Investors looking for winners from the artificial-intelligence spending boom have moved steadily beyond the processors that train and run AI models. Memory and storage companies are increasingly part of the same calculation.
That shift has transformed Kioxia's standing in the market.
The Japanese NAND flash-memory producer has seen its shares rise roughly 456% this year and is now preparing another step in its international expansion: a planned U.S. listing through American Depositary Shares.
Kioxia is targeting roughly April to June 2027 for the listing. If it proceeds, the move would give American investors a more direct way to trade the company's shares without displacing its existing Tokyo listing.
The timing puts Kioxia at the intersection of two powerful trends: rising demand for memory across AI infrastructure and investors' search for semiconductor exposure beyond the industry's most familiar names.
A US Listing Is Primarily About Access
Kioxia already has substantial exposure to the growth of artificial intelligence. A U.S. listing would not change that underlying business.
What it could change is who can invest in it, and how easily.
The company plans to use American Depositary Shares, or ADSs, which represent shares of a foreign company and trade in the U.S. market. Kioxia has said the proposed listing is intended to broaden its investor base and support corporate value.
That could be particularly useful with institutional investors.
A company listed only in Japan can face practical barriers when attracting some international portfolios, including differences in trading hours and market access. A U.S.-traded security gives those investors a more familiar route into the stock and can potentially improve liquidity.
For Kioxia, that matters after a rally that has already dramatically raised its profile at home.
A Major AI Investor Is Paying Attention
One indication of the opportunity comes from Voya Investment Management.
Sebastian Thomas, a portfolio manager involved with Voya's roughly $14 billion artificial-intelligence strategy, has pointed to the potential benefits of greater liquidity if Kioxia establishes a U.S. listing.
The firm's Global Artificial Intelligence fund has returned about 600% over the past decade and already owns stakes in memory manufacturers Micron and SK Hynix.
Kioxia is not currently among its holdings.
That distinction is important. Thomas's comments should not be interpreted as a commitment by Voya to buy the stock. They show instead why access matters to large international funds already investing across the AI semiconductor supply chain.
A U.S. listing could make Kioxia easier to evaluate alongside companies those investors already own.
AI's Storage Requirements Are Expanding the Investment Story
For much of the current AI cycle, processors have commanded the attention.
Nvidia's graphics processing units became the clearest symbol of the boom as technology companies raced to build computing clusters capable of training increasingly sophisticated models.
Those systems also require memory and storage.
Kioxia specialises in NAND flash memory, a technology used extensively in solid-state drives and data-storage systems. As data centres process larger volumes of information, the infrastructure surrounding AI processors has become more important.
Kioxia has increasingly framed its own strategy around what it describes as the “AI Inference Era.”
Inference is the process of using a trained AI model to produce responses or predictions. It becomes especially important as AI moves from model development into widespread commercial use, where potentially millions of requests must be processed continuously.
Those workloads create storage requirements alongside demand for computing power.
For Kioxia, enterprise solid-state drives and advanced flash technology provide a way to participate in that expansion without competing directly in the market for AI processors.
Memory Companies Are Benefiting From a Different Market
The broader memory industry has also moved into a much stronger phase.
AI-related infrastructure demand and tighter supply have supported higher prices across parts of the memory market, improving the financial position of major manufacturers.
Kioxia reported record earnings of ¥596.8 billion for the quarter ending in March.
The company and Sandisk have also announced plans to invest more than $31 billion in Japan through 2032 to develop advanced semiconductor technology and increase production capacity, subject to government support.
That is a long investment horizon.
Building semiconductor capacity requires substantial capital and takes time, meaning decisions being made now reflect expectations about demand years into the future rather than conditions in a single quarter.
AI is central to those expectations, but it is not the industry's only source of demand. Smartphones, PCs, enterprise systems and conventional cloud infrastructure continue to consume large quantities of flash memory.
The investment case for Kioxia therefore depends on a broader storage market even as AI becomes its most closely watched growth driver.
A 456% Rally Changes the Conversation
Kioxia's extraordinary stock-market performance creates opportunity as well as pressure.
A roughly 456% rise in 2026 means investors have already priced in a considerable improvement in the company's prospects.
The rally makes a U.S. listing more prominent, but it also raises the standard Kioxia will have to meet if it wants to attract new shareholders at current valuations.
Memory has historically been one of the semiconductor industry's most cyclical businesses.
Shortages push prices and profits higher. Manufacturers respond by investing in capacity. If production eventually grows faster than demand, prices can fall sharply and margins can contract.
AI could lengthen or strengthen the current demand cycle, but it does not make those economics disappear.
That leaves investors with a more complicated question than whether artificial intelligence needs additional storage. They must judge whether demand can grow fast enough to absorb the industry's next wave of production.
Competition for AI Capital Is Already Intense
Kioxia would not arrive in the United States as the only way to invest in memory.
Micron already provides U.S. investors with direct exposure to the sector, while SK Hynix has become one of the industry's most important AI-related companies because of its position in high-bandwidth memory.
Sandisk offers another familiar storage name.
Kioxia would therefore need to compete for capital rather than simply benefiting from scarcity.
Its strength lies in NAND flash and its position in the storage market, which gives investors a somewhat different route into the infrastructure supporting AI.
That could become more relevant as attention moves from training massive models toward deploying them at scale.
The semiconductor investment story is also broadening beyond chips themselves. Data-centre operators, networking companies, power-equipment manufacturers and cooling specialists are all competing for capital tied to the same infrastructure expansion.
Simply carrying an AI connection is no longer enough to distinguish a company.
The Proposed Listing Comes at a Useful Moment
Kioxia's timing nevertheless has logic.
The company is profitable, its shares have attracted intense investor interest in Japan, memory-market conditions have strengthened and AI has made storage infrastructure more strategically important.
A U.S. listing could put that combination in front of investors who currently have easier access to Kioxia's competitors.
The planned ADS structure would also allow the company to increase its international visibility without abandoning its domestic market.
Whether that translates into sustained investor demand will depend less on the listing itself than on Kioxia's ability to convert favourable memory conditions into durable earnings.
The company's large investment plans make that challenge more important. Expanding semiconductor capacity requires money long before the resulting products generate revenue, and the industry can change considerably during that period.
For investors, Kioxia's proposed U.S. listing would therefore arrive with both sides of the memory cycle on display: exceptional current demand and the possibility that aggressive investment eventually creates more supply.
That tension is what makes the company an increasingly interesting part of the AI trade.
Kioxia does not manufacture the processors at the centre of the AI boom. It supplies something those processors cannot operate effectively without: somewhere to keep the enormous quantities of data they consume and produce.
A U.S. listing could make that less visible part of the AI infrastructure story considerably easier for global investors to buy.
