SK Hynix record $26.5B US Nasdaq stock offering

Most large US listings price at a discount to lure buyers. SK Hynix did the opposite. It priced its 177.9 million ADRs at $149 each, a roughly 2.9% premium to its Korea close of 2.19 million won ^[1], in what the deal's backers described as the only premium-priced IPO of its kind in US market history. The structure is a 10-to-1 ADR - ten receipts to one Seoul-traded common share - so the $149 tag maps onto a fraction of the Korean price, and the premium is the tell: buyers were willing to pay up rather than demand a sweetener.

The scale is the headline. At about $26.5 billion, the deal is the largest-ever US IPO by a foreign company, the largest ADR offering on record, and the second-biggest US IPO behind SpaceX's roughly $85.7 billion offering ^[1]. It tops the previous ADR record, Alibaba's roughly $25 billion 2014 raise ^[1]. Demand ran hot enough to justify the premium: book-building drew orders more than seven times the ADRs on offer ^[1], and cornerstone investors Baillie Gifford, Coatue Management and Situational Awareness Partners indicated interest of up to $7 billion between them ^[2]. Bank of America, Citigroup, Goldman Sachs and JPMorgan ran the book ^[1]. The point of the raise is concrete, not financial engineering: proceeds are earmarked to expand chip-making facilities in South Korea and buy equipment including ASML's extreme ultraviolet scanners ^[1].

Strip away the deal mechanics and this is a bet on one product: high-bandwidth memory, the stacked DRAM that feeds AI accelerators. HBM has been called the scarcest commodity in AI computing, with demand outrunning supply, and SK Hynix commands roughly 58-60% of the market ^[2][3]. That is why the raise reads as an anchor customer's insurance policy as much as a capital event. SK Hynix supplies the HBM inside Nvidia's H100, H200 and Blackwell GPUs and is co-designing next-generation HBM4, a partnership Jensen Huang reinforced with a June visit and a multiyear commitment ^[3]. Hyperscalers like Google and Microsoft are buying as much memory as they can secure ^[3], which is exactly the dynamic that lets a chipmaker price an IPO at a premium.

The fundamentals under the story are unusually strong for a memory company, which historically earns thin, cyclical margins. Q1 2026 revenue came in around $34 billion, up 198% year over year, with a 79% gross margin, a 72% operating margin, and a forward P/E of roughly 8 ^[3]. TrendForce analyst Ellie Wang argues that HBM leadership "has positioned SK Hynix as one of the biggest beneficiaries of the rapid growth in AI infra" ^[4]. In other words, the Nasdaq listing is less about needing the cash and more about planting a flag where the AI-trade capital lives.

The tension in this deal is whether SK Hynix is a structural AI winner or a cyclical memory stock dressed up as one. The bull case leans on scarcity and lock-in. The bear case leans on history: the memory industry is infamous for boom-and-bust cycles, and the capacity this raise funds could itself fuel the next oversupply ^[5]. Capital Economics' James Reilly points to the price action as a warning, calling the volatility "evidence of excessive froth" that "calls into the question the sustainability of this rally" ^[5]. The numbers back the nerves: across 127 trading days in 2026 the Korea-listed stock moved more than 1% on 103 occasions, with 53 days swinging more than 5% ^[6], after a 12-month run of roughly 700-770% ^[2][7].

On Reddit, the community read sharpened this into a specific contrarian thesis. Retail voices flagged the irony of selling shares near a possible cycle peak to fund the very capex that ends the shortage, and questioned whether the fabled boom-bust pattern still applies when a large share of volume and pricing is said to be locked into long-term agreements. That framing - demand-durability as the real debate, not deal size - is the sharper way to read the risk than the froth-versus-fundamentals headline. Meanwhile Samsung and Micron have both secured HBM4 certification ^[8], a reminder that SK Hynix's lead is defensible, not permanent.

Calling this an IPO is convenient shorthand, but the community-level discussion homed in on what it actually is: a cross-listing of an already-public company, not a fresh float. That distinction matters for how the price behaves. The dominant analyst thread on YouTube framed the move around an "accessibility discount" - the idea that Korea's emerging-market index classification and the friction of accessing its market have long kept SK Hynix trading at a discount that a US listing could narrow. Wedbush's Dan Ives captured the mainstream optimism, calling the debut "a positive indicator of the AI trade, with Korea chip plays now front and center" ^[4].

The skeptics push back on the arbitrage. If ADRs and Korean shares are fungible, any US premium should compress against the home-market price rather than persist, so the premium pricing is a snapshot of demand, not a durable re-rating. A quieter contrarian read holds that the Korea discount may be structural - rooted in governance, geopolitics and index constraints - and therefore not something a Nasdaq ticker simply arbitrages away. For a general reader the takeaway is that the record-breaking headline number is real, but the more interesting question is whether US access changes the stock's long-run valuation or just its trading venue.

Because retail investors were not offered the ADRs, the sharpest community discussion turned to proxies and second-order plays rather than the deal itself. On Reddit, one recurring "picks and shovels" argument held that the durable exposure is not SK Hynix but its supply chain - the equipment makers behind the EUV and etch/deposition steps and the specialty-materials suppliers that every HBM stack depends on. That maps directly onto SK Hynix's stated use of proceeds: expanding facilities and buying gear such as ASML's EUV scanners ^[1], spend that flows to the toolmakers regardless of which memory vendor ultimately wins.

The retail conversation across X, YouTube and Reddit was overwhelmingly momentum-driven and positive, fixated on the record scale and the mechanics of a premium-priced ADR, but the most useful signal was skeptical rather than euphoric: debate over which listed proxy best captures the theme, and whether the equipment and materials layer offers steadier exposure than a single memory maker priced for perfection. It is a reminder that when a $26.5 billion deal is closed to the crowd, the crowd's attention migrates to whatever it can actually buy - and that hunt often surfaces the more resilient part of the value chain.