- Lead. SK Hynix confirmed plans on 24 June to raise approximately $29 billion by listing American depositary receipts on the Nasdaq on 10 July — a deal that would rank among the largest US equity offerings in history.
- Fact. The chipmaker holds roughly 60% of the global market for high-bandwidth memory, the component at the heart of AI accelerators built by Nvidia and others, and posted a record 37.6 trillion won operating profit in Q1 2026.
- Stake. All proceeds are earmarked for new fabrication capacity: the Yongin semiconductor cluster and a Cheongju advanced-packaging facility purpose-built for next-generation HBM production.
The structure and timeline
SK Hynix’s planned Nasdaq offering — valued by Bloomberg at 45.45 trillion won, or roughly $29.4 billion at current exchange rates — would use American depositary receipts, with each common share represented by 10 ADRs, allowing dollar-denominated trading for US institutional investors. The company filed a confidential S-1 with the Securities and Exchange Commission in March 2026 and publicly confirmed the deal in late June.
The listing timetable has book-building running ahead of a 10 July Nasdaq debut, with subscription and payment completing 14 July. New Korean shares arising from the offering are set to begin trading on the Korea Exchange on 29 July. Citigroup, JPMorgan, Goldman Sachs and Bank of America are the lead underwriters.
The strategic case for a US listing
SK Hynix’s rationale is explicit: the company wants to trade alongside Micron Technology on the Nasdaq and capture the premium that US markets are assigning to AI semiconductor exposure. The company recently overtook Samsung to become South Korea’s most valuable listed company, but Korean equity markets have struggled to reflect the full value of its AI memory dominance. The KOSPI’s sharp sell-off earlier this month illustrated how domestic market volatility can compress valuations even for structurally well-positioned exporters.
The company’s position in high-bandwidth memory is the core of the investment case. HBM is a stacked chip architecture that sits alongside Nvidia’s H100 and H200 graphics processors in AI data centres, providing the memory bandwidth that large language models require. SK Hynix holds approximately 60% of global HBM supply; its nearest competitor, Samsung, has faced yield challenges on newer generations. On 17 June the company began shipping samples of its HBM4E chip in a 12-layer configuration, the most advanced format currently available.
Proceeds and capacity plans
The entire net amount from the offering is designated for capital expenditure. The primary destination is the first fabrication plant of the Yongin semiconductor cluster south of Seoul, a four-fab campus where the first building is due for completion in 2027. A second tranche of spending will go to the Cheongju P&T7 facility, an advanced-packaging fab specifically designed for HBM production that requires extreme-ultraviolet lithography equipment.
The scale of the commitment reflects a broader industry bet that AI infrastructure spending will sustain demand for HBM through the end of the decade. If the offering prices at the top of its range, the Nasdaq listing would surpass even SpaceX’s record-setting June debut as the largest US IPO in history — a context that underscores how rapidly the AI-adjacent capital market has expanded in 2026.