revenue came up 5.6% short. Both numbers are still company records.
The miss was shipping dates, not the business
This is the part that took sitting through the call to get. Their operating margin came in at 76%, exactly where analysts had it. So they didn't get squeezed on price. What slipped was how much product actually went out the door. Some HBM4 volume meant for the June quarter is shipping in the second half instead, and HBM4 is the expensive stuff, so pushing it back dragged down the average price of everything they sold.
Those are two very different misses if you own the stock. A cracked margin means the memory cycle is rolling over. That's the one that should scare you. A shipment sliding a few weeks means they sold it and book it later. That's a calendar problem. HBM4 yields are already running at HBM3E levels and production ramps from here. We broke the full print down Tuesday night, including the weirdest line in the release, a net profit bigger than revenue.
Five years of it is already sold
Song Hyun-jong, SK Hynix's president, told the call the company had "wrapped up LTA negotiations with more than 10 customers." Those are long-term supply deals. They run 3 to 5 years, the price is locked into an agreed range, and some come with deposits or purchase guarantees up front. Korean brokerages think they cover about half of revenue.
The release is blunter than the selling was. As the big tech companies pour money into AI, it says, "additional supply requests continue to mount." Customers are paying for that out of what they earn selling AI services, so "the momentum in memory demand is expected to persist." Then there's the spending line. SK Hynix says it'll build capacity "while maintaining capital expenditure discipline." That's close to the opposite of what the American hyperscalers keep saying, where the capex number itself has turned into a reason to sell the stock.
That's why I'm still in this. Everyone's arguing about whether 2027 turns into an oversupply year. A five-year deal signed in 2026 runs to 2031, and half the revenue already has a customer, a volume and a price attached to it. It doesn't guarantee the profit, since costs move and prices float inside those ranges. But it beats guessing.
Seoul dumped it, New York took it better
Seoul opened strong, then fell all the way to 1,246,000 won ($858 USD). That's 19.6% under Tuesday's close. It finished at 1,401,000 won ($964 USD), down 9.61%. The Kospi closed down 5.98%.
New York took it a lot lighter. SKHY closed Tuesday at $130.17 and was $126.78 by midday Wednesday, down 2.6%. Seoul lost 9.61% on the same news. Jim Cramer said the bears were "really trying to take apart a good story with Hynix." Worth saying though: Tuesday's crash in Seoul was about Chinese memory supply and had nothing to do with this print.
The part US holders keep forgetting: SKHY isn't the Korean stock. It's a receipt, and ten of them equal one Seoul share. Run that math at Wednesday's close and the Seoul shares are worth $96.42 per receipt. SKHY trades at $126.78. You're paying about a 31% premium for the convenience of buying it in dollars. I'm long the stock and long that premium, and only one of those has anything to do with memory chips.
What I'm doing about it
Nothing. I'm not selling into a 19.6% flush at a company that just locked up five years of demand. Four Korean brokerages cut their price targets after the print. All four kept a buy rating.
The 2027 HBM contract talks are the next thing I'm watching. Regular DRAM prices moved about 30% this quarter, and the company said market conditions could feed into those negotiations. That's the number that tells me whether I'm early or just wrong.
Figures and quoted company language are from the SK hynix 2Q26 results release. Won at 1,453 to the dollar. SKHY prices intraday as of 12:02 p.m. ET, July 29, 2026. None of this is investment advice.
