AI Selloff Sweeps Chip Stocks and Data Center Debt; Pharma Defense, IEA Reserve Release and Rate Repricing Become Main Themes

Global technology and AI-related assets are undergoing a broad selloff, from chip stocks and data center bonds to emerging cloud computing company debt, as rising financing costs and doubts about AI monetization reinforce each other; meanwhile, capital is shifting to defensive assets such as large-cap pharma, the International Energy Agency is accelerating oil reserve releases to address diesel tightness, and Apple and Meta are showing another industry path through on-device AI cooperation.^{[3][5][1][2][4]}

AI and chip stocks hit by global selloff

Thursday's global tech selloff continued. In U.S. stocks, Nvidia closed down 1.84% that day, AMD fell 1.5%, Intel dropped about 2.8%, and Dell Technologies plunged 6.8%; Asian markets suffered heavier losses, with SK Hynix plunging more than 10% in a single day, Samsung Electronics falling 8.73%, and Japan's SoftBank Group dropping 7.9%. According to CNBC, as of Thursday this week, the 20 highest-market-cap AI and semiconductor stocks globally had collectively lost more than $1 trillion. The iShares Semiconductor ETF (SOXX), which tracks the chip sector, recorded its eighth consecutive trading day of declines on Thursday, with a cumulative drop of about 12.7% over that period, according to Dow Jones Market Data, the ETF's longest losing streak in 11 years.^[3]

This selloff was not triggered by a single event. Inflation data was the spark: U.S. September consumer price index (CPI) released on Wednesday came in above expectations, and the market immediately increased bets on further Federal Reserve rate hikes. The Nasdaq Composite closed down 1.88% at 26,692.87 on Wednesday, and the S&P 500 fell 1.11% to 7,718.94. Anthropic data added to the gloom: according to the Financial Times, AI startup Anthropic disclosed at a private investor meeting on Wednesday that third-quarter revenue reached $16.8 billion, up about 50% quarter over quarter, but this growth rate was far below the roughly 145% quarter-over-quarter increase in the second quarter. D.A. Davidson analyst Gil Luria believes the market seems to have interpreted Anthropic's update as a negative signal; but he also noted that Anthropic achieved positive cash flow in the third quarter, showing that the company and OpenAI are not inherently unprofitable business models.^[3]

Chinese competition is also intensifying. ChangXin Memory Technologies announced over the weekend that it had achieved a breakthrough in a new architecture that can store more data in a smaller space; Elon Musk then posted on X that China will produce a large number of memory chips in 2028. Meanwhile, the CEOs of Broadcom and Intel both warned on Monday this week that financing issues could become a major obstacle to AI expansion, with Broadcom CEO Hock Tan saying the challenge could even exceed chip or power shortages. Rising rates directly hit the return on AI infrastructure investment. James Vrachas, executive director at investment bank Post Oak Group, told MarketWatch that chip stocks have already priced in a great deal of future growth expectations, and a slight shift in sentiment or doubts about spending can significantly hit share prices. Baird investment strategist Ross Mayfield noted that rising Treasury yields are putting pressure on the chip sector, especially high-growth stocks and companies on the more aggressive end of the AI risk spectrum that rely more on external financing.^[3]

Asian and European markets fell more deeply. On Thursday, South Korea's SK Hynix plunged more than 10% in a single day, Samsung Electronics fell 8.73%, LG Innotek fell 8.95%, Seoul Semiconductor fell 9.28%, and Samsung SDI fell 6.26%; the KOSPI approached its lowest closing level since April. Japanese markets also tumbled, with SoftBank Group down 7.9%, semiconductor test equipment maker Advantest plunging nearly 14%, chip manufacturing equipment supplier Tokyo Electron down 6.45%, Renesas Electronics down 8%, and TSMC down 1.81%. In Europe, Dutch semiconductor equipment giant ASML fell slightly, Infineon dropped 5%, and ASM International fell 7%. Some analysts believe this selloff has not yet touched a substantive deterioration in AI demand. Yorkville Ives technology analyst Dan Ives wrote that this selloff reflects market concerns about financing costs, not doubts about AI computing demand; what is truly worth watching are demand signals, including hyperscale cloud providers cutting capital expenditure plans, contract cancellations, or falling GPU rental prices.^[3]

Data center bonds under collective pressure

According to a Financial Times report on Thursday, from special financing bonds for large data centers leased to Meta, Microsoft, and Nvidia to debt issued by emerging cloud computing companies such as CoreWeave, prices have fallen into stressed territory, reflecting deep investor doubts about the sustainability of returns on AI infrastructure investment. The direct triggers for this selloff include: OpenAI postponing its IPO to next year, Anthropic's quarterly revenue growth slowing, and OpenAI disclosing last week that actual annualized revenue was far below previous expectations. These signals intensified market doubts about AI monetization and triggered a linked decline in stocks and bonds. Nvidia lost about $270 billion in market value over the past week, CoreWeave shares fell about 17% during the week, and competitor Nebius dropped 18%.^[5]

Bond prices fell across the board, with some already at distressed levels. Bonds issued for the Hyperion project by Beignet Investor, a special purpose vehicle jointly funded by Blue Owl and Meta, fell from above par in early July to about 92 cents; bonds for two projects in Louisiana and Texas developed by Hut 8, the latter leased by Nvidia, fell to about 94 cents and 92 cents respectively; Blackstone-owned QTS Data Centers' Project Excalibur bond in Georgia, with Microsoft as tenant, has fallen below 88 cents. Emerging cloud computing companies have been hit even harder: CoreWeave-issued bonds have fallen below 80 cents, a level generally seen by the market as the symbolic threshold for distressed debt, with the corresponding yield approaching 15% at one point on Thursday. About $18 billion in loans related to Oracle's Project Jupiter data center in New Mexico also slid into stressed territory due to delays in power supply permits amid local opposition; construction bonds for Oracle's Michigan AI campus also fell from par to 87 cents.^[5]

The deeper logic behind the shift in market sentiment is the huge gap between AI infrastructure investment and predictable returns. A Gartner report released this week predicts that because of a multi-trillion-dollar gap between the pace of capital expenditure and AI return expectations, 80% of emerging cloud computing companies will stop operating independently before 2030. Gartner estimates that AI data center operators need to generate cumulative revenue of $8.5 trillion by the end of 2030 to cover capital and financing costs and achieve reasonable returns; however, it forecasts that consumer and enterprise AI service spending flowing to data center providers will not exceed $3.6 trillion. Gartner warns that if AI infrastructure investment maintains its current trajectory, more than $4 trillion in assets could face impairment risk by 2032. Ludovic Subran, chief investment officer of Germany's Allianz Group, said data center bonds are entering a 'natural selection' moment, and the weakest credits will struggle to survive.^[5]

In addition to fundamental pressure, two types of structural risk are drawing increasing investor attention. One is local political resistance: backlash against data centers has spread across many parts of the United States, triggered by concerns over water consumption, noise, and air pollution; New York State and Texas have imposed strict restrictions on new projects, and other states are considering similar measures. The other is circular financing risk. A recent report by the Bank for International Settlements pointed out that a considerable proportion of financing and business relationships in the AI field are self-referential, meaning customers and suppliers finance each other, and problems in one link could trigger a chain reaction. Despite overall pressure, signs of divergence are still emerging within the market. Tom Murphy, head of U.S. investment-grade credit at Columbia Threadneedle, said data center assets that have already been built and put into operation may instead gain a relative premium under current headwinds because of supply contraction. Nvidia CEO Jensen Huang said in a Bloomberg interview on Thursday that he does not think CoreWeave needs funding support from Nvidia, and said CoreWeave's cost of capital is actually falling and its strength is 'improving every day.'^[5]

Capital shifts to defense: large-cap pharma stocks approach record highs

While tech stocks are being sold off, capital is accelerating its shift into the large-cap pharma sector. Biopharma giants such as AbbVie and Eli Lilly saw their share prices approach record highs this week, reflecting a clear signal that investors are turning to defensive assets amid turbulence in AI concept stocks. On Thursday, the State Street Health Care Select Sector SPDR ETF hit a record high of $177.19, up 5.6% cumulatively over the past five trading days. Eli Lilly closed up 3.4% on Thursday, its third-highest close ever; AbbVie closed at $283.40, its second-highest ever, after setting a record of $283.48 the previous day. Neither company was driven by major clinical data or M&A news; they benefited purely from defensive rotation of capital.^[1]

Other large pharmaceutical companies in the sector also strengthened. Gilead Sciences and Amgen rose 7.7% and 7.2% respectively over the past five trading days, Merck rose 2.3%, and UnitedHealth Group surged 9.1% on strong third-quarter results. Eli Lilly is one of the core targets in this pharma sector rally. With a dominant position in the GLP-1 weight-loss drug market, Eli Lilly has become one of the most widely recognized success stories in the industry. It is worth noting that this week's share price rise was not driven by a specific catalyst, but shows a regular pattern: whenever AI concept stocks face headwinds or the industry hits a major milestone, Eli Lilly's share price often rises accordingly. Even though its GLP-1 business has already achieved breakthrough success, Eli Lilly still maintains a high-intensity forward-looking layout. According to recent remarks by CEO Dave Ricks, the company has completed 40 deals this year worth a total of $20 billion, and two new weight-loss drugs are approaching the regulatory approval stage.^[1]

AbbVie's strong performance has more specific fundamental support. The latest data for the subcutaneous formulation of the company's core drug Skyrizi show that the formulation has confirmed efficacy in Crohn's disease patients after 24 weeks of treatment, laying the foundation for its application to the U.S. Food and Drug Administration (FDA) for approval and potentially extending the drug's patent protection period. According to recent remarks by AbbVie executives, the subcutaneous formulation is expected to bring 'considerable' incremental sales; compared with the existing clinic injection method, the subcutaneous formulation allows patients to self-administer at home, significantly improving convenience. Skyrizi had sales of $17.6 billion last year and is widely used in patients with inflammatory bowel disease, Crohn's disease, and other autoimmune diseases. However, the strength of large-cap pharma stocks does not mean the entire pharmaceutical sector is doing well. The high-risk, high-reward biotech sub-sector is showing clear divergence. The State Street SPDR S&P Biotech ETF fell 2.8% over the past month, after the sector had seen a long-awaited recovery earlier in the year. Biotech sector performance is highly correlated with interest rate trends and the financing environment, and its fragility contrasts sharply with the stability of large diversified pharmaceutical companies. In the current market environment, investor preference is concentrating on the latter, and a certainty premium is becoming the core logic of this pharma rally.^[1]

Rising rates: supply-driven repricing of capital markets

A recent research report from Bridgewater Associates pointed out that the continued climb in global bond yields is driven by factors far more complex than rising energy prices. Bridgewater believes the recent sharp rise in rates reflects more the unexpectedly resilient economic growth than a simple energy price shock. Expanding AI capital expenditure, persistently high government spending, and rate-insensitive borrowing activity together form the deeper drivers of rising rates. At present, market expectations for the cumulative tightening of policy in major developed economies this cycle have converged, centering on a range of 125 to 150 basis points. At the same time, Bridgewater warns that economic growth has remained resilient mainly because the relatively loose financial conditions before the energy price rise are still providing support; now, rising rates combined with persistently high energy prices reduce the likelihood that economic growth will maintain its previous resilience, especially outside the United States.^[6]

Bridgewater characterizes the current shift in the market landscape as a fundamental reversal of the post-global-financial-crisis logic. Over the past decade or more, quantitative easing led to negative net public-sector asset issuance, weak private-sector borrowing, and low yields plus abundant liquidity jointly suppressed risk premiums. Now, the rise of modern mercantilism, the AI capital expenditure boom, and continued central bank balance sheet reduction are together driving a massive expansion in net asset supply, and the scale of asset issuance the market needs to absorb is close to the high point of the past half-century. At the same time, demand from asset-price-insensitive investors, including central banks and liability-driven institutions, has structurally declined, and a growing share of new assets must be absorbed by price-sensitive investors. Bridgewater believes this mechanism will push asset yields systematically higher: price-sensitive investors need higher return compensation to absorb increasing supply, and limited leverage capacity will keep raising that requirement as supply increases.^[6]

Bridgewater points out that the impact of intensified competition for capital is already visible in some markets: yields on the longest-duration bonds continue to rise, credit spreads of hyperscale cloud providers are widening, and financing pressure on some CCC-rated companies is beginning to show in spreads. However, overall risk premiums remain low, and the gap between stock earnings yields and real bond yields is at its narrowest historical level outside the dot-com bubble period. Bridgewater warns that as asset supply continues to increase, pressure from rising funding costs will be difficult to confine to only some markets for long. Once it transmits to stock prices, it could trigger a self-reinforcing chain reaction: this economic expansion relies heavily on asset prices to support corporate capital expenditure and household consumption, and a substantial stock market reversal would weaken both pillars at the same time; with fiscal deficits still high during a period of strong economic growth, once private demand contracts, the government's room to offset it through fiscal stimulus will be very limited.^[6]

IEA accelerates reserve release: diesel and Hormuz risks intertwine

Member countries of the International Energy Agency (IEA) agreed to complete the 100 million barrels remaining in the March oil reserve release plan and promised to release 20 million barrels of diesel from it 'as soon as possible.' According to Bloomberg, the IEA governing board met on Thursday, October 15 local time, and said in a statement afterward that member countries will speed up allocation of remaining reserves, including releasing 20 million barrels of diesel 'as soon as possible.' The IEA also noted that because some member countries released more than their initial commitments, the final actual release total will exceed the 400 million barrel cap announced in March. The 100 million barrels are an old account: the IEA announced in March 2026 a release of 400 million barrels of oil reserves to respond to supply disruptions caused by the Iran war; so far, member countries have released 330 million barrels, and this 100 million barrels is to make up the remaining gap. JPMorgan analysts clearly pointed out that this reserve release is 'to a large extent an accelerated delivery of previous commitments, rather than a new intervention of 100 million barrels.'^[2]

Oil prices reacted limitedly. WTI crude rose about 6% cumulatively this week, with a single-day gain of 3% on Thursday, and is currently trading below $98 per barrel; Brent crude is stable near $106 per barrel. Tensions in the Strait of Hormuz remain the main source of upside risk for the market. On diesel, discussions within EU member states show the direction is feasible but the scale is expected to be limited. So far, no EU member state has explicitly committed to releasing additional diesel inventories within the 20-day window set by the G7. Polish Energy Minister Miłosz Motyka said that if all parties reach consensus, Poland is ready to release oil and diesel inventories; France and Italy may consider releasing some additional diesel on top of their March commitments, but no final decision has been made. The IEA separately listing diesel this time reflects the most urgent contradiction in the current market. The Iran war and the Ukraine conflict have damaged refining facilities and disrupted tanker shipping, and diesel prices have hit record highs in multiple markets.^[2]

The IEA statement shows that member countries currently hold about 1.1 billion barrels of public emergency oil inventories, including more than 200 million barrels of diesel. U.S. Energy Information Administration (EIA) data show that U.S. crude inventories fell 8.6 million barrels last week, the largest weekly decline since February this year, while distillate inventories also fell 2.5 million barrels to a seasonal low. Geopolitically, according to Bloomberg, U.S. Central Command confirmed on Thursday that a container ship that sank in the Strait of Hormuz this week was hit by a projectile from Iran's Revolutionary Guard; the Singapore-flagged GFS Pride sank while being towed to shore. Nevertheless, more than 40 vessels passed through the Strait of Hormuz in the past 24 hours, showing that shipping has not been completely interrupted. Eurasia Group analysts wrote in a research note that the conflict is likely to escalate further; the firm raised the probability of escalation from 35% to 65% and noted that the current standoff is no longer sustainable for the United States.^[2]

Apple and Meta cooperation: on-device AI breaks the rivals' deadlock

Apple and Meta announced they are joining forces to bring Meta's strongest open-source AI model to Apple's high-end desktop computers, marking a break by two Silicon Valley rivals from their historical feud and a new alignment under the AI wave. According to CNBC, the two companies announced on Thursday (October 15) that developers will be able to run Meta's latest open-source model, Muse Spark Open, locally on Apple Mac Studio computers. Users can combine two to four Mac Studios to complete coding, video analysis, and other agent tasks; the model runs locally rather than relying on the cloud, balancing security and cost control. Meta AI head Alexandr Wang said Apple's Mac Studio provides enough memory to make running Muse Spark Open on a personal computer a reality.^[4]

This cooperation comes just one month after former Apple CEO Cook stepped down and John Ternus took the helm of Apple. Ternus responded on social media to Zuckerberg's post, saying it is amazing to run a model of Muse Spark Open's scale on desktop so efficiently, and that Mac Studio keeps pushing the boundaries of on-device AI. However, this solution has quite high hardware requirements. Currently, a Mac Studio with an M5 Ultra chip and 256GB of memory starts at $10,799, and a higher-spec version with 512GB of memory will be released later this month; in addition, fully running Muse Spark Open requires purchasing multiple machines at the same time. Muse Spark Open is Meta's strongest open-source model to date, using a mixture-of-experts architecture with 816 billion parameters, released under a license that allows limited commercial use. As an open-weight model, developers can download model weights directly from platforms such as Hugging Face and run them on Apple chips without additional configuration. Meta also specifically optimized the model for Apple hardware, making it run faster on Apple devices.^[4]

The rift between Apple and Meta has a long history. In 2021, Apple launched a new privacy policy that limited the targeted advertising capabilities of digital advertising companies; Meta immediately estimated that the change would reduce its 2022 revenue by about $10 billion, and relations between the two fell to a freezing point. Previously, Cook had publicly mocked Zuckerberg during the 2018 Cambridge Analytica scandal; Zuckerberg said in an interview with Joe Rogan last year that Apple 'has not invented anything great since the iPhone.' The last public cooperation between the two companies was in 2017. Meta is the latest link in Apple's recent broad alliance strategy in AI. In 2024, Apple announced cooperation with OpenAI to integrate ChatGPT into Siri, but according to documents disclosed by OpenAI in its legal dispute with Musk's xAI, that integration 'performed far below expectations' and failed to drive OpenAI user growth. Apple subsequently sued OpenAI in July this year for trade secret theft, and OpenAI denied the allegations. Earlier this year, Apple also announced an agreement with Google to incorporate the Gemini model and Google Cloud infrastructure into Apple Intelligence. Despite an expanding lineup of partners, Apple's AI strategy still faces criticism from investors and users, and the new version of Siri released last month is widely seen as still lagging behind chatbots and agents from top AI labs. For Meta, this cooperation is an important endorsement for its AI strategy revival, but Meta's revenue sources are still almost entirely dependent on advertising.^[4]

Common trends and follow-up watch

The common thread across these developments is that AI is shifting from a 'narrative premium' to a 'financing and return interrogation': chip stocks and data center bonds are under simultaneous pressure, rising rates and expanding asset supply are raising funding costs, and capital is rotating toward defensive assets such as pharma. In energy, the IEA's accelerated reserve release reflects that diesel and Hormuz risks remain important variables for inflation and growth. Going forward, watch how the tech earnings season prices inflation concerns, hyperscale cloud providers' capital expenditure plans, GPU rental prices and contract changes, and whether data center bond credit spreads can stabilize; whether the certainty premium in the pharma sector can continue will depend on the rate path and the biotech financing environment.^{[3][5][6][2][1][4]}