Kioxia and SK Hynix Eye U.S. Listings as AI Semiconductor Supercycle Fuels Capital-Raising Race

Kioxia valuation surges on soaring NAND prices
Competition intensifies to tap U.S. market premium for semiconductor financing
Sustainability of valuations remains uncertain amid mounting AI bubble concerns

Japanese memory semiconductor company Kioxia Holdings is pursuing a U.S. stock market listing. The move is widely interpreted as an effort to improve access to global investors and attract additional capital inflows amid the semiconductor boom driven by expanding artificial intelligence (AI) investment. SK Hynix is also reviewing a U.S. listing, signaling the emergence of a full-scale race across the global semiconductor industry to secure massive financing through American capital markets. At the same time, growing concerns in the U.S. market over an AI bubble and slowing memory demand are raising the possibility that shifts in investor sentiment could become a critical variable affecting semiconductor firms’ listing strategies.

Kioxia’s U.S. Listing Drive After 300% Rally in One Year

According to Bloomberg on the 20th local time, Kioxia is preparing an American Depositary Share (ADS) listing on the New York Stock Exchange (NYSE). ADS instruments allow shares of foreign companies to trade in U.S. markets. Although Kioxia already has depositary receipts in place, an ADS structure lowers arbitrage costs and enhances accessibility for American investors.

Kioxia has emerged as one of the biggest beneficiaries of the global memory semiconductor rally, with its stock price surging more than 300% this year alone. Its market capitalization has climbed to approximately $177 billion, making it the fourth-largest company on the Japanese stock market behind Toyota Motor, Mitsubishi UFJ Financial Group, and SoftBank Group. Kioxia reported first-quarter operating profit of approximately $4.1 billion, up 314% from the previous quarter. Total operating profit for fiscal year 2025 reached roughly $6 billion, marking a 90% increase from a year earlier.

In particular, Kioxia projected second-quarter operating profit of around $8.9 billion. The guidance reflects confidence that soaring NAND prices could allow the company to surpass its entire previous fiscal year’s operating profit in just a single quarter. Despite the explosive share-price rally over the past year, Kioxia’s forward 12-month price-to-earnings ratio (PER) remains around 10 times earnings, roughly half the Nikkei average of 21–23 times. If the company were to receive a market-average valuation multiple, it could potentially challenge Toyota for the top spot in market capitalization.

Spun off from Toshiba, Kioxia primarily manufactures NAND flash memory used in laptops and AI data centers. As global memory manufacturers increasingly focus resources on high-bandwidth memory (HBM) for AI servers, market observers believe Kioxia has been reaping substantial spillover benefits in the NAND market. Bloomberg noted that Kioxia is already among the most actively traded stocks in Japan and that a U.S. listing could further accelerate capital inflows.

SK Hynix Targets U.S. ADR Listing This Year

Kioxia’s U.S. listing initiative reflects a broader strategic push to preemptively absorb liquidity from U.S. capital markets at a time when AI memory supercycle dynamics are rapidly inflating semiconductor valuations. The U.S. equity market has become the epicenter of global AI investment flows, and semiconductor firms are increasingly embracing a cycle of “earnings growth → valuation re-rating → U.S. capital raising.”

SK Hynix is likewise accelerating procedures for a U.S. market listing, targeting a launch window between June and July this year. The company has reportedly begun assembling an underwriting syndicate centered on major U.S. investment banks. Securing large-scale funding is viewed as essential for maintaining leadership in the AI semiconductor market, where capital expenditure requirements are immense. Although SK Hynix held nearly $25 billion in cash reserves as of the end of last year thanks to the AI semiconductor boom, the scale of future investment demands is expected to be even larger. Facility investments required to meet next-generation semiconductor demand are reaching astronomical levels.

In February, SK Hynix decided to invest approximately $15 billion by the end of 2030 for the first fabrication plant within the Yongin semiconductor cluster. Including previously announced capital expenditures from July 2024, the total investment rises to roughly $21 billion. The company ultimately plans to invest around $420 billion by 2050 to operate four fabs in Yongin. In parallel, SK Hynix is pursuing the establishment of an AI solutions company in the United States to strengthen its position as a core partner for American AI data centers, with up to $10 billion in planned investment for the venture.

As SK Hynix’s U.S. listing becomes increasingly tangible, market attention is focusing on the scale of potential new share issuance. Earlier market expectations centered on an ADR listing backed by treasury shares. However, after the company effectively retired approximately $8.5 billion worth of treasury stock in January, leaving only employee compensation shares outstanding, new share issuance has become the primary remaining financing route. Consequently, large-scale equity issuance appears unavoidable if SK Hynix aims to secure the valuation re-rating it seeks. Investment banking industry estimates place the potential issuance size at around $10 billion. Achieving the company’s targeted valuation uplift would likely require broad participation from institutional investors.

AI Rally Drives U.S. Semiconductor Market as Bubble Warnings Intensify

The trend toward valuation re-rating through U.S. listings has become increasingly entrenched across the global semiconductor industry. Taiwan Semiconductor Manufacturing Co. (TSMC) and Dutch EUV lithography monopoly ASML are already listed in the United States through ADR structures, while British semiconductor design company Arm absorbed massive AI-related investment demand through a direct Nasdaq listing. Following its listing, Arm’s valuation surged sharply amid enthusiasm over generative AI server investment, reinforcing the perception that U.S. markets apply the most aggressive premium to AI semiconductor stocks.

The AI semiconductor rally led by Nvidia continues to propel U.S. markets higher, with gains in the Philadelphia Semiconductor Index (SOX) substantially outpacing those in Korean and Japanese markets. Global capital flows are likewise concentrating in New York. U.S. pension funds, technology-focused exchange-traded funds (ETFs), and long-term growth capital are all increasing allocations to AI semiconductor companies, intensifying the perceived necessity among chipmakers to secure a U.S. market premium.

However, the growing AI bubble narrative inside the U.S. market remains a major risk factor. Much of the concern stems from the share-price trajectory of Micron Technology. Just three years ago, Micron was posting massive losses, yet the company has since become one of the most profitable firms in the U.S. market thanks to explosive demand for HBM used in AI servers. Some projections even suggest Micron’s expected net profit over the next 12 months could surpass that of Meta and Berkshire Hathaway.

What has particularly alarmed analysts is the pace of growth itself. The memory semiconductor industry has long been regarded as one of the most cyclical sectors in the global economy. When demand surges, companies engage in aggressive capacity expansion, only for oversupply to later trigger price collapses and earnings deterioration. Micron itself was once viewed as undervalued in 2022 due to its low PER, before a sharp industry downturn subsequently drove the stock sharply lower.

Technological shifts in AI are also emerging as a key variable. In March, Google researchers unveiled “TurboQuant,” a technology capable of reducing AI memory usage to roughly one-sixth of existing levels. The announcement triggered a sharp selloff in global memory semiconductor stocks listed on U.S. exchanges. Concerns quickly surfaced that the technology could significantly reduce memory requirements while maintaining AI performance, potentially undermining current HBM demand forecasts altogether. In addition, market participants are increasingly discussing the possibility that major technology companies may slow the pace of data center investment. Although leading U.S. tech firms are currently building AI data centers at unprecedented scale, analysts caution that investment reductions cannot be ruled out if AI service monetization takes longer than expected.