SK Hynix is preparing to bring a major piece of the AI memory trade closer to U.S. investors. The South Korean memory chipmaker, a rival to Samsung and U.S.-based Micron, is planning to sell nearly 17.8 million shares in a U.S. IPO, the company said on Monday.
The listing arrives at a moment when memory chips have become central to the infrastructure behind AI. Systems that run AI require large amounts of memory, and the scramble to build more AI computing capacity has pushed demand ahead of supply.
What SK Hynix Is Offering
SK Hynix will not be offering ordinary common shares directly to U.S. investors. Instead, the company plans to offer American depositary receipts, or ADRs.
An ADR is a certificate that allows U.S. investors to buy exposure to a foreign company without trading directly on an overseas exchange. In this case, each ADR will represent a tenth of a common share.
The securities are expected to price on Thursday and begin trading on Friday. If the offering performs well, there is a path to a very large raise: based on SK Hynix’s closing share price last Friday in Seoul, the company could raise around $28 billion, Bloomberg reports.
That potential scale matters because U.S. investors have been looking for more ways to participate in the companies supplying the AI buildout. Nvidia has become the defining name in that trade, but the hardware stack behind AI extends beyond graphics processors. Memory is one of the areas now drawing intense attention.
Why AI Has Changed The Memory Market
The reason SK Hynix is receiving this level of attention is simple: AI systems are highly memory intensive. They need chips that can store and move data quickly inside the systems that train and run AI models.
The source of demand is broad. Hyperscalers like Amazon, Microsoft, Google, and Oracle are racing to build so-called AI factories. At the same time, new AI data centers are multiplying nationwide.
That expansion has created pressure on several types of memory chips, including:
- High-bandwidth memory (HBM), used to move data rapidly inside AI systems.
- DRAM, a key memory technology used in computing systems.
- NAND, another form of memory used to store data.
Demand has outpaced supply, creating a shortage across these categories. The situation has been called “RAMageddon.”
The shortage is not just a problem for companies building AI data centers. Apple executives said the shortage is forcing it to raise prices on Mac computers and iPads. That shows how pressure in the memory supply chain can move beyond AI infrastructure and affect consumer technology as well.
The Numbers Behind The Surge
SK Hynix has already benefited from the current cycle. The company said its first-quarter revenues were up nearly 200% over the same quarter last year. Its stock is also up about 260% so far this year.
Those gains reflect how sharply investor expectations have shifted around memory. For years, memory chipmakers have been viewed through the lens of supply cycles, pricing swings, and capital spending. The AI boom has changed the immediate story by creating a surge in demand for the chips that move and store data inside AI systems.
Micron offers the closest U.S. comparison. Its stock has shot up nearly 700% over the past year to a more than $1 trillion valuation, fueled by record AI-driven memory demand and revenue.
That comparison helps explain why a U.S. listing from SK Hynix could receive strong attention. Wall Street is looking for another Nvidia, and memory chipmakers are among the closest options available. SK Hynix’s ADRs would give U.S. investors a more direct way to take a position in one of the major companies tied to that memory demand.
The Capacity Race Comes With Risk
SK Hynix and Samsung are also part of a much larger manufacturing push. South Korean tech companies, led by those two firms, have vowed to spend over $550 billion on building out new manufacturing capacity.
The logic is straightforward: if AI systems keep demanding more memory, suppliers need more factories and more output. Building that capacity could help address shortages in HBM, DRAM, and NAND.
But the strategy is not risk-free. New facilities take time to build. By the time they are ready, memory needs for AI may change. If that happens, companies could end up with more supply than the market wants.
That outcome could put pressure on prices. In memory markets, too much supply can quickly become a problem because pricing is central to profitability. The same expansion that looks necessary during a shortage can become dangerous if demand shifts before new capacity arrives.
What Investors Should Watch
The planned SK Hynix ADR offering is not just another cross-border listing. It is a test of how much U.S. investor appetite exists for memory companies tied to AI infrastructure.
Several facts will shape that appetite. SK Hynix is already showing strong revenue growth. Its stock has already climbed sharply this year. The AI data center buildout continues to pressure memory supply. And the broader market is still searching for companies that can benefit from the same demand wave that lifted Nvidia and Micron.
At the same time, the long-term risk is embedded in the same story. A shortage can support high prices and strong investor enthusiasm, but a major manufacturing expansion can eventually shift the balance. If AI memory requirements change or supply catches up too quickly, the market could look very different.
For now, SK Hynix is arriving in the U.S. market at a moment when memory has become one of the clearest hardware bottlenecks in AI. That timing may be exactly what makes the IPO stand out.