Key Points
Semiconductor memory stocks led a sharp selloff in AI-related shares, with Micron and Marvell falling heavily after a crowded rally tied to AI data center demand.
Broadcom’s results cooled investor sentiment because the market had already priced in very high expectations for AI-related growth. Strong U.S. jobs data and rate fears added pressure to growth stocks.
The upcoming SpaceX IPO and potential listings by major AI companies are making investors reconsider where capital should sit within the broader AI infrastructure trade.
News
Semiconductor memory stocks fall as AI-related shares come under pressure
U.S. semiconductor stocks fell sharply on June 5, 2026, led by memory and AI-related names.
According to MarketWatch, the PHLX Semiconductor Index dropped 10.3%, marking its worst one-day performance since March 2020. Marvell fell 16.7%, Micron dropped 13.3%, AMD lost 10.9%, Broadcom declined 7.9%, and Nvidia fell 6.2%. The selloff hit many of the companies that had benefited from enthusiasm around AI data center investment.
Broadcom’s earnings became one of the key triggers for the decline. Its AI-related revenue continued to grow, but the results and outlook did not meet the level of optimism investors had built into the stock. That disappointment spread across the broader semiconductor sector, especially to names that had risen sharply in recent months.
Rate concerns added another layer of pressure. Strong U.S. jobs data reduced expectations for near-term Federal Reserve easing and raised concerns that interest rates could stay higher for longer. Kevin Warsh took office as Federal Reserve chair on May 22, 2026, and markets are watching how the new Fed leadership will approach inflation and rate policy at the June FOMC meeting.
In Japan, the focus now turns to whether the U.S. semiconductor selloff will spill over into stocks such as Kioxia, Tokyo Electron, Advantest, Lasertec, and Disco. Kioxia is especially sensitive to sentiment around memory stocks because its core business is NAND flash memory and SSDs, which are tied to data storage and high-speed storage demand in AI data centers.
Background
Memory stocks face profit-taking after strong AI-driven gains
Semiconductor memory stocks had risen on expectations that AI data center demand would continue to expand.
The AI boom has increased demand not only for GPUs but also for memory, storage, networking chips, power equipment, and cooling systems. As AI models become larger and more widely used, data centers require more computing power and greater storage capacity.
Memory companies such as Micron are positioned to benefit from AI server demand. In Japan, Kioxia has also been viewed as a company linked to AI data center expansion through NAND flash memory and SSDs.
Stock prices had already reflected a significant amount of future growth. Even when AI-related revenue continues to rise, a stock can fall if investors had expected an even stronger upside surprise. The latest pullback reflects a combination of profit-taking and a reassessment of how much growth had already been priced into semiconductor memory names.
Broadcom’s earnings cooled expectations for AI semiconductors
Broadcom’s earnings played a major role in changing market sentiment toward AI chip stocks.
The company’s AI-related revenue continued to show strong growth. The problem for investors was that expectations had already become extremely high. The market wanted not only growth, but growth that could exceed already ambitious forecasts.
AI-related stocks are now being judged by a stricter standard. It is no longer enough for companies to show that AI demand remains strong. Investors are also asking whether growth, margins, and guidance can justify valuations that have already moved far ahead.
Memory stocks are particularly sensitive to this shift. AI demand remains a tailwind, but the memory industry is also cyclical. Supply expansion, pricing pressure, and inventory cycles can quickly change investor sentiment. When stock prices rise in anticipation of future demand, even a small disappointment can trigger a large move.
Rate fears and the new Fed leadership add pressure
Interest-rate concerns also contributed to the selloff in growth-oriented stocks.
Strong U.S. employment data reduced expectations that the Federal Reserve would quickly move toward rate cuts. When rates stay high, the present value of future earnings falls. That matters especially for AI-related companies whose valuations depend heavily on long-term growth assumptions.
The arrival of Kevin Warsh as Federal Reserve chair has added another layer of uncertainty. Investors are watching whether the new Fed leadership will keep a firm focus on inflation, how cautious it will be about rate cuts, and whether markets need to price in a higher-for-longer policy environment.
For semiconductor memory stocks, the main pressure came from AI expectation resets and profit-taking after a strong rally. Rate fears made the selling broader and easier to justify, especially in stocks that had already moved sharply higher.
Kioxia is tied to AI infrastructure through storage demand
Kioxia is now one of the key Japanese stocks to watch.
The company’s core business is NAND flash memory and SSDs. In AI data centers, Kioxia is closer to data storage and high-speed read-write demand than to the main computing layer handled by GPUs.
AI workloads require not only processing power but also large-scale storage. Training data, inference data, logs, images, video, and enterprise datasets all need to be stored and accessed quickly. As AI usage expands, storage becomes an important part of the infrastructure stack.
Kioxia is therefore linked to AI data center demand through memory and storage. The U.S. selloff in Micron and other memory names could make Japanese investors more cautious toward Kioxia in the short term. The key question is not whether AI infrastructure demand has disappeared, but how much future growth had already been priced into the stock.
Analysis
The AI trade is moving from broad enthusiasm to capital selection
The semiconductor memory selloff does not mean that AI demand has collapsed.
AI data center investment remains large. Demand for computing chips, memory, storage, networking equipment, power systems, and cooling infrastructure remains strong. As generative AI adoption expands, data centers need more computing power and larger storage capacity.
The market issue is valuation. Nvidia, Broadcom, Micron, Marvell, and other AI-related names had already priced in a significant amount of future growth. When a stock has moved far ahead, even solid earnings may not be enough if the outlook falls short of elevated expectations.
The latest correction points to a shift in the AI trade. Investors are moving from a phase in which almost anything connected to AI could rise, toward a more selective phase focused on growth quality, margins, valuation, and the ability to beat expectations.
Memory stocks benefited from the AI infrastructure story, but that also made them vulnerable to profit-taking. AI demand may remain strong, while individual stocks still face sharp corrections when expectations become too stretched.
The SpaceX IPO creates a major capital demand
The SpaceX IPO is not the direct cause of the semiconductor selloff. Still, it is becoming an important factor in how investors think about capital allocation.
Reuters has reported that SpaceX is expected to price its IPO on June 11, 2026, and begin trading on Nasdaq the following day. The offering has been reported at $135 per share, with a potential $75 billion raise and a valuation around $1.75 trillion.
When a very large IPO approaches, investors often look for capital inside existing portfolios. Stocks that have already risen sharply can become natural sources of cash. In this case, AI-related semiconductor stocks and memory names are among the assets that had already delivered large gains.
SpaceX is also being viewed as more than a traditional space company. Its satellite network, launch infrastructure, communications platform, and potential links to data flow and AI infrastructure make it part of a broader investment theme involving space, communications, cloud infrastructure, and AI.
That makes the IPO relevant to the AI trade. Investors are not simply deciding whether to own semiconductor stocks. They are also deciding how much capital should remain in already-listed AI beneficiaries and how much should be reserved for the next large infrastructure story.
Capital may be moving from listed AI stocks to upcoming mega listings
AI investing is no longer limited to semiconductor stocks.
Public companies such as Nvidia, Broadcom, Micron, and Marvell have already attracted major investor capital. Their stock prices have benefited from expectations that AI data centers will continue to require more computing power, memory, networking, and storage.
The next wave of capital demand is broader. Companies such as OpenAI and Anthropic represent the model layer of the AI economy. SpaceX represents space, communications, and infrastructure. Cloud and data center companies are also part of the same capital-intensive ecosystem.
For investors, this creates a new choice. They can continue holding listed AI stocks that have already rerated, or they can prepare for future listings and new large-scale opportunities. Taking profits in stocks that have already risen sharply can be a way to prepare for the next round of capital allocation.
This is not necessarily money leaving AI. It may be money rotating within the wider AI infrastructure complex, from listed semiconductor winners toward upcoming model, infrastructure, and space-linked companies.
The semiconductor memory pullback can be read as one early sign of that transition. AI demand remains, but investors are beginning to compare already-valued stocks with the next set of companies that may absorb market capital.
Japan’s memory stocks face a test after the U.S. selloff
Japan’s semiconductor sector now faces a short-term test.
The sharp decline in Micron and other U.S. memory stocks could weigh on Japanese memory-related shares. Kioxia is especially exposed to sentiment around memory and storage because its business is tied to NAND flash memory and SSDs.
Kioxia is linked to AI data centers through storage demand. AI infrastructure requires fast and scalable data storage, not only computing chips. This gives Kioxia a place in the AI infrastructure story, even though it is not primarily a GPU company.
In the near term, profit-taking may spread from U.S. memory stocks to Japanese names. Kioxia, Tokyo Electron, Advantest, Lasertec, and Disco could all be affected by broader semiconductor-sector selling.
The medium-term question is whether AI data center demand continues strongly enough to support valuations. Memory stocks benefit from structural demand growth, but they remain exposed to supply cycles and pricing pressure. After a sharp rally, investors are likely to focus more closely on whether expected growth can justify current valuations.
For the Japanese market, U.S. Treasury yields, Nasdaq futures, the SOX index, and currency moves will also matter. The coming sessions will show how much of the U.S. semiconductor correction has already been priced in, and where buyers begin to return to memory-related stocks.
Conclusion
The AI trade is not ending. It is becoming more selective.
The latest semiconductor memory selloff does not signal the end of AI demand.
AI data center investment remains strong, and demand for computing chips, memory, storage, networking equipment, and related infrastructure continues to expand. The market is not rejecting AI. It is becoming more selective about which AI-related assets deserve high valuations.
Broadcom’s results reminded investors that AI-related companies now face a very high bar. Growth alone may not be enough. Stocks must also show that they can beat elevated expectations, sustain margins, and justify valuations that already reflect a large amount of future demand.
Micron and other memory stocks were hit because they had benefited from the AI infrastructure story. In Japan, Kioxia and other semiconductor-related names may face similar pressure as investors reassess how much future growth has already been priced in.
Rate concerns add to the pressure. Strong U.S. jobs data and uncertainty around the new Federal Reserve leadership make high-growth stocks more vulnerable to valuation resets.
SpaceX and other potential AI-related mega listings add another layer. Investors are beginning to decide whether to keep capital in already-listed AI winners or reserve funds for the next large infrastructure story. The movement is not capital leaving AI. It is capital being reallocated within a much broader AI, space, communications, data center, and semiconductor ecosystem.
This article is for informational and analytical purposes only. It does not constitute investment advice or a recommendation to buy or sell any specific stock or financial product. Investors should make their own decisions based on their risk tolerance and independent research.
See you again in the next article.
Reference Links
- Marvell, Micron Shares Tumble as the Chip Sector Suffers Its Worst Day in 6 Years (MarketWatch)
- Broadcom’s Sales and AI Chip Forecast Comes in Below Expectations, Shares Tumble (Reuters)
- Broadcom Set to Shed $300 Billion in Value as AI Results Fail to Impress (Reuters)
- Wall Street Selloff Deepens as Strong Jobs Data Fuels Rate Worries (Reuters)
- SpaceX Accelerates IPO Timeline, Targets June 12 Listing on Nasdaq (Reuters)
- SpaceX Plans to Raise $75 Billion in IPO at $135 Per Share (Reuters)
- Anthropic Confidentially Files for Initial Public Offering on US Stock Market (The Guardian)
- OpenAI Plans ChatGPT ‘Superapp’ Overhaul Ahead of Listing, FT Reports (Reuters)
- Kioxia Corporation Home (KIOXIA)
