As of the close on the 17th (U.S. Eastern Time), the S&P 500 rose 7.05 points, or 0.10%, to 6,843.22; the Dow Jones Industrial Average gained 32.26 points, or 0.07%, to 49,533.19; and the Nasdaq Composite added 31.713 points, or 0.14%, to 22,578.384. The Nasdaq 100 fell 31.135 points, or 0.13%, to 24,701.597.

 

Tech stocks were mixed. Google and Microsoft dropped more than 1%, with Google Class C down 1.05%, Microsoft 1.11%, Meta 0.08%, and Tesla 1.63%. Meanwhile, Nvidia rose 1.2%, Apple 3.17%, and Amazon 1.19%. Software stocks remained under pressure: AppLovin fell 3.66%, Salesforce 2.86%, and Datadog 2.11%.

 

A number of investment bank traders and frontline investors told Yicai that U.S. stocks have recently been stuck in a dilemma of “strong fundamentals but weak technicals”. Simply put, after the S&P 500 hit 6,800 points last October, the index has moved sideways for nearly half a year. Even though last week’s non-farm payroll data far exceeded expectations and the Consumer Price Index (CPI) cooled, the index failed to gather enough upward momentum to break above 7,000. Instead, institutional investors have continued selling tech stocks. It appears that U.S. stocks are starting to pay off their “AI debt”, as the market is genuinely concerned again about the massive capital spending by tech companies on artificial intelligence. In 2026, Google, Amazon, and Meta have committed to spending over $500 billion on AI, setting a new record.

 

 

The bull market in U.S. stocks has surged since October 2022, driven almost entirely by tech stocks and AI-related trades. However, this pattern has changed significantly over the past three months, especially this year.

 

Year-to-date, the tech sector of the S&P 500 has fallen 6%, while energy (+21%), materials (+16%), and consumer staples (+16%) have led the gains. Investors are not only worried whether the massive investments by big tech will deliver expected returns, but also concerned about the disruptive impact of AI on traditional industries such as software, and have shifted their focus to sectors with more reasonable valuations.

 

The surge in capital expenditure is the main driver. Over the past two years, higher capital spending was consistently rewarded with rising share prices, leading investors to believe that spending equaled future returns. Today, however, investors are far more focused on actual returns.

 

Several tech giants delivered disappointing results this earnings season, further denting market sentiment. For instance, Amazon’s stock plunged more than 11% in after-hours trading following its latest earnings report. The company announced $200 billion in capital expenditure for 2026, well above analysts’ expectations. Most of this funding will go toward expanding AI infrastructure, including data centers and Nvidia GPUs, to support the growth of its AWS cloud service.

 

One trader told Yicai that objectively, the earnings were not bad. AWS cloud revenue grew 24%, and the advertising business became consistently profitable. Earnings per share (EPS) only slightly missed estimates by about one cent. Normally, this might have been overlooked by the market, but investors are clearly more demanding now. Moreover, Amazon’s free cash flow plummeted 71% in the quarter. Since institutional investors prioritize cash flow, algorithmic selling in the stock intensified.

 

Microsoft, which also sold off sharply earlier, faces similar pressure. Markets are anxious about tech firms’ runaway AI capital spending. The commitment by Google, Amazon, and Meta to spend over $500 billion on AI in 2026 has amplified market fears.

 

Meta has also sparked another layer of concern. Last October, Meta teamed up with private equity giant BlueOwl to raise $27 billion through private bond issuance to build data centers, setting a record for private debt placement. Recently, however, it emerged that to secure this huge funding, Meta signed a “residual value guarantee”. This means that if the AI bubble bursts and the data centers lose value in the future, Meta must compensate creditors with its own cash. Risk was essentially not transferred at all — it was merely hidden in the footnotes of the financial statements.

 

At the same time, a run on private equity (PE) has emerged. Traditional PE funds are typically locked up for 10 years and insulated from liquidity crises, but the fund pool BlueOwl used this time included many investors demanding “semi-liquidity”. Amid panic, redemption requests hit 17% in a single quarter, forcing BlueOwl to suspend redemptions. Markets now worry: if PE runs out of money, who will keep funding the $27 billion data center? While this concern may be overblown, panic can be amplified as negative sentiment rises.

 

 

Hindered by market worries over tech stocks, the S&P 500 has struggled to climb above the 7,000-point mark. Going forward, the key factor may be whether institutional investor confidence can be restored.

 

Recent selling in big tech has come from institutional investors, while retail investors have been “buying the dip”, raising questions over whether professionals are sensing unknown risks.

 

According to Yicai, Goldman Sachs trader Lee Coppersmith wrote in a recent client note that the macro environment remains broadly supportive, fundamentals (earnings) are solid, but technical conditions have deteriorated, and the AI landscape has grown increasingly complex.

 

Specifically, overall U.S. corporate earnings are healthy and largely in line with expectations. About 75% of S&P 500 companies have reported results, with earnings per share rising roughly 12% year-on-year, and the median company posting high single-digit growth. Revenue has been supported by economic activity and a weaker dollar, profit margins have beaten forecasts, and some companies have raised their 2026 earnings guidance. None of this has been particularly controversial, nor has it been the main driver of recent stock price action. Pressure has instead come from the forward-looking debate among institutional investors.

 

The software sector, feared to be disrupted by AI, has fallen roughly 24% over the past three months, yet earnings estimates for the next two years are actually about 5% higher. “We see a similar dynamic across multiple AI-related industries — solid results, rising earnings estimates, but significant compression in price-to-earnings (valuation) ratios. Investors are increasingly focused on the durability of profits rather than near-term earnings strength,” Coppersmith said.

 

More importantly, capital expenditure by hyperscale cloud providers in the U.S. in 2026 is expected to reach approximately $660 billion, about $120 billion higher than estimates just a few weeks ago. What has shaken investor confidence is that this growth is coming at the cost of share buybacks.

 

Statistics show that S&P 500 buybacks declined roughly 7% year-on-year in the fourth quarter, and capital spending by hyperscale cloud operators will consume more than 90% of their free cash flow this year.

 

Even so, given the still-supportive underlying fundamentals, investment institutions tend to view the current pullback as a healthy correction. The question is how long the correction will last and how deep it will go.

 

Invesco Chief Global Market Strategist Brian Levitt told Yicai: “Most sectors of the S&P 500 remain higher year-to-date. Stock markets outside the U.S. have also maintained their prior gains. The market was previously led by a tiny number of companies, and leadership is now broadening out as expected. What we’re seeing is not a cyclical reversal or a so-called ‘AI bubble’ bursting, but a healthy market adjustment after slowing momentum. Despite increased volatility and somewhat weaker sentiment over the past week, the overall fundamental backdrop remains fairly positive.”