According to AP, on June 23, U.S. tech stocks and the AI supply chain collectively declined, with the Nasdaq index closing down 2.2% and the S&P 500 down 1.4%. This pullback was not due to issues at a single chip company, but rather a simultaneous pressure on the most crowded AI hardware trades over the past year. The first pressure came from a sudden rise in expectations for Federal Reserve rate hikes, and the second from investors beginning to question when the cloud providers’ continued heavy AI capital expenditures would translate into clearly defined profits.
The most direct pressure fell on the hardware chain. Market data showed that NVIDIA (NVDA) dropped about 4% on Tuesday, with its market capitalization falling below $5 trillion. Micron plunged 13.2%, Qualcomm declined around 8%, and SanDisk and Western Digital also tumbled significantly. Memory, storage, AI chips, and smartphone chips all weakened together, indicating that the selling pressure was not confined to a single segment.
Asian markets also came under pressure. On June 23, South Korea’s KOSPI index fell nearly 10%, with SK Hynix and Samsung Electronics both posting double-digit losses. For months, tight supply of HBM and memory chips had supported South Korean tech stocks, but this time, the market chose to book profits first.
The order of this downturn is highly significant. Investors did not first retreat from software or internet platforms; instead, they initially targeted chip and memory stocks that had previously benefited the most from AI capital spending.
NVIDIA remains the core asset in the AI boom. Its GPUs have nearly defined this cycle of data center expansion and have become the primary outlet for market risk appetite. A market cap falling below $5 trillion does not alter the company’s industrial standing, but it serves as a prominent price signal on the trading front. When both interest rates and return cycles are questioned, assets with the largest gains and most crowded positions are often sold first.
Micron's decline was steeper, partly due to its upcoming earnings report. The company announced that it will release its third-quarter fiscal year 2026 results and hold an earnings call on June 24. The market had already priced in sustained strong demand for high-bandwidth memory driven by AI servers. If guidance is weak, investors fear that previous price gains lack new earnings catalysts; even if guidance is strong, the company must demonstrate that high memory prices and AI demand are not merely short-term stockpiling.
The market reaction in South Korea further amplified these concerns. SK Hynix and Samsung, both key players in the global memory and HBM supply chain, posted double-digit declines, indicating that this correction has spread from U.S. tech leaders to the global AI hardware supply chain.
Previously, Broadcom's AI revenue guidance falling short of the most optimistic expectations triggered a sell-off in chip stocks. Tuesday's market movement resembled a concentrated release of these concerns. Demand for AI remains strong, but the market is no longer willing to pay ever-higher prices solely for the promise of "huge potential in the future."
The trigger at the macro level comes from changes in expectations regarding Federal Reserve policy.
According to a Federal Reserve announcement, Kevin Warsh was sworn in as Chair of the Federal Reserve on May 22. Citing a Bank of America forecast, Reuters reported that the Fed may raise interest rates by 25 basis points in September, October, and December 2026, for a total annual increase of 75 basis points, citing resilient labor markets and persistent inflationary pressures.
This is particularly unfavorable for tech stocks. The valuation of AI leaders relies heavily on long-term growth expectations; rising interest rates increase the discounting pressure on future cash flows and make low-risk assets like U.S. Treasuries more attractive again. Recently, U.S. Treasury yields have remained elevated, and futures markets have clearly intensified bets on rate hikes this year, causing market expectations for policy path to adjust rapidly.
The market isn’t suddenly doubting the existence of AI; it’s recalculating a more realistic question: If the cost of capital is higher and future profits are further away, how much should we be willing to pay for AI assets today?
This is also why adjustments in chips, memory, and high-growth tech stocks have been so synchronized. They previously benefited together from the combination of “sustained explosive demand for AI” and “eventual decline in interest rates.” Once one of these pillars weakens, the segments with the largest gains and highest valuations come under pressure first.
Another source of pressure comes from AI capital expenditures themselves.
Major hyperscale cloud and AI investors such as Alphabet, Amazon, and Meta continue to maintain intense data center construction. Over the past year, such spending has been viewed by the market as a guarantee of demand for NVIDIA, memory chips, power equipment, and data center assets. As long as cloud providers continue to invest heavily, the hardware supply chain will sustain its revenue.
But now the question is, can this money eventually be recovered?
Training and inference for AI models require massive computational power, electricity, and server investments. Cloud providers can monetize through enterprise customers, advertising tools, developer platforms, and consumer subscriptions, but it remains unproven whether service pricing can fully cover capital expenditures. The market is beginning to scrutinize AI product pricing, customer usage intensity, and whether enterprises are willing to pay high long-term fees for generative AI.
This is also why the "sell the heavy spenders" trade has gained popularity. Investors are not only selling chip stocks but also becoming more cautious toward internet and cloud computing giants that continue to increase their AI budgets. The more aggressive their prior spending, the more they are questioned about profitability and free cash flow.
The volatility of overvalued assets is amplifying this sentiment. According to Axios, SpaceX's stock dropped more than 16% on Monday following its IPO, wiping out approximately $400 billion in market value. While not the primary driver of the recent chip stock decline, it illustrates that strong-narrative, high-valuation assets are facing stricter market scrutiny.
This correction is more accurately described as a concentrated pullback following significant gains in AI trading, rather than a confirmed bubble burst.
Demand for AI hardware remains strong, and cloud providers have not halted their capital expenditures. The fundamentals of companies like NVIDIA, Micron, and SK Hynix remain closely tied to data center construction, HBM supply, and AI server shipments. The real question is whether current stock prices have already priced in too many positive developments.
The first verification point is Micron’s earnings report. The market will focus on three things: whether demand for memory driven by AI servers remains strong, whether price increases can be sustained, and whether management’s guidance for upcoming quarters is sufficient to support prior gains. If the earnings report is strong, the chip supply chain may get a reprieve; if guidance falls short of expectations, selling pressure could continue to spread to more AI supply chain companies.
The second checkpoint is interest rates. Whether the Fed under Walsh truly raises rates starting in September will depend on inflation, employment, and energy prices. If inflationary pressures remain stubborn, growth stock valuations will continue to face pressure; if the data shows cooling, the market may reprice expectations of a policy pivot, providing room for tech stocks to recover.
The current market divergence lies in whether this is merely a normal profit-taking within the AI bull market, or the beginning of a shift by investors from “buying growth at all costs” to “demanding tangible returns.” Tuesday’s decline at least indicates that the AI narrative remains strong, but it can no longer alone offset the pressures of higher interest rates and a longer path to profitability.
律动 BlockBeats