SK Hynix's first earnings report after its Nasdaq debut is not just a victory lap. It is a test of whether investors still believe the AI memory boom can support prices, margins, and a $26.5 billion U.S. listing.
SK Hynix reports second-quarter earnings on July 29, and you don't need to be a memory-chip specialist to see why the market is jumpy. The company sits in the hottest part of the AI supply chain, high-bandwidth memory for Nvidia-class systems, but the stock has already been through a sharp July wobble. These numbers have to do more than look good. They have to calm people down.
The forecasts are still large enough to make ordinary chip cycles look small. TrendForce reported this week that Korea Investment & Securities expects second-quarter revenue of 80.9 trillion won and operating profit of 60.4 trillion won, while BLOTER, citing FnGuide data, put SK Hynix's expected operating margin near 77%. That would sit above the 72% margin SK Hynix reported in the first quarter, a level TrendForce said already outpaced TSMC and Micron.
That is not normal profitability. It is what happens when one product category becomes a bottleneck for the whole AI buildout.
The product is HBM. SK Hynix and Nvidia announced a multi-year technology partnership on June 7 covering next-generation memory for Nvidia's AI infrastructure roadmap, including Vera Rubin AI supercomputers and Vera CPUs. Yonhap reported in January that Nvidia had allocated about 70% of its HBM4 demand for the Vera Rubin platform in 2026 to SK Hynix, with samples already supplied. That is a real advantage, and it is not the sort of advantage Samsung or Micron can erase by issuing a louder press release.
Here is the problem. Investors have started asking whether high prices are still a sign of strength, or whether they are becoming a strain on the customers paying for all this infrastructure. Chey Tae-won, SK Group's chairman, has been unusually direct on that point. Bloomberg reported in March that he expected the memory shortage to last another four to five years, with supply unlikely to fully meet demand until around 2030. Chosun Biz reported that he repeated that view at Computex in Taipei on June 2 and said SK Hynix plans to double memory wafer capacity within five years.
Shortage cuts both ways. It gives SK Hynix pricing power today. It also tells you that relief will be slow, expensive, and operationally messy.
The selloff was about price discipline
The July slide did not come from a sudden collapse in AI demand. It came from fear that the trade had run ahead of itself. The Financial Times reported that U.S. semiconductor stocks suffered their worst week since April 2025, with the Philadelphia Semiconductor Index down nearly 10% and more than 20% below its June peak. Memory names were right in the middle of that selloff.
SK Hynix's new U.S. ADRs made the move feel even sharper. Barron's reported that the ADRs jumped as much as 8.3% on July 17 after a 14% drop the previous day, a move amplified because South Korea's market was closed for Constitution Day. MarketWatch reported on July 24 that the ADRs were trading at a large premium to the Seoul shares after Korea Securities Depository limited new ADR creation to 2.5% of outstanding shares. That premium is not a footnote. If you buy the U.S. line, you are paying for access as well as earnings.
Long-term supply agreements are the other pressure point. TrendForce said the expanding use of LTAs is reshaping SK Hynix's earnings outlook because more memory pricing is being set through customer agreements rather than pure spot-market moves. Mirae Asset Securities, cited by Financial Post and summarized by TrendForce, cut its operating profit estimate after lowering DRAM and NAND average selling price forecasts. That is not the same as saying demand has cracked. It says some of the upside is already spoken for.
Frankly, that is the distinction investors need to keep straight.
The Nasdaq debut raises the stakes
SK Hynix began trading on Nasdaq on July 10 under the ticker SKHY. Reuters reported that the company raised about $26.5 billion by selling 177.9 million American depositary shares at $149 each, making it the largest first-time U.S. share sale by a foreign company and topping Alibaba's $25 billion 2014 debut. Nasdaq said the shares opened at $170.
That timing matters. The company came to U.S. investors while the AI memory story was near peak excitement. Now its first earnings report as a Nasdaq-listed company arrives after a market scare, with hyperscaler spending under the microscope. Barron's reported this week that Micron and SK Hynix rose after Alphabet lifted its 2026 capital expenditure guidance to $195 billion to $205 billion, up from its previous $180 billion to $190 billion range. You can see the chain clearly: Alphabet spends, Nvidia sells systems, SK Hynix ships memory.
Samsung gives the other side of the picture. Reuters reported that Samsung estimated second-quarter operating profit of 89.4 trillion won, up 1,810% from a year earlier, yet its shares still fell on worries that AI infrastructure spending could slow. Good numbers are no longer enough by themselves.
SK Hynix still has the cleaner story. It has the Nvidia relationship, the HBM lead, and a shortage that its own chairman says could run to 2030. The July selloff was a valuation argument. The July 29 earnings report is a fundamentals argument. Right now, the company needs to show that those fundamentals are strong enough to carry the price investors have already paid.
Also read: Intel posts its fastest revenue growth in 15 years and still can't build chips fast enough • A $700 Billion Sovereign Fund Just Made the Chinese AI Cost Argument Impossible to Ignore • Meta and BlackRock's $12 billion El Paso bond deal shows the AI arms race is finally moving the credit market
