SK Hynix plans to list ADRs in the U.S. as early as August, with total fundraising reaching up to $14 billion.

Memory giant SK Hynix plans to go public in the U.S. by issuing American Depositary Receipts (ADRs) as early as August. Reuters cited two sources familiar with the matter, indicating that the SEC expects to complete its review as soon as the week of June 22; market estimates suggest this fundraising could reach up to $14 billion.
(Background: The Korean stock market rebounded 8% in one day to recover from a crash! SK Hynix surged 16%, Samsung rose 9%, and Jensen Huang joked, "You should be happy about the discount.")
(Additional context: Goldman Sachs and Morgan Stanley are competing to be lead underwriters for OpenAI and Anthropic IPOs! Hidden returns could reach $7 billion.)

Table of Contents

Toggle

• From Seoul to Wall Street: The Significance of an ADR Application
• HBM Turns Memory from a Cyclical Drag into a Computing Bottleneck
• The $14 Billion Fundraising Premise Is That AI Capital Expenditures Won't Cool Down

Memory used to be the least glamorous business in the semiconductor industry. Overcapacity meant bleeding losses; only when the economy improved could companies break even, and they chased the cyclical waves for twenty years. Now, SK Hynix, which is profiting handsomely from HBM, is preparing a large IPO-level fundraising plan to open the doors to Wall Street.

From Seoul to Wall Street: The Significance of an ADR Application

Reuters reported, citing two sources, that SK Hynix’s U.S. depositary receipt (ADR) application is progressing smoothly. The SEC is expected to complete its review as early as the week of June 22, with a potential listing as soon as August.

In simple terms, an ADR is a certificate that allows foreign companies’ stocks to be traded on U.S. stock exchanges, effectively opening a direct buying channel for overseas investors.

SK Hynix responded that they are still preparing to issue ADRs, aiming to list by the end of the year, but the size and exact timing are not yet determined. In fact, the company submitted a confidential filing to the SEC in March, clearly planning to list in the U.S. in the second half of the year. Market estimates suggest the total fundraising could exceed $14 billion.

Why go to the U.S.? There are three specific reasons.

First, Nvidia, the largest customer, is based in the U.S. Listing in the U.S. can strengthen capital ties with the American AI ecosystem and make it easier for institutional investors to directly hold shares of this core supply chain company.

Second, U.S. stocks offer deeper liquidity and higher valuation multiples, helping lock in capital during high-growth cycles.

Third, an ADR is also a brand statement: this company is not just an Asian memory manufacturer but a core supplier of AI infrastructure.

HBM Turns Memory from a Cyclical Drag into a Computing Bottleneck

To understand SK Hynix’s current position, one must first grasp the importance of HBM. High Bandwidth Memory (HBM) is essentially multi-layered memory stacked together and placed close to AI chips, serving as a critical component for running large language models on AI servers.

SK Hynix’s dominance in HBM is backed by numbers: about 57% of the global HBM market in Q4, maintaining an estimated 70% to 80% overall share, with the new generation HBM4 expected to hold over 50% market share by 2026. As Nvidia’s AI chips are shipped more, their orders become more concentrated. This relationship has shifted from “component procurement” to “strategic dependence.”

The key point is that HBM has rewritten the pricing logic of memory. Historically, this industry was known for cyclical losses, but HBM’s complex manufacturing process and high yield thresholds mean that expanding capacity isn’t just a matter of increasing production. During the explosive growth of AI computing power, HBM has become a truly scarce resource, and the pricing of scarce resources is entirely different from that of ordinary commodities.

This explains why SK Hynix’s stock price has surged over 210% this year, with a market cap surpassing $1 trillion at the end of May, making it the third Asian company after TSMC and Samsung Electronics to cross this threshold.

The $14 Billion Fundraising Premise Is That AI Capital Expenditures Won’t Cool Down

However, there is a crack that cannot be ignored. The valuation at which SK Hynix plans to raise funds in the U.S. is based on two assumptions: that AI clients like Nvidia will continue to increase capital expenditures rapidly, and that HBM supply and demand will not reverse.

Currently, both assumptions still hold. AI infrastructure investments by Microsoft, Amazon, Google, and Meta are still accelerating, and Nvidia’s order visibility extends into next year. Competitor Micron is also benefiting from HBM price increases but still lags behind in market share and technological accumulation by a generation.

However, market pricing has always moved faster than fundamentals. A market cap exceeding $1 trillion and a doubled stock price imply that investors are already pricing in significant future profits. The $14 billion fundraising and August listing essentially ask: are global investors willing to bet today’s valuation on continued expansion of AI computing demand for how many more years? If HBM supply and demand loosen or AI capital expenditures unexpectedly slow, the fragility of this logic will be exposed.