According to reports, the US semiconductor sector is undergoing a profound market reassessment. As September inflation data came in hotter than expected, further reinforcing expectations of Federal Reserve rate hikes, and amid rare warnings from the CEOs of Broadcom (AVGO.US) and Intel (INTC.US) regarding AI infrastructure financing bottlenecks, the global AI-themed trade centered on semiconductors suffered massive selling over the past week. The Philadelphia Semiconductor Index (SOX) fell approximately 12% over the past six trading sessions, wiping out about $1.6 trillion in market value and marking its longest losing streak since May 2025; on Wednesday alone, the index plunged 4.30%. This selloff not only affected the US market but quickly spread to major global semiconductor hubs such as South Korea and Japan, sparking widespread discussion among investors about the sustainability of the AI investment cycle. More than half of the index's constituent stocks fell at least 10% during this period. Arm Holdings Plc (ARM.US) plummeted 21%, Applied Materials (AMAT.US) dropped 16%, and Intel (INTC.US) declined 13%. GlobalFoundries (GFS.US) was the only stock that gained, outperforming its peers during this period.
The semiconductor sector's decline echoed broader market weakness 鈥?the tech-heavy Nasdaq 100 fell more than 2%, its worst single-day performance since late July, while the S&P 500 dropped 1.1%. This wave of selling was not confined to the US market alone. Global semiconductor stocks experienced synchronized selling; in early Asian trading on Thursday, South Korea's KOSPI opened down more than 5%, hitting a new low since July, while Japan's Nikkei 225 plunged more than 3%. The synchronized decline in global semiconductor stocks reflects deepening systemic concerns among investors about the sector. "Pressure on global chip stocks has been mounting in recent days," wrote Ulrike Hoffmann-Burchardi of UBS's Chief Investment Office in a research note on Wednesday. "While intensifying competition in AI, rising financing needs, and higher interest rates may periodically amplify volatility, we believe this latest selloff is more of a recalibration of expectations than a fundamental break in the AI investment thesis."
Inflation Pressures Intensify Across the Board, Core CPI Rebounds Above Expectations
The latest round of pressure emerged after September inflation data came in higher than expected, further reinforcing market expectations that the Federal Reserve will continue raising rates, even though policymakers may hold steady at their next meeting later this month. Data released by the US Labor Department on October 14 showed that driven by rising energy and food costs, the September Consumer Price Index (CPI) rose 0.6% month-over-month, the largest increase since April; the year-over-year rate climbed to 3.7%, up from 3.4% in August and exceeding economists' expectations of 3.6%. Excluding volatile food and energy prices, core CPI rose 0.3% month-over-month, unchanged from August but above the market expectation of 0.2%; the year-over-year rate rose from 2.4% in August to 2.6%, exceeding the expected 2.5%.
Specifically, energy prices were the primary driver pushing overall inflation higher. The September energy price index rose 5.2% month-over-month, with the gasoline price index surging 9.1% month-over-month, contributing more than half of the month's CPI increase. Ongoing Middle East conflicts continued to push fuel costs higher; energy prices have risen 20.4% year-to-date through September, and rising fuel costs further pushed up airfare prices, with airline ticket prices rising 1.2% month-over-month in September and 22.4% year-to-date.
Notably, the scope of price increases has extended beyond the energy sector, showing a broader spillover pattern. Analysts pointed out that prices for entertainment, used cars and trucks, education, airfare, healthcare, and apparel all rose, with some increases stemming from tariffs and energy factors, while others suggest inflation is more persistent. Olu Sonola, head of US economic research at Fitch Ratings, said the latest inflation data "flashes a red light" for the Federal Reserve, and the transmission of price increases to broader services sectors could raise the risk of sustained inflation.
The deeper issue is that AI infrastructure building itself is becoming a driver of inflation. The data center construction boom is pushing up prices for semiconductors and other goods and transmitting to the broader economy through channels such as construction labor costs. Chicago Fed President Austan Goolsbee explicitly stated that inflationary pressures spawned by the AI boom are extending beyond directly related industries. "The price increases are no longer confined to data center components and directly related industries. I am increasingly concerned that this is classic overheating of aggregate demand."
The hotter-than-expected inflation data quickly reshaped market expectations for the Fed's rate path. According to the CME FedWatch tool, after the CPI release, traders' bets on a 25-basis-point rate hike in October briefly rose from about 18% before the data to 50%. The probability of a December rate hike rose to nearly 90%. Although market expectations have since cooled somewhat, the probability of a December hike remains above 60%.
Fed policymakers密集 released hawkish signals around the inflation data release. New York Fed President John Williams made clear he still believes one more rate hike this year is appropriate, though he does not think the central bank needs to raise rates further in 2027. Williams revealed that the Iran conflict pushing up energy prices and widening refined product spreads changed his assessment of the near-term inflation outlook, making him "strongly support" the September rate hike decision. Boston Fed President Susan Collins also said she expects "modest further rate increases will be appropriate" by year-end to ensure inflation returns to the Fed's 2% target in a timely and sustained manner. The Cleveland Fed president expressed a similar stance, suggesting further monetary tightening may be needed.
Citigroup adjusted its forecast after the CPI data, predicting the Fed will hike rates by 25 basis points in December 2026, whereas it had previously forecast the Fed would hold rates steady through the end of this year. In a special interview, Williams noted that the AI infrastructure boom, alongside tariffs and energy prices, has become one of the three core variables affecting the US inflation path. He observed that rising costs and commodity prices related to AI construction are already evident: "We do see rising goods and other costs related to AI construction: data centers, enormous demand for chips. Prices in AI-driven subsectors have seen rather sudden and significant increases, and we are also seeing some signs of spillover into construction labor costs." But he also offered a relatively optimistic long-term assessment 鈥?as chip capacity increases and productivity improves, future supply-side conditions may improve and create disinflationary forces.
Broadcom CEO's Warning Triggers Chain Reaction: Capital Constraints on AI Investment Surface
Beyond macro-level rate pressures, financing constraint warnings at the industry level became another important catalyst for this selloff. Broadcom CEO Hock Tan earlier this week likened some investors chasing the AI boom to "gamblers," saying bluntly that many AI investors do not understand the technology they are investing in yet hold extremely high expectations. He further warned that capital 鈥?not chip supply 鈥?will become the biggest constraint on AI development. Hock Tan also questioned the actual returns of corporate AI spending, noting that AI model developers must find customers who can generate returns on investment through adopting AI technology, rather than relying on users who are still in trial phases and constantly "wasting money" on tokens. He used a vivid analogy to illustrate the problem: "It's like when you were a kid going to the arcade, your parents gave you $10 worth of tokens, you spent them all in 10 minutes and learned nothing. That is exactly what is happening in the corporate world today." Hock Tan also questioned the actual returns of corporate AI spending, noting that AI model developers must find customers who can generate returns on investment through adopting AI technology, rather than relying on users who are still in trial phases and constantly "wasting money" on tokens. He used a vivid analogy to illustrate the problem: "It's like when you were a kid going to the arcade, your parents gave you $10 worth of tokens, you spent them all in 10 minutes and learned nothing. That is exactly what is happening in the corporate world today."
These remarks sparked widespread resonance and debate in the market. Michael Burry, the renowned investor and real-life prototype of the film "The Big Short," reposted Hock Tan's comments on social platform X, specifically highlighting the issue that companies spend money using AI tokens without necessarily creating real value.
Broadcom's warning was corroborated by reality. AI cloud computing service provider CoreWeave (CRWV.US) saw its unsecured bonds maturing in 2032 fall below 80% of face value on Wednesday, with yields exceeding 15%; its credit default swaps rose to record highs, with market pricing indicating the company's probability of default within the next five years exceeds 50%. CoreWeave shares closed down more than 15% on Wednesday. D.A. Davidson analysts had previously called CoreWeave the "subprime tranche" of AI, warning that if the AI financing boom fades, the company may be unable to repay its high debt. Daniel O'Regan, executive director of equity trading at Mizuho Securities, noted that in recent weeks investors have become "increasingly unwilling to fund AI-related spending at any cost."
AI capital expenditure has exceeded companies' ability to generate free cash flow, meaning companies must seek external financing to purchase needed hardware, and when external financing becomes more difficult, spending on chips and other AI infrastructure could be materially affected. A JPMorgan analyst team led by Nikolaos Panigirtzoglou wrote in an October 14 report that while the current stock price action resembles the late-July pullback, the background environment for AI infrastructure stocks is now more unfavorable. Panigirtzoglou said the reason is that "although overall equity funds still see continuous inflows, retail investors have so far lacked 'buy-the-dip' behavior, buyback flows have seen a cliff-edge decline, short positioning is insufficient, and leveraged equity ETF assets under management are higher." "Despite the recent drawdown, cyclicals do not appear oversold from a positioning or flow perspective." "We recommend investors focus on where AI spending is translating into sustainable earnings growth, as well as broader corporate and consumer adoption," said UBS's Hoffmann-Burchardi.
Comments