Investment Review №349. High Hopes

On Friday, July 10, SK Hynix (SKHY), Nvidia’s (NVDA) key memory supplier, made its Nasdaq ADR debut and raised $26.5bn. It was the largest U.S. market debut by a foreign company on record, surpassing Alibaba’s (BABA) $25bn listing in 2014. Strictly speaking, this was not an IPO. SK Hynix has traded in Seoul for nearly three decades. But in market terms, it looked and felt like one: demand was seven times covered, and the ADRs finished their first session up 13%. Then Monday arrived. SK Hynix shares in Seoul collapsed 15.4%, their worst day in almost 20 years. The KOSPI sank 9%, and trading was halted for 20 minutes.

There was no negative business update. There did not have to be one. This was not a fundamental selloff. It was a positioning accident. Part of it was the classic “buy the rumor, sell the news” trade. After a 630% rally over the past year, local investors used the Nasdaq listing as the moment to take profits. The ADRs were also trading at roughly a 26% premium to the Korean shares, inviting arbitrage flows back into the domestic listing. But the bigger story sits inside Korea’s leveraged ETF market. In January, regulators approved 2x single-stock ETFs, and positioning quickly became extreme. By late June, assets in those products had ballooned to $50bn. ETF dealers have to buy into strength and sell into weakness. On June 23, their rebalancing alone triggered roughly $6bn of forced selling — about 14% of the combined daily turnover in Samsung and SK Hynix. That matters because Samsung and SK Hynix together account for more than 40% of the KOSPI. When one of them cracks, the whole index feels it. The issue has now become political: on July 15, the ETF structure is expected to be discussed at a meeting with the president, while the industry is proposing to cut leverage to 1.5x to reduce the risk of another forced unwind.

And yet the fundamental story is moving in the opposite direction. SK Hynix controls 58% of the high-bandwidth memory market by revenue, versus 21% each for Samsung and Micron. This is the memory layer without which no AI accelerator works. The economics of the entire memory industry are also changing. ASML’s EUV lithography machines—the most expensive tools in chip manufacturing—used to be absorbed mainly by processor fabs. This year, for the first time, memory accounted for 51% of ASML’s sales. That is not a small shift. If the most advanced equipment in the semiconductor industry is moving toward memory, the market is being told very clearly where the AI supply chain is most constrained. SK Hynix CEO Kwak Noh-Jung expects a severe shortage in 2027 and demand to exceed capacity for another decade. Chairman Chey has promised to double capacity over five years, but customers are already telling him the same thing: it will not be enough. The proceeds from the Nasdaq listing will go toward buying more equipment from ASML.

The valuation does not scream bubble either. SK Hynix trades at about 5.8x forward P/E, below Micron (MU) at roughly 7x—the same “Korea discount” the Nasdaq listing was supposed to close. The next catalyst is already visible. In December, the stock is likely to be added to the Nasdaq-100, which would force passive funds to buy it regardless of market sentiment. Every IPO-style stock goes through this phase. Even SpaceX (SPCX) was not spared. That is why I look at these stories through a long-term lens. Volatility in the stock will eventually fade. After that, a second bullish wave could follow over a 12–18-month horizon. The SK Hynix case shows that IPO-style listings are drawing intense attention, which can quickly turn into outright frenzy. The shares then correct, but that correction is part of the process of building a stronger base for the next leg higher. And this principle will likely apply to every major “IPO-style” stock, including the brightest names still ahead—OpenAI and Anthropic.