SK Hynix’s Korean Listing Takes a Hit While Its US ADRs Command a 33% Premium — A Tale of Two Markets

SK Hynix shares slid 8% in Seoul's tech rout, while a $644M divorce ruling and $4.83B packaging bet highlight AI-driven demand and market disconnects.

A brutal session on the KOSPI sent SK Hynix shares sliding more than 8 percent on Friday, but the damage had little to do with the memory giant’s own fundamentals. The broader index cratered 5.72 percent as a tech-led selloff swept through Seoul, dragging the stock down to 1,759,000 won. For investors watching the company’s Nasdaq-listed ADRs — which trade under the ticker SKHY at a 33 percent premium to the Korean shares — the divergence underscores a growing disconnect between the two markets.

The ADR premium stems from a technical bottleneck: the Korea Securities Depository caps the conversion of local shares into new ADRs at 2.5 percent, creating artificial scarcity for US-based buyers. That structural constraint has kept the American depositary receipts trading well above their Seoul-listed counterparts, even as Korean retail investors piled into the dip on Friday, snapping up shares at the lower price.

A $644 Million Divorce Ruling That Actually Brought Relief

One of the week’s most eye-catching headlines involved SK Group Chairman Chey Tae-won, whom South Korea’s Supreme Court ordered to pay 944 billion won — roughly $644 million — to his ex-wife Roh Soh-yeong in what stands as the country’s largest-ever divorce settlement. The sum is substantial, yet market participants greeted the ruling with something close to relief. The final figure came in well below the plaintiff’s original demand, and crucially, the court mandated a cash payment rather than a transfer of SK Group shares. That removed the specter of Chey being forced to sell down his stake in SK Hynix to raise funds — a scenario that could have destabilized the conglomerate’s ownership structure.

Capacity Constraints and a $4.83 Billion Packaging Bet

The bigger story for SK Hynix remains the relentless demand for high-bandwidth memory, where it commands a 56.4 percent market share. The board recently approved a $4.83 billion investment to expand packaging capacity at its Cheongju site, specifically for the P&T7 facility. The move doesn’t alter the company’s overall capital expenditure budget; rather, it accelerates spending to bring clean-room capacity online sooner. CEO Kwak Noh-jung has been blunt about the industry’s central challenge: customer demand keeps climbing, but production capacity is hitting hard limits.

First-quarter results for 2026 already illustrated the pace of the memory super-cycle: revenue hit 52.6 trillion won ($34.5 billion), with a gross margin of 79.3 percent — a 198 percent year-over-year surge. The company reports second-quarter numbers on July 29, and consensus forecasts point to another massive jump, fueled by the AI chip boom.

Silicon Valley Diplomacy and a 70% Nvidia Stake

While the Seoul stock suffered, SK Hynix was quietly preparing for a high-profile diplomatic push. South Korean President Lee Jae Myung is currently touring San Francisco and Silicon Valley, where his delegation is scheduled to meet Nvidia CEO Jensen Huang and OpenAI chief Sam Altman. Government officials have signaled that SK Hynix is on the verge of announcing “significant” long-term supply agreements with leading US technology companies, likely centered on next-generation HBM chips and AI infrastructure.

The company already holds a commanding 70 percent share of orders for Nvidia’s forthcoming Vera Rubin platform — a position that makes it indispensable to the AI supply chain, geopolitical headwinds notwithstanding. Rising oil prices and tensions in the Middle East have weighed on broader market sentiment, but SK Hynix’s role as a critical link between memory production and hyperscaler demand has so far insulated it from the worst of the macro turbulence.

The Broader Chip Landscape: ASML, AMD, Intel, and Qualcomm

SK Hynix is far from the only semiconductor heavyweight navigating a volatile moment. ASML posted €9.3 billion in second-quarter net revenue and raised its 2026 guidance for the second time this year, now forecasting €43 billion to €45 billion. The Dutch lithography giant also confirmed that Intel Foundry has begun commercial production using High-NA EUV technology for its Panther Lake processors on the 18A node — a milestone for the entire chipmaking ecosystem. ASML plans to boost EUV and immersion DUV system capacity by 30 percent annually in 2027 and 2028 to meet AI-driven demand.

AMD, meanwhile, officially launched its Helios rack system and the Instinct MI455X accelerator at its “Advancing AI 2026” event in San Francisco, securing deals with Microsoft, Anthropic, OpenAI, and Meta. Despite controlling just 4.5 percent of the data-center GPU market versus Nvidia’s 95 percent-plus, analysts remain bullish: 51 of those surveyed rate AMD a “Strong Buy,” with Benchmark reiterating a $685 price target.

Intel delivered the week’s biggest surprise, posting 25 percent revenue growth in the second quarter — its strongest quarterly expansion in over 15 years. Data-center and AI revenue jumped 59 percent to $6.3 billion, though a GAAP net loss of $11 billion, tied to an accounting revaluation related to the CHIPS Act agreement, kept headline numbers messy. The stock has rallied roughly 48 percent above its 200-day moving average, though it eased 3.5 percent on Friday to €85.08.

Qualcomm finds itself in a more precarious position. Its Snapdragon platforms will power Samsung’s latest Galaxy Z Fold8 devices and new intelligent eyewear, but the company’s Dragonfly data-center push has yet to generate meaningful revenue. Of 37 analysts covering the stock, 22 rate it a “Hold,” and the shares have fallen 14.8 percent in the past 30 days to €147.88. The RSI of 37.4 suggests oversold conditions, but the market is waiting for proof of execution beyond the smartphone business.

What’s Next

The coming week will test the narratives across the sector. SK Hynix and Qualcomm both report earnings on July 29, offering a real-time read on HBM pricing trends and Chinese Android smartphone demand. For AMD, the focus shifts to whether Helios can ship in volume during the second half of the year — a question that depends on HBM4 supply and the timing of Nvidia’s competing Vera Rubin platform. Intel must show that its data-center momentum can hold despite ongoing wafer, memory, and substrate shortages. And ASML’s capacity expansion plans will provide a crucial signal on whether the current lithography cycle has legs beyond the initial AI buildout.