The global memory chip giant SK Hynix Inc., one of the world's top three memory chip manufacturers and the leader in market share for High Bandwidth Memory (HBM), has filed a revised F-1 registration statement with the U.S. Securities and Exchange Commission (SEC) to list its shares on the Nasdaq exchange in the United States.
The company plans to list its American Depositary Shares (ADSs) under the ticker symbol "SKHY." Each ADS will represent a portion of the company's ordinary shares listed on the Korea Exchange's KOSPI market, where the shares have a par value of 5,000 Korean won each.
In its latest filing, the company stated that on June 24th, its board resolved to issue a maximum of 17.79 million new ordinary shares in connection with this offering. This represents approximately 2.50% of the total 712.702 million issued and outstanding ordinary shares as of the resolution date.
The net proceeds from this offering are intended to fund a portion of the company's 45.5 trillion won capital expenditure plan related to a major capacity expansion facility for memory chips in South Korea. The company also plans to use funds to procure expensive EUV lithography equipment, with an estimated cost of about 11.9 trillion won, with bulk deliveries expected by December 2027.
SK Hynix further anticipates using cash flow from operations, existing and future credit facilities, certain borrowings under debt securities, and other financing resources to cover the portion of the total funding required for its major capacity expansion infrastructure that exceeds the net proceeds from this U.S. listing.
Wall Street financial giants including Bank of America, Citigroup, Goldman Sachs, and JPMorgan are set to act as global coordinators for this sizable IPO. The company stated last week that it expects its shares to begin trading on U.S. markets on July 10th.
The company aims to raise approximately 45.45 trillion won (about $29.4 billion) through this listing.
The AI-Driven U.S. Listing and Global Capital Flows
Based on the proposed fundraising and issuance size, and depending on exchange rates, SK Hynix's U.S. ADS offering would rank among the top three initial stock offerings in global market history by size. According to compiled institutional data, it would be comparable to Saudi Aramco's massive $29.4 billion IPO in 2019. The largest IPO globally remains SpaceX's unparalleled fundraising of at least $75 billion.
SK Hynix is the core supplier of High Bandwidth Memory (HBM) systems to AI chip leaders like NVIDIA Corp (NASDAQ: NVDA). This U.S. listing plan marks a significant milestone in the company's remarkable ascent. After becoming the preferred HBM supplier for NVIDIA, which holds a dominant position in AI chips, it has become the global market share leader for HBM systems. This has allowed it to surpass its long-time Korean rival, Samsung Electronics Co., Ltd., in market capitalization and overall DRAM share at times.
The timing of this U.S. ADS offering follows a staggering approximately 850% surge in SK Hynix's shares traded on the Seoul market over the past 12 months, which propelled its market capitalization above 1 trillion won, briefly exceeding that of Samsung Electronics, South Korea's long-standing largest company by market value.
Investor sentiment around semiconductor companies tied to the AI computing infrastructure boom has been volatile recently, with the year's overall gains interspersed with sharp swings in both directions, due to increasingly crowded positions and growing leveraged strategies.
SK Hynix is actively seeking to raise 45.45 trillion won ($29.4 billion) through a U.S. stock market listing to capitalize on robust global investor demand for shares of the high-flying and supply-constrained memory chip giant.
The influx of capital into SK Hynix's U.S. listing plan, coupled with the exceptionally strong earnings and outlook just reported by U.S. memory chip giant Micron Technology, Inc. (NASDAQ: MU), highlights that the memory chip super-cycle is far from over.
The combination of SK Hynix's plan to raise about $29.4 billion and Micron's stellar recent performance and guidance significantly reinforces a firm optimistic cyclical view in the market. This view posits that the so-called "memory super-cycle" is not only ongoing but has evolved from a traditional PC/smartphone cycle into a structurally supply-constrained super-cycle led by AI data centers. Micron's results validate the real strength of demand, while SK Hynix's listing validates the fervor of capital, underscoring that investors are still willing to pay a high premium for the "memory chip supply bottleneck."
The grand investment narrative of global capital seeking "silicon-based inflation" over "carbon-based" assets this year essentially involves capital shifting from traditional manufacturing, automotive, consumer goods, real estate, energy, and other "carbon-based assets" reliant on population, resources, and linear economic growth, towards the high-end manufacturing supply chain centered on silicon wafers for AI computing infrastructure.
The core logic for global capital embracing "silicon-based inflation" undoubtedly lies in the fact that in the AI era, the scarcest resources are not traditional labor, real estate, or general production and manufacturing capacity, but rather "silicon-based means of production" such as GPUs/ASICs, HBM/DRAM/NAND memory chips, data center CPU components, high-performance Ethernet infrastructure, advanced packaging capacity, cutting-edge semiconductor manufacturing equipment like EUV, data center power chains, and data center optical interconnect/communication.
From Cyclical Commodity to Structural Necessity
The core significance of SK Hynix's U.S. listing may extend beyond fundraising. It represents a shift in the valuation framework for the HBM leader from a discounted Korean domestic cyclical stock to a U.S. AI computing infrastructure asset pricing system. As described, the listing plan and the strong demand for its U.S. ADS validate capital enthusiasm, highlighting that investors remain willing to pay a premium for the memory chip supply constraint.
As the core supplier of HBM for AI systems required by companies like NVIDIA and Alphabet Inc. (NASDAQ: GOOGL), if SK Hynix can achieve valuation multiples in the U.S. market comparable to peers like Micron, its Korean-listed shares could also be revalued upwards.
From an industrial engineering perspective, HBM is no longer a standard memory component but a key constraint determining the throughput, power efficiency, and inference concurrency capability of GPU/ASIC clusters. By using U.S. market financing to lock in advanced packaging and EUV capacity in Yongin and Cheongju, SK Hynix is essentially securing "memory bandwidth supply rights" in advance during an era of AI computing shortage.
Goldman Sachs's June DRAM survey reveals that HBM pricing power is being re-anchored by traditional DRAM tightness. DDR5 spot prices have risen 20% since May 1st, representing a 25% premium to May contract prices; DDR4 prices rose 11% over the same period, with a premium as high as 45%. This indicates the spot market is significantly leading the contract market, suggesting further upward pressure on future contract prices.
More critically, Goldman Sachs sharply raised its 2027 HBM price growth forecast for Samsung from +14% to +44% year-over-year, noting that given tight HBM supply and demand and a widening price gap between traditional DRAM and HBM, the forecast still carries upside risk.
South Korea's May DRAM exports increased 21% month-over-month and 370% year-over-year. Combined revenue from Taiwanese server ODMs grew 53% year-over-year, while revenue for the world's largest server BMC chip supplier, Aspeed, grew 69% year-over-year. This collectively indicates that this round of memory price increases is not merely inventory restocking but systemic tightness caused by the combined effects of AI GPU rack-scale servers, AI ASIC server racks, HBM packaging, and the reallocation of traditional DRAM supply.
In other words, the logic behind HBM price increases is not isolated; it is the result of the entire AI data center memory layer upgrading from a "capacity resource" to a "bandwidth resource, power resource, and delivery resource."
Furthermore, Wall Street firm Bernstein significantly raised its price target for SanDisk to $3,000 in a recent report, representing a significant shift in the valuation logic for the NAND memory chip segment. The market is no longer looking solely at short-term NAND spot prices but is beginning to reprice the cash flow visibility and downside protection offered by Long-Term Agreements (LTAs). Bernstein raised its price target for SanDisk from $1,700 to $3,000, maintaining an Outperform rating.
Bernstein noted that the new generation of LTAs differs from historical "supply guarantee agreements" that favored customers. They now incorporate fixed prices or price ranges, customer capital commitments or financial guarantees, and 3 to 5-year contract terms, giving memory manufacturers stronger profit buffers during price downturns for the first time.
According to Bernstein's latest stress test, if 60% of SanDisk's shipments are covered by LTAs, even if NAND prices crash 72% from their peak to $0.11/GB, EPS could still reach $214 by FY2030, compared to only about $81 without LTA protection. Under an extreme pressure scenario of $0.06/GB ASP by FY30, 60% to 80% LTA coverage could still support a blended ASP between $0.19 and $0.24/GB, with gross margins remaining above 80%.
These latest market dynamics suggest that AI data centers are simultaneously revaluing the "scarcity pricing power" of DRAM/HBM and the "contractual cash flow attributes" of NAND. While cycles won't disappear, the long-term low valuation, high volatility, and pure commodity labels of memory chip manufacturers are being systematically and significantly weakened by the AI data center construction wave's seemingly endless demand for AI computing resources and long-term price-lock contracts.
Unfinished Rally for Memory Leaders?
Nomura is currently the most bullish large financial institution on the memory chip sector. In a recent research report, Nomura raised its price target for Samsung Electronics from 340,000 won to 590,000 won, and for SK Hynix from 2.34 million won to 4.00 million won, representing potential upside of approximately 118% and 120%, respectively.
Nomura's core bullish logic is that AI has transformed memory from a traditional PC/phone cyclical product into a long-term growth asset for data centers. AI agent inference requires massive Key-Value (KV) Cache, and HBM supply is significantly lagging demand. Nomura expects global data center total capital expenditure to increase from $1.16 trillion last year to $6.13 trillion by 2030, with memory's share of data center investment potentially rising from the current 9% to 23%. Therefore, the approximately 6x 12-month forward P/E ratios for Samsung and SK Hynix are clearly undervalued, with room for revaluation towards the ~20x valuation system of Taiwan Semiconductor Manufacturing Company (NYSE: TSM).
Whether it's Google's massive TPU AI computing clusters or NVIDIA's immense AI GPU computing clusters, all require fully integrated HBM memory systems alongside the AI chips. This is compounded by tech giants accelerating the construction or expansion of AI data centers, necessitating large-scale purchases of server-grade DDR5 memory and enterprise-grade high-performance SSDs/HDDs. Samsung Electronics, SK Hynix, and Micron Technology are positioned across these three core memory areas: HBM, server-grade high-performance DRAM (including DDR5/LPDDR5X), and high-end data center-grade SSDs. They are the most direct beneficiaries within the "AI memory + storage stack," capturing the "super dividend" of the AI infrastructure wave.
GPUs are responsible for generating intelligence, HBM/DRAM is responsible for high-speed data feeding, enterprise NAND/eSSD handles hot data and caching, while HDDs manage the long-term retention of massive amounts of cold/warm data. Therefore, Goldman Sachs and other Wall Street firms believe the AI computing arms race led by cloud giants is transforming memory chips from cyclical products into scarce strategic assets. Price increases for DRAM/NAND in 2026 are not the end but could be the initial phase of a super-cycle.
Recent research from the well-known semiconductor industry analysis firm SemiAnalysis shows that although Chinese DRAM giant ChangXin Memory Technologies (CXMT) will maintain one of the world's fastest capacity expansion paces in the coming years, SemiAnalysis does not believe this will alter the overarching trend of tight DRAM supply.
Conversely, the firm believes that AI servers, high-performance computing, and cloud computing infrastructure construction are continuously driving up memory demand, while the pace of new capacity release still struggles to keep up with demand growth. According to SemiAnalysis's calculations, even incorporating the expansion plans of major players like CXMT, Samsung, SK Hynix, and Micron into its model, the global DRAM market will still face a high-single-digit percentage supply deficit in 2026. By 2027, the supply-demand gap could even widen further to a low-to-mid double-digit percentage level.
Based on this assessment, the research firm expects the tight DRAM supply situation to persist until around 2028. In other words, even the rapid rise and continued expansion of CXMT is unlikely to fundamentally reverse the industry's supply shortage, which forms a key part of SemiAnalysis's rationale for remaining bullish on DRAM prices and the overall HBM/DRAM/NAND memory industry upcycle.
Comments