AI Stocks Jolt Global Markets as Valuation Fears Grow

December 15, 2025 at 07:10 UTC

5 min read

AI stocks sell-off illustration with declining charts and AI icons, reflecting market volatility

Key Points

A sharp pullback in AI leaders like Nvidia and Broadcom has rattled Wall Street and Asia
Investors are questioning whether massive AI capex will deliver profits fast enough
Oracle and Broadcom earnings sparked renewed doubts about the AI trade’s sustainability
Despite volatility, AI chip demand and spending plans from big tech remain very strong

AI-led tech sell-off spreads from Wall Street to Asia

Global markets are starting the final full trading week of 2025 on the defensive as renewed doubts about artificial-intelligence-driven earnings hit high‑growth technology names. In the US, a sell-off in AI bellwethers dragged the S&P 500 and Nasdaq lower last week even as the Dow Jones Industrial Average rose more than 1%, helped by financial stocks. Broadcom shares tumbled more than 11% on Friday after investors focused on lower margins and uncertain deals, despite earnings and guidance that beat expectations. Oracle slid 12.7% over the week after a disappointing fiscal second-quarter report and concerns about the pace at which its heavy AI infrastructure spending will translate into cash flow. The weakness in US tech spilled into Asia on Monday. Tokyo’s Nikkei 225 fell 1.5% and South Korea’s Kospi dropped 1.2%, with South Korean chipmakers Samsung and SK hynix among the biggest losers. MSCI’s broad Asia Pacific ex‑Japan index fell about 1%, and Hong Kong’s Hang Seng declined 0.7%, as investors reassessed richly valued AI beneficiaries across the region.

Nvidia and peers face scrutiny despite surging AI demand

Nvidia, one of the main winners of the AI boom, has become a focal point of the reassessment. Its shares are down about 17% from a 52‑week high of $212.19 reached in late October, closing last week at $175.02. The pullback comes amid growing concerns about a potential AI bubble and investor demands for clearer evidence that current AI spending will generate durable returns. Yet Nvidia’s latest results show demand remains intense. Fiscal third‑quarter revenue rose 62% year over year to $57.0 billion, an acceleration from 56% growth in the prior quarter. Data center revenue, which captures most AI hardware demand, climbed 66% to $51.2 billion, also faster than in fiscal Q2. Operating income increased 65% to $36.0 billion and earnings per share grew 67% to $1.30. Chief executive Jensen Huang said Blackwell chip sales are “off the charts” and cloud GPUs are sold out. Nvidia guided for fiscal fourth‑quarter revenue of $65.0 billion, plus or minus 2%, implying roughly 14% sequential and about 65% year‑over‑year growth. Even so, with the stock trading around 43 times earnings, analysts note that any sign of slower growth, margin pressure, tighter export rules or rising competition from in‑house chips at customers such as Alphabet and Amazon could trigger a sharper correction.

Oracle, Broadcom and the AI capex–profitability mismatch

Oracle’s latest earnings underscored the tension between surging AI investment and near‑term profitability. The company reported fiscal second‑quarter revenue of $16.1 billion, up 14% year over year, and diluted earnings per share of $2.10, up 91%. Net income nearly doubled to $6.1 billion, while remaining performance obligations jumped 438% to $523 billion, reflecting a large backlog of cloud and AI contracts. Oracle Cloud Infrastructure revenue grew 69% year over year, but capital expenditures of $12 billion were $4 billion above consensus. Management guided to a $15 billion increase in capex to $50 billion for fiscal 2026 to build out infrastructure for its AI backlog. Bank of America described the current gap between spending and revenue as an “investment curve issue” rather than a change in fundamentals and reiterated a buy rating, while cutting its price target from $368 to $300. Other analysts have taken a more cautious stance, pointing to Oracle’s growing debt load and heavy exposure to large AI customers. Broadcom’s results, meanwhile, “overdelivered” on AI according to one analyst, but its stock still fell sharply as investors focused on margins and deal uncertainty, highlighting how sensitive the market has become to any perceived weakness in AI‑related earnings.

Macro backdrop: Fed cuts, data watch and crypto volatility

The AI‑driven tech volatility is unfolding against a shifting macro backdrop. The Federal Reserve cut the federal funds rate by 25 basis points last week to a range of 3.50%–3.75%, its third consecutive reduction, and Chair Jerome Powell downplayed the prospect of renewed hikes. Markets are now reassessing the path of further cuts in 2026, with some traders pricing in more easing than the Fed has signaled. US futures were steady to slightly higher on Sunday and Monday after a turbulent week, as investors rotated from high‑multiple technology stocks into cheaper parts of the market. Attention is turning to a busy US data calendar, including November consumer price inflation, nonfarm payrolls, retail sales and other releases that could influence the Fed’s January decision. In digital assets, Bitcoin slipped below $90,000, trading around $89,000–$88,000 and contributing to a 0.9%–1.8% drop in total crypto market capitalization. Analysts noted rising derivatives open interest alongside falling prices and a shift in sentiment back to “extreme fear,” suggesting increased short positioning as crypto traders track swings in AI‑linked tech stocks and broader risk appetite.

Key Takeaways

AI remains a powerful earnings engine for leaders like Nvidia and Oracle, but markets are no longer willing to overlook the timing and profitability of massive capex plans.
Valuation risk has become central: richly priced AI stocks are reacting sharply to even modest disappointments in margins, guidance or contract visibility.
Macro factors such as Fed rate cuts and upcoming US inflation and jobs data are now interacting with AI sentiment, shaping risk appetite across equities and cryptocurrencies.