Thomas of Voya stated that Kioxia's potential listing in the United States could position it as a focal point in the global artificial intelligence sector. Kioxia, whose share price has surged 456% year-to-date, is leading the Nikkei 225 Index. The company announced in May that it is preparing for a U.S. initial public offering to broaden its investor base.
Zhitong Finance APP has learned that, as analysts widely speculate that Japan's leading NAND memory chip manufacturer $Kioxia Holdings (285A.JP)$ is poised to knock on Wall Street's door and list on the U.S. stock market for public trading, similar to international semiconductor giants such as Taiwan Semiconductor and SK Hynix, the memory chip sector, which has recently returned to a bull market trajectory, is seeing dual catalysts from both fundamental demand and capital flows. The core investment implication of the market's speculation regarding Kioxia's plan to list in the U.S. lies in significantly expanding the tradable market accessible to institutional investors, thereby providing additional funding sources for the company, which is already driven by AI demand.
Kioxia's share price has surged 456% in the Japanese stock market year-to-date, outperforming the Nikkei 225 Index. However, Sebastian Thomas, an investment manager at Voya Investment Management—which oversees $14 billion in AI-focused funds—has not yet taken a position in the stock, primarily due to liquidity concerns.
Thomas stated that for large funds, beyond corporate fundamentals, factors such as actual position-building capacity, transaction impact costs, ease of exit, and liquidity strength also influence allocation decisions. He noted that following SK Hynix's recent listing in the U.S. market via American Depositary Receipts (ADRs) could improve these conditions for Kioxia, enhancing its investability within global institutional AI portfolios.
Memory Chips Break Out of Mid-Summer Consolidation; Goldman Sachs Detects Signs of a New Memory Bull Market
Wall Street financial giant Goldman Sachs has renewed its bullish outlook on the trading environment for the memory chip sector. This view is primarily based on the expansion of demand for cutting-edge high-performance AI computing power driven by the launch of OpenAI's Astra, and the rising expectations for memory chip demand as the RSI (Recursive Self-Improvement) training paradigm begins to dominate AI training. This coincides with low positioning among traditional Wall Street asset managers and hedge funds. In particular, stocks such as Micron Technology and SanDisk have begun to break through their summer downtrend technical lines, while implied volatility in the semiconductor sector has declined, leaving room for hedge funds to rebuild positions after previously reducing their risk exposure.
Regarding the fundamentals of the memory chip sector, Timothy Arcuri, a senior analyst at UBS Group, expects that the quarter-on-quarter increase in average selling prices for memory chips in the third quarter will exceed 20%, against the record-high base established in the second quarter. The supply shortage for both DRAM and NAND is expected to persist until 2027. The combination of these two bullish investment perspectives forms a logic where 'earnings expectations are supported, and there is room for capital replenishment given low positioning.' Whether sustained revaluation can occur still depends on the realization of actual selling prices, shipments, and profits.
The strong U.S. Producer Price Index (PPI) data released on Thursday led to market sell-offs, but prior market performance had already reflected the recovery of this memory theme. The South Korean KOSPI Index, known as the 'bellwether for AI computing power,' rebounded approximately 21% from its late-July lows by mid-August, reaching the threshold commonly referred to as a technical bull market. It rose another 4.6% on September 7, with Samsung Electronics and SK Hynix gaining 5.7% and 8.1%, respectively. In the U.S. market, as of the close on September 9, the Philadelphia Semiconductor Index had risen for five consecutive trading days, with a cumulative gain of nearly 6%. During the same period, the Roundhill Memory ETF rose by nearly 12%, indicating that the memory sector led this rebound.
A significant industrial signal brought by Astra is that more complex tasks are beginning to hold commercial value when executed by AI. On September 10, media reports indicated that OpenAI launched a ChatGPT product tailored for the financial services industry, combining GPT-6 Astra with professional data sources to support research, financial modeling, and client material creation. By extension, the growth variables of AI demand will further expand to include the number of concurrent agents, task execution duration, frequency of tool calls, and context size. As the cost of completing a task decreases and success rates improve, enterprises have greater justification for deploying more workflows. This opens up vast new space for cloud-based AI inference computing power and high-performance storage demand related to AI, serving as the latest basis for the market to reassess the sustainability of AI infrastructure growth.
From Tokyo's Stock King to Wall Street's Focus? A U.S. Listing Could Make Japan's Kioxia the Center of Global AI Investment
From an underlying architecture perspective, Kioxia's direct benefits mainly stem from NAND flash memory and enterprise-grade SSDs. Agents calling upon enterprise knowledge bases, retrieving vector data, loading models, saving task states, and maintaining audit logs all require persistent storage. Long contexts and multi-turn tasks will also increase the capacity demand for reusable key-value caches, driving some caching to be stored in SSDs through tiered architectures.
HBM handles high-bandwidth computing access, server DRAM manages low-latency working sets, and NAND provides cost-effective large-capacity storage, with the three technologies working in concert. Kioxia has positioned SSDs with high random read/write performance as a key product direction for enhancing AI inference efficiency. Relevant studies also indicate that optimizing SSD cache scheduling can reduce redundant computations and GPU wait times. Therefore, Kioxia's long-term growth logic should rest on increasing enterprise SSD capacity, performance value, and customer share. While listing in the U.S. improves conditions for capital market participation, sustained profit growth still requires support from product competitiveness, supply and demand dynamics, and order volumes.
The chief portfolio manager of a $14 billion fund focused on artificial intelligence investments stated that Kioxia's plan to list in the United States could return this Japanese chipmaker to the global investment spotlight. Kioxia's shares have surged 456% year-to-date, leading gains among constituents of the benchmark Nikkei 225 Index. According to informed sources, the company indicated during an internal meeting in May that it was preparing for the listing and trading of American Depositary Shares (ADS) in the U.S. stock market to expand its investor base.
Voya Investment Company's Global Artificial Intelligence Fund does not hold Kioxia, nor does it hold any technology stocks driving the rise of the Nikkei Index. The fund includes $5 billion in capital from Japan.
Sebastian Thomas, Portfolio Manager at Voya, stated that the issue lies in liquidity, and trading in the U.S. market could alter this consideration, similar to the surge in AI financing and investment seen after South Korean memory chip giant SK Hynix listed on Nasdaq in July.
"There are many noteworthy companies in Japan, particularly those in the AI computing infrastructure supply chain," Thomas said in an interview with media on Thursday. "The challenge is finding companies with sufficient liquidity to enable our investment."
It is understood that this AI investment fund, affiliated with Sumitomo Mitsui DS Asset Management, has achieved a cumulative return of approximately 600% since its inception ten years ago this month, calculated on a pre-tax distribution reinvestment basis.
The fund focuses on investing in core AI computing infrastructure companies within the computing power industry chain, such as AI chip leader NVIDIA (NVDA.US), AMD, and AI ASIC pioneer Broadcom, with NVIDIA being its largest holding. It also invests in cutting-edge software application developers related to AI applications, as well as manufacturing companies likely to benefit from adopting advanced AI technologies, such as Eli Lilly and Co (LLY.US).
Thomas noted that Voya has previously invested in Japanese companies. Although it does not currently hold Kioxia, it has exposure to other memory chip manufacturers, including SK Hynix and Micron Technology. He added that a U.S. listing makes such investments easier to execute. "We generally prefer technology companies with better liquidity, larger position capacity, and greater market capitalization," he said.
Edited by Deng