NEW YORK — Dec. 12, 2025 (6:00 p.m. ET) — A significant pullback in the U.S. stock market occurred on Friday as many of the most scrutinized AI stocks faced fresh concerns regarding profitability, margins, and the pace of payback from the expansive AI infrastructure investments. By the end of the trading day, the Nasdaq Composite fell 1.69%, the S&P 500 dropped 1.07%, and the Dow slipped 0.51%, marking the Nasdaq’s lowest close since November 25, according to Reuters’ market wrap.
The selloff was particularly pronounced for Broadcom and Oracle. Broadcom’s commentary on margins reignited skepticism about the profitability of custom AI silicon, while Oracle’s ambitions in AI-driven cloud services and connections to OpenAI’s data centers came under increased scrutiny.
On December 12, the market’s message was clear: although AI remains a dominant theme, investors are demanding more tangible proof of long-term earnings potential rather than just promises of substantial capital expenditures. Rising Treasury yields, prompted by policymakers voicing resistance to looser monetary policy, further pressured high-valuation growth stocks.
This macroeconomic backdrop intensified an ongoing debate on Wall Street regarding the viability of AI investments. Questions included: How profitable are AI chips and systems? How swiftly will hyperscaler and enterprise AI spending translate into recurring cash flow? Are data center timelines being delayed due to labor, materials, and power constraints? The Philadelphia Semiconductor Index suffered a notable decline of 5.1%, marking its weakest session since October 10, with every stock in the index losing ground.
Broadcom’s selloff was particularly alarming for AI investors, as the company occupies a pivotal role in two of the market’s most dynamic sectors: custom AI accelerators (ASICs) designed for hyperscalers and high-speed networking essential for connecting AI data centers. On Friday, Broadcom shares fell more than 11% after the company indicated that increasing sales of lower-margin custom AI processors were straining profitability, raising fears that the AI boom may not be as lucrative as some investors had hoped.
Despite forecasting a first-quarter revenue of approximately $19.1 billion, exceeding analysts’ predictions, Broadcom’s CFO cautioned that gross margins would likely fall by about 100 basis points sequentially, primarily due to a higher mix of AI-related revenue. While the company cited a significant $73 billion backlog expected to ship over the next 18 months, investor concerns centered on customer concentration, with a large portion of this backlog deriving from just five clients.
Oracle’s stock also faced pressure, serving as a barometer for broader market concerns about the financial implications of mega data center projects. Shares plummeted after Oracle projected its fiscal 2026 capital expenditures would be $15 billion higher than earlier estimates, casting doubt on the speed of AI profitability. The company’s ambitious plans to expand its AI infrastructure, particularly through a $300 billion deal with OpenAI, raised alarms over debt levels associated with such a buildout.
Despite Oracle’s assurances that all milestones remain on track, reports of potential delays in the construction of OpenAI-related data centers fueled further investor unease. Analysts indicated that concerns about data center construction bottlenecks, including delays and power availability, are becoming increasingly significant in assessing returns on AI investments.
While many AI chip stocks faced declines, Nvidia remained in focus due to its robust fundamentals and shifting policy landscape concerning exports to China. Nvidia announced it is considering augmenting production capacity for its H200 AI chips after receiving demand that exceeds current output levels. This follows recent reports that the U.S. will permit Nvidia to export H200 processors to China, albeit with uncertainties surrounding approval from Chinese authorities.
The market’s retreat also impacted companies benefiting from AI enthusiasm, such as SanDisk, CoreWeave, and Oklo, which suffered steep declines on the day. This broader downturn indicates growing investor caution, extending to leveraged AI infrastructure models and related sectors.
On a more positive note, Rivian saw shares surge by 17.9% after analysts praised its strategy shift toward a custom chip and clearer autonomy roadmap, exemplifying that the AI narrative encompasses more than just chips and cloud technologies.
Looking ahead, analysts are shifting focus from narrative-driven AI success to concrete monetization metrics. JPMorgan named Alphabet and Amazon as leading AI stocks for 2026, citing their robust AI and cloud infrastructure investments and anticipated improvements in operating performance. Predictions for Alphabet included a $385 price target and expectations tied to AI integration across Google’s ecosystem, while Amazon is projected to see enhanced growth in AWS and increased free cash flow.
As the market grapples with these shifts, additional factors are at play, including President Donald Trump’s executive order aimed at creating a uniform federal AI regulatory framework. Concerns linger regarding local resistance to AI infrastructure projects, as seen with recent zoning rejections in Arizona, which highlight the complex landscape surrounding AI growth. The convergence of these elements signals that AI’s trajectory is not merely a software story; it increasingly involves considerations of land, power, and local regulations.
In the upcoming week, investors will be closely monitoring various data points that could influence the AI market, including labor market and inflation figures that might affect Treasury yields and risk appetite. Key points to watch include AI margins, potential data center construction delays, and the ever-evolving landscape of U.S.-China trade relations. As the distinction between AI adoption and profitability becomes more pronounced, it is evident that the dynamics defining AI stocks will significantly shape their performance as we approach 2026.
See also
Bloom Energy (BE) Plummets 10% Amid AI Infrastructure Selloff Post-Oracle Update
AI’s Energy Demands Threaten U.S. Infrastructure and Market Stability Amid Growing Vulnerabilities
Google DeepMind Launches Gemini AI Lab in UK to Accelerate New Materials Discovery by 2026
AI Wildlife Videos Surge as Realism Blurs; Experts Warn of Authenticity Crisis
AI Trade Retreats: Broadcom, SanDisk, and Oracle Face Significant Midday Losses Amid Margin Concerns
