Goldman Sachs strategist Ben Schneidau said that companies are increasing their investments in artificial intelligence (AI) at an unprecedented pace, yet for most firms, this technology has not yet translated into tangible improvements in earnings.

In a report published by Goldman Sachs on August 14, it said that during this year’s second-quarter earnings season, only 2% of the constituents of the $S&P 500 Index (.SPX.US)$ had quantified the specific impact of AI on earnings. Another 11% said they had observed measurable productivity gains in specific areas such as software programming and customer support.

However, the companies that have implemented these efficiency improvements have not significantly outperformed overall market levels in terms of profit growth. Data shows that their median year-over-year profit growth was 17%, while companies that have not quantified the contribution of AI efficiencies were at 14%. Goldman Sachs noted that this gap is not statistically significant.

For investors, this finding helps explain the current market landscape: on the one hand, AI-infrastructure beneficiary stocks such as semiconductor manufacturers and cloud computing service providers continue to attract strong investor interest; on the other hand, companies that promise future efficiency improvements are generally met with caution. Infrastructure spending has already produced immediate revenue and profit growth, while the potential returns that companies may achieve by using AI to improve efficiency are still difficult to quantify and may only gradually become visible over several quarters.

AI infrastructure drives earnings growth

Overall, performance during the second-quarter earnings season has been particularly strong. Goldman Sachs said that after excluding one-off gains related to some private equity investments, S&P 500 earnings per share grew 31% year over year.

Among ultra-large-scale enterprises and other beneficiaries of AI capital expenditures, earnings growth can be as high as 54%, contributing roughly half of the index’s overall earnings growth.

Even so, growth momentum is not limited to large technology stocks. The median earnings growth among S&P 500 index constituents is 14%. And excluding energy companies that benefit from rising oil prices, the growth rate for non-AI infrastructure companies is also 14%.

This broader improvement may help ease concerns in the market that earnings growth is overly dependent on a small number of tech giants. However, the performance gap between infrastructure providers and AI application-focused companies remains significant.

Goldman Sachs noted that investors favor infrastructure stocks because their returns are immediate and relatively easy to track. By contrast, the portfolios of companies that frequently highlight AI productivity plans have broadly performed in line with the overall S&P 500 index over the past few years.

Corporate spending accelerates expansion

Various signs indicate that the impact of AI will become clearer in the financial performance of companies over the next few quarters.

Goldman Sachs cited the Ramp AI Index showing that a company’s monthly average AI spending per employee has risen from $5 at the start of the year to $12 in July; for the top 10% of companies, this spending jumped from $240 to $650.

During second-quarter earnings calls, about 7% of S&P 500 companies discussed AI deployment costs. Most companies said the scale of such spending is still small, or emphasized that investments will proceed on a cautious basis. Some companies also said that the benefits from AI have already exceeded the costs.

Goldman Sachs estimates that, at present, the share of AI inference costs in the revenue of S&P 500 companies is less than 0.5%. Its latest IT spending survey also shows that 89% of respondents say AI spending accounts for 1% to 5% of their IT budgets.

It is worth noting that the above estimates do not cover all costs related to AI deployment, such as staffing for specialists and building technical infrastructure.

Shifting spending within existing budgets to support the transition

Goldman Sachs’ survey found that about two-thirds of companies support their AI investment by reallocating resources from existing budgets, rather than relying entirely on new funding.

Specifically, 35% of respondents said their AI spending comes from new budgets; 18% fund it through cost-efficiency improvement initiatives; another 18% reallocate funds from software budgets, 11% shift from labor costs, 10% come from cloud services budgets, and 9% come from data analytics spending.

Goldman Sachs believes that structural adjustments within IT budgets are more likely to reallocate profits across different companies rather than significantly change the overall earnings level of the S&P 500. But if labor costs are cut drastically, it could lead to broader economic impacts.

However, for now, the labor market impact remains concentrated in areas such as marketing, graphic design, customer service, and some technical roles, while new jobs created by data center construction to a certain extent offset the reduction in the roles mentioned above.

Goldman Sachs economists expect that AI will ultimately replace some labor, but they believe this impact will be temporary and smaller than many investors have anticipated.

Software industry impact is not yet obvious

The market had previously worried that customers would develop applications using AI themselves, thereby reducing reliance on external software vendors. But Goldman Sachs found that there has not been widespread industry reshuffling so far.

In its IT survey, only 17% of respondent companies said they plan to increase internal software development and reduce purchases of off-the-shelf software. Among software companies covered by Goldman Sachs, the median annual recurring revenue (ARR) growth rate rose from 18% in Q4 2025 to 22% in this year’s first quarter, accelerating further to 23% in the second quarter.

Of course, this does not mean that individual vendors can rest easy. Goldman Sachs noted that there are reports that $Starbucks (SBUX.US)$ is developing in-house AI tools to replace some of the software currently provided by companies such as $Microsoft (MSFT.US)$ and $IBM Corp (IBM.US)$. But based on industry-wide data, there has not been a widespread deterioration trend yet.

Goldman Sachs screens potential beneficiary targets

Goldman Sachs believes that companies with higher labor costs and where many positions have potential for automation will ultimately benefit the most from AI.

Currently, labor costs account for about 12% of total revenue for S&P 500 companies, or roughly $2.1 trillion per year. The differences across sectors are significant: industrial companies account for 21%, the information technology sector for 16%, and the energy sector for only 5%.

Goldman Sachs screened companies from the Russell 1000 index constituents with higher labor costs, greater potential for AI automation to replace roles, and management that had mentioned AI-related efficiency improvements in earnings reports. The list includes: $Costar (CSGP.US)$, $Dollar Tree Inc. (DLTR.US)$, $eBay (EBAY.US)$, $Arthur J. Gallagher (AJG.US)$, $Axon Enterprise (AXON.US)$, $The Trade Desk (TTD.US)$, $Airbnb (ABNB.US)$, $Boeing (BA.US)$, $Lockheed Martin (LMT.US)$, $Charles Schwab (SCHW.US)$ and $Morgan Stanley (MS.US)$.

However, Goldman Sachs also pointed out that this screened list does not mean these companies have already achieved significant AI-driven cost reductions and efficiency gains— in fact, the earnings data of these potential beneficiaries has not yet shown any meaningful improvement.

For now, the investment logic for AI remains sharply distinct: on one side are infrastructure providers, whose profits are already clearly visible; on the other side are technology deployers, whose financial returns still remain more in the realm of expectations.

The above information is for reference only and does not constitute any investment advice!