Second-quarter earnings across the S&P 500 were robust, with companies tied to artificial intelligence infrastructure continuing to account for a substantial portion of overall profit growth.
S&P 500 earnings per share increased 31% from a year earlier during the quarter when excluding one-off income related to private investment holdings. AI infrastructure companies, including hyperscalers, generated approximately half of that increase, with earnings among the group climbing 54% year-over-year.
Strength was not confined to AI-related businesses. Excluding the energy sector, which received a boost from higher oil prices, the median S&P 500 company recorded earnings growth of 14% compared with the same period last year.
However, Goldman Sachs strategists led by Ben Snider said evidence that corporate adoption of AI is translating directly into earnings improvements remains limited. Only 11% of S&P 500 companies quantified productivity benefits from AI for a specific business application during their earnings calls, such as software coding or customer support. Just 2% put a figure on AI’s direct contribution to earnings, unchanged from the first quarter.
Goldman’s analysis also found little difference in earnings performance between companies reporting measurable AI productivity improvements and those that did not.
“Q2 results showed a small and statistically insignificant difference in earnings growth between the companies quantifying AI productivity gains this quarter and other S&P 500 companies,” the strategists wrote.
Goldman nevertheless expects the financial impact of AI adoption to become clearer as corporate spending increases. Data from the Ramp AI Index shows that monthly AI expenditure per employee at the median company rose from $5 in January to $12 in July. Among companies in the top decile for spending, the figure surged from $240 per employee per month at the beginning of the year to $650 by July.
Despite that rapid increase, AI inference expenses remain relatively modest, with Goldman estimating they represent less than 0.5% of S&P 500 revenues. Around two-thirds of companies are financing AI investment by shifting money from existing budgets rather than adding entirely new spending, according to Goldman’s IT Spending Survey. Software is the most common source of redirected funds at 18%, followed by labour at 11%.
Evidence that AI is broadly displacing traditional software spending also remains limited. Some businesses, including Starbucks, are developing proprietary AI systems to replace third-party software, but software industry revenue growth has actually accelerated modestly in recent quarters. Software shares have also recovered from earlier declines.
Goldman believes this lack of clarity is reflected in how investors are positioning themselves. Markets have favoured AI infrastructure companies where the earnings benefits are already visible and relatively near term, while remaining more cautious about trying to identify which businesses will ultimately secure the largest productivity and profit gains from wider AI adoption.