During this earnings season, the U.S. stock market has seen profit growth that "exceeded expectations," with tech giants experiencing a dual increase in "capital expenditure and AI returns."

The U.S. stock market is seeing positive news during the second quarter earnings season, with corporate profits growing well beyond expectations. At the same time, "hyperscalers," represented by ultra-large tech companies, are not only continuing to ramp up investments in AI infrastructure but are also sending positive signals about the accelerating realization of returns on AI investments, providing fundamental support for the market. As of July 31, 61% of the S&P 500 companies have reported their second-quarter results. According to insights from the Momentum Trading Desk and the mid-term earnings report from the Goldman Sachs strategy team, 64% of these companies exceeded market consensus expectations for earnings per share (EPS) by at least one standard deviation, marking one of the highest levels in history. Excluding the "other income" from equity investments by tech giants, the year-on-year growth rate of EPS for the S&P 500 reached 26% in the second quarter, surpassing the first quarter and marking the fastest growth since 2021; if the aforementioned earnings are included, the overall growth rate would soar to 45%, far exceeding the previous market consensus expectation of 22%. Meanwhile, hyperscalers such as Alphabet, Amazon, and Microsoft reported a combined capital expenditure (capex) of $182 billion for the second quarter, with cloud business revenues growing by 48% year-on-year, significantly accelerating from 39% in the first quarter. Analysts predict that by 2027, capital expenditure for hyperscalers will exceed $1 trillion, an increase of over $125 billion from forecasts before the earnings season, as the scale of investments in AI infrastructure continues to expand, leading to a heightened external financing demand that has drawn significant market attention towards changes in corporate capital structures. The proportion of companies exceeding earnings expectations has reached a historical high, but TMT stocks showed a lukewarm response. The breadth of earnings surprises in this earnings season is notable. Among the S&P 500 companies that have reported results, 64% exceeded consensus expectations by more than one standard deviation, a ratio only surpassed in the previous quarter, the third quarter of 2025, and during the reopening period of the COVID-19 pandemic in 2020-2021. However, the market's reward response to exceeding expectations has been relatively restrained, especially in the technology, media, and telecommunications (TMT) sector, which is underperforming. In the TMT sector, stocks with EPS that exceeded expectations lagged the S&P 500 average by 192 basis points on the day after results were announced, while comparable stocks outside the TMT sector outperformed by an average of 75 basis points. In contrast, in the past twenty years, the average performance of S&P 500 stocks exceeding expectations on the day following the announcements outperformed the index by 95 basis points, whereas this quarter, the figure has narrowed to only 39 basis points. Despite this, the equal-weighted S&P 500 index price and EPS have continued to rise in tandem, indicating that the fundamentals beyond the AI complex remain robust, and the structural differentiation in the market has not affected the overall bullish trend. EPS growth exceeding expectations accelerates, with AI infrastructure contributing about one-third of the growth. In terms of the structure of profit growth, the overall year-on-year EPS growth tracking value for the S&P 500 in the second quarter was 45%, of which 19 percentage points came from the combined "other income" of approximately $151 billion from Alphabet and Amazon, with Microsoft contributing another approximately $3 billion. Excluding these one-time items, the growth rate was still 26%, significantly higher than the 20% in the first quarter. For the median stock, the tracking value for the median EPS growth year-on-year in the S&P 500 stands at 12%, also above the market consensus expectation of 9% prior to the earnings season, indicating that the profit exceeds expectations is not solely driven by a few giants. Notably, stocks related to AI infrastructure contributed about one-third of the EPS growth of the S&P 500 in the second quarter. Analysts forecast that this proportion will further increase to over 50% in the second half of 2026 and into 2027, becoming the core driver of the index's profit growth. Alphabet, Amazon, Micron Technology, and NVIDIA were the top individual contributors for the quarter, with the top ten contributors accounting for 79% of the EPS growth of the index in the second quarter. Earnings expectations for 2027 have been revised upward, while cost pressures pose marginal risks. Strong performance in the second quarter has spurred widespread upward revisions of analysts' long-term earnings expectations. Since the beginning of the third quarter, consensus forecasts for the S&P 500's EPS for 2027 have been revised upward by approximately 1%, with the energy and financial sectors showing the most significant adjustments, while the majority of sectors have seen positive revisions and the breadth of these revisions remains positive. However, cost pressures eroding profit margins are a key risk facing corporate earnings at present. Over the past several quarters, the net profit margin of the median S&P 500 stock has essentially stalled, with companies exhibiting some resilience in dealing with tariff and energy price pressures, but overall improvements have been lacking. Among the companies that have reported results, the market consensus for third-quarter profit margins has been lowered for most sectors, with the consumer staples sector showing the largest downward adjustment (-53 basis points), and slight downgrades also seen in consumer discretionary, industrials, and materials sectors. The high profit margins of large tech companies have supported the weighted average profit margins of the S&P 500 to some extent, but this effect has not transmitted to the majority of stocks. Hyperscalers: Capital expenditures surge, and accelerated monetization of AI becomes apparent. The second quarter financial reports from hyperscalers are sending positive signals about the accelerated realization of returns on AI investments. The cloud business revenues of Alphabet, Amazon, and Microsoft collectively increased by 48% year-on-year, significantly surpassing market expectations and further accelerating from 39% in the first quarter. Meta Platforms revenue grew by 28% year-on-year, aligning closely with expectations. Analysts currently predict that the collective revenue for the four hyperscalers (Alphabet, Amazon, Meta, and Microsoft) will grow at an annualized rate of 18% over the next two years, with expectations of continued upward revisions. At the same time, the scale of capital expenditures and financing demand is expanding significantly. In the second quarter, these companies (including Oracle) disclosed a total capital expenditure of $182 billion, while free cash flow was only about $5 billion, raising approximately $101 billion through debt and equity financing combined. Analysts estimate that capital expenditures for hyperscalers in 2026 will reach about $793 billion, while the forecast for 2027 has jumped from $929 billion prior to the earnings season to over $1 trillion (a year-on-year increase of 33%), and it is expected that capital expenditures will continue to exceed operating cash flow from 2026 to 2028. Goldman Sachs credit strategists anticipate that these companies will issue about $400 billion in investment-grade bonds globally next year to support continued expansions in AI infrastructure investments. Management teams of various companies have also commented on financing strategies during earnings conference calls. Alphabet stated that it has significantly expanded its debt portfolio and has recently conducted equity financing to maintain a solid balance sheet; Meta reported that it is evolving its capital structure to increase debt proportion while expanding long-term capital sources, including partnerships with BlackRock; Amazon indicated it would flexibly evaluate various financing tools based on business growth needs; Oracle disclosed plans to raise about $40 billion (including debt and equity) in fiscal year 2027, which includes a $20 billion announced immediate equity offering. Volatility in AI stocks continues to follow historical momentum cyclic patterns, with improved prospects following deleveraging. In the past week, volatility in AI-related stocks has continued the consolidation pattern often seen following sharp upward movements in momentum factors. According to analysis from Goldman Sachs, since 1980, momentum factors have typically shown a period of adjustment following a rise of over 20% within three months, and the recent pullback aligns with this historical trend. Currently, the three-month realized volatility of Goldman Sachs long-short momentum factor is at some of the highest levels in decades (excluding recession periods), but the significant deleveraging actions taken by hedge funds and ETF investors recently, along with the robust fundamentals confirmed during the earnings season, suggest that subsequent rotation volatility is expected to stabilize. Goldman Sachs maintains a target price of 8,000 for the S&P 500 by the end of 2026 and an EPS forecast of $385 for 2027. NVIDIA, the largest unreported stock in the S&P 500 by market capitalization, is scheduled to announce its second-quarter results on August 26, which is expected to provide further directional signals for the market. This article is reprinted from Wall Street Insights, authored by Zhang Yaqi; edited by Wenwen.