The biggest keyword in this round of earnings season for U.S. stocks - AI anxiety

This is a financial reporting season with a significant divergence between fundamentals and market sentiment. Goldman Sachs believes that the core contradiction of this earnings season lies in the notable divergence between strong corporate fundamentals and the market's panic over AI disruption. Earnings have exceeded expectations, and revenue has shown steady growth, but the market's pricing logic has been dominated by the AI narrative.

On March 4th, according to news from the Chasing Wind Trading Desk, Goldman Sachs' economic research team stated in their latest report that the fourth quarter earnings season for the S&P 500 index is nearing its end. From a purely performance perspective, it has been a remarkably bright quarter—corporate earnings have grown by 13% year-on-year (with initial expectations of only 7%), and actual revenue growth excluding the energy sector has increased by 4.6% year-on-year, surpassing the 3.5% average during the past fifty years of economic expansion. However, all of this is overshadowed by one term: AI anxiety (AI-nxiety).

The report systematically outlines three core themes of this earnings season—the realization of AI productivity, expectations of AI's impact on employment, and the explosive growth of capital expenditures by mega-scale tech companies. Meanwhile, the "K-shaped economy" narrative has once again gained traction, but actual sales data shows that the degree of divergence is not as severe as the market fears.

Goldman Sachs states that the uncertainty surrounding AI is reshaping market pricing logic—industries considered highly exposed to the risks of AI automation (such as software, financial services, media, and entertainment) have shown significant underperformance since the beginning of the year; while the capital expenditure expectations for mega-scale tech companies have been significantly raised, driving overall business investment to become the strongest component of GDP growth by 2026. Investors need to find pricing anchors between "AI disruption panic" and "delayed realization of AI dividends."

Fundamentals Remain Strong, but Fully Suppressed by the AI Narrative

The report states that the hard data from this earnings season is undoubtedly positive:

The overall earnings of the S&P 500 have grown by 13% year-on-year, far exceeding the initial expectation of 7%;

The median company's earnings have grown by 10%, reflecting the broadness of growth;

Analysts' revisions for 2026 earnings expectations are positive, breaking the historical trend of typically lowering next year's expectations in the fourth quarter;

Actual revenue growth excluding the energy sector has increased by 4.6% year-on-year, consistent with the pace of the past few quarters and above the 3.5% average during the expansion period of the past fifty years.

However, Goldman Sachs believes that these solid numbers have been almost completely ignored at the market level. The frequency of discussions about AI during earnings conference calls has reached an all-time high, with investors' attention dominated by concerns over AI potentially disrupting specific industries.

Goldman Sachs data shows that industries with higher exposure to AI automation—those with a higher proportion of labor costs relative to revenue and where labor is more easily replaced by AI—have significantly underperformed since the beginning of the year, including sectors such as software and services, financial services, and media and entertainment.

AI Productivity: Heated Discussions, Few Quantifiable Realizers

Goldman Sachs states that client discussions mainly focus on three macroeconomic-related backgrounds. The first core topic of AI during the earnings season is productivity enhancement, but the data reveals a significant "gap between words and actions":

70% of S&P 500 management mentioned AI during earnings call meetings, a record high;

54% of management mentioned AI in the context of productivity and efficiency;

However, only 10% of management quantified the impact of AI on specific business scenarios;

Only 1% of management quantified the actual impact of AI on profits.

The adoption of AI by small and medium-sized enterprises is lagging further behind: among the broader Russell 3000 constituents, only 50% of management discussed AI; the U.S. Census Bureau's "Business Trends and Outlook Survey" shows that currently, less than 20% of businesses use AI for any business function.

Despite not finding a significant correlation between productivity and AI adoption rates at the macroeconomic level, Goldman Sachs noted that among companies that have quantified the productivity impact of AI, the median productivity increase is about 30%. The most commonly quantified application scenarios are concentrated in customer support and software development.

AI and Employment: Emerging Concerns, Macro Impact Not Yet Significant

According to research reports, the second AI topic during earnings season is employment and hiring intentions, one of the most sensitive topics in the market:

In the fourth quarter, the proportion of management discussing AI in relation to layoffs and hiring freezes increased, although the absolute proportion remains limited;

At the macroeconomic level, Goldman Sachs has not yet found a significant correlation between labor market outcomes and the degree of AI exposure or adoption;

However, Goldman Sachs pointed out in the research report that an early signal worth noting has emerged: Companies that discussed the relationship between AI and labor during earnings calls saw a 12% decrease in job vacancies over the past year, while the average decrease for all companies was 8%—this suggests that some companies may have begun to restrain their hiring intentions in anticipation of productivity gains from AI.

The research report states that from a longer-term perspective, multiple business surveys show that respondents generally expect AI to reduce their labor demand, and this impact is expected to grow over time. Goldman Sachs' baseline estimate is: In the long term, AI automation will lead to the replacement of 6% to 7% of workers (equivalent to about 11 million jobs).

Capital Expenditures: Mega Tech Companies Exceed Expectations, Driving Business Investment

Goldman Sachs stated that the most explosive data during earnings season comes from capital expenditures, especially from mega tech companies (Amazon, Meta, Google, Microsoft, Oracle):

Analysts have significantly raised their 2026 capital expenditure expectations for mega tech companies by 24% from the beginning of earnings season to $667 billion, corresponding to a 62% year-over-year increase compared to 2025;

Goldman Sachs believes that this figure represents analysts' expectations and serves as a proxy indicator for corporate guidance, indicating that investments in AI infrastructure by mega tech companies are still accelerating.

According to the research report, beyond mega tech companies, overall capital expenditure expectations are also strong, partly due to the more generous corporate tax incentives provided by the One Big Beautiful Bill Act (OBBBA):

Analysts expect the median capital expenditure of S&P 1500 companies to grow by 7% in 2026, up from 3% in 2025;

Industries more significantly impacted by the OBBBA tax incentives (where the reduction in capital costs is more pronounced) are expected to see higher capital expenditure growth rates.

Goldman Sachs predicts that business investment will become the strongest component of GDP growth in 2026, with a year-over-year Q4/Q4 growth rate of 5.2%. AI investment is expected to contribute approximately 1.5 percentage points to capital expenditure growth, but due to a significant portion of spending being allocated to imported equipment, its net contribution to GDP growth is only 0.1 to 0.2 percentage points.

K-shaped Economy: Narrative Stronger than Reality, Low-Income Consumer Resilience Exceeds Expectations

Meanwhile, during this earnings season, the narrative of the "K-shaped economy" has once again gained prominence.

The "K-shaped economy"—where consumption diverges between high-income and low-income groups—has become another high-frequency term this earnings season, but Goldman Sachs' data analysis shows that the degree of narrative divergence is far greater than the actual sales data divergence:

For retailers primarily located in low-income postal code areas, the sentiment index related to consumer discussions during earnings call conferences has fallen to its most pessimistic level since 2020;

However, these retailers' nominal same-store sales grew by 1.4% year-over-year, significantly accelerating from 0.2% in the third quarter of 2025;

In contrast, the same-store sales growth rate for retailers targeting middle to high-income consumers was 2.5%;

The sales gap between the two narrowed from 2.3 percentage points in the third quarter to 1.1 percentage points in the fourth quarter.

However, Goldman Sachs believes that the K-shaped economy narrative is somewhat exaggerated. Looking ahead to 2026, low-end consumption indeed faces more headwinds:

A decrease in immigration will further suppress employment and income growth for low-income groups, and the impact of government spending cuts will also be more concentrated on low-income groups. Middle to high-income groups will benefit from new personal tax cuts and the stock wealth effect.

Goldman Sachs expects overall consumer spending to grow at a robust rate of approximately 2.2% in 2026 (Q4/Q4)