The third-quarter reporting season for U.S. stocks will officially kick off next week. Although the earnings growth rate of the S&P 500 index is expected to reach its highest level since 2021, this growth is highly concentrated among a few tech and energy giants. In contrast, the earnings momentum of smaller companies and most industries is weakening, leading to a further widening gap between the performance of the index and individual stocks.

According to the latest forecast from Goldman Sachs, the year-over-year growth rate of earnings per share for the S&P 500 index is expected to reach 27% in the third quarter, the highest since 2021. However, this growth rate is highly dependent on a few companies: memory chip manufacturers Micron and NVIDIA are expected to contribute about one-third of the index's earnings growth, while the median growth rate of earnings per share for S&P 500 constituents is only 9%, a further decline from 14% in the second quarter.

The concentration of profits and the narrowing of market breadth are becoming the core risks that investment banks are concerned about. Goldman Sachs pointed out that market breadth has dropped to its lowest level since the internet bubble—the S&P 500 index is only 2% away from its historical high, but the average decline for median stocks from their respective highs has reached 17%.

Morgan Stanley strategist Mike Wilson believes that the divergence between the index approaching record highs and a large number of stocks experiencing bear market-like pullbacks “will eventually converge in some way,” and that bond market volatility may become the ultimate “referee.”

Profit growth is becoming increasingly dependent on a few stocks.

The latest report from Goldman Sachs shows that Micron is expected to contribute 19% to the third-quarter earnings growth of the S&P 500, while NVIDIA contributes 15%. The combined contribution of these two companies is roughly equivalent to the total contribution of the remaining 490 constituents.

The contributors that follow closely include Meta, Alphabet, and Broadcom, contributing 7%, 6%, and 5% respectively; ExxonMobil, Chevron, and Marathon Petroleum benefited from the rising oil prices due to the situation in the Strait of Hormuz, and Boeing also made it into the top ten. The top ten contributors are expected to account for a total of 68% of the third-quarter profit growth, which is significantly higher than 47% in the first quarter and 48% in the second quarter.

From the perspective of the topic, AI infrastructure is the main source of profit growth. Goldman Sachs estimates that “AI infrastructure (excluding hyperscale cloud providers)” will contribute 54% to the third-quarter profit growth, with hyperscale cloud providers contributing another 19%. Together, these two sectors account for about three-quarters of the profit growth of the S&P 500.

The industry differentiation is also evident. Affected by the situation in the Strait of Hormuz and rising oil prices, the earnings per share of the energy sector are expected to increase by 109% in the third quarter; the information technology sector is expected to grow by 64%, with both contributing about 80% of the quarterly profit growth. In contrast, the discretionary consumption sector is expected to see virtually no profit growth, while the essential consumption sector is expected to experience negative growth.

Goldman Sachs also pointed out that analysts expect profit growth rates in nearly all sectors to slow down quarter-on-quarter, with information technology and communications services being the only two exceptions. Since the third quarter, the profit margin forecasts for the median companies in the S&P 500 have been lowered by 11 basis points, and all sectors have seen reductions except for information technology.

Micron takes the lead in reporting results; beyond high growth, the market is more concerned about sustainability.

Micron announced its financial results on September 30th, showing strong performance: the adjusted earnings per share were $33.42, higher than the market's expectation of $31.83; adjusted revenue increased from $11.2 billion in the same period last year to $54.2 billion; the gross margin reached 87%, significantly higher than 45.7% in the same period last year. The company provided a revenue guidance range for the next quarter of $60 billion to $63 billion, which is also significantly higher than the market's consensus forecast of $56.77 billion.

However, the significantly better-than-expected performance did not lead to a corresponding surge in stock prices, as the market had already factored in higher growth expectations. Before the release of the financial report, Morgan Stanley analyst Joe Moore pointed out that market discussions regarding Micron "had clearly shifted from 'how good can it get' to 'how long can this momentum last'."

Since the beginning of this year, Micron's stock price has risen by more than 231%, making it one of the best-performing stocks in the Philadelphia Semiconductor Index. With the significant increase in both performance and stock price, market attention has shifted from whether short-term performance will exceed expectations to whether the demand for AI storage and the high level of profitability can continue.

Super-large cloud providers are increasing their capital expenditures, but market expectations may still be low.

The core driving the profit growth of companies related to AI is the continuous increase in capital expenditure by large-scale cloud providers. The market expects that these companies' capital expenditure in the third quarter will increase by 116% year-on-year, higher than the 87% in the second quarter. Goldman Sachs predicts that the growth rate of capital expenditure will exceed 50% in 2027, and the total scale will also surpass the current market consensus of about $1.1 trillion.

Historical experience shows that market forecasts for capital expenditures of ultra-large cloud vendors have consistently lagged behind the actual growth rates: at the beginning of 2024, the market expected a full-year growth of 19%, but the actual growth was 54%; at the beginning of 2025, the forecast was 22%, while the actual growth was 73%; at the beginning of this year, the forecast was 36%, and the current actual growth rate is already close to 96%. Therefore, Goldman Sachs believes that the market's expectation of a 37% capital expenditure growth rate for 2027 may also be on the low side.

The acceleration in cloud business revenue is also providing support for this round of capital expenditures. Cloud revenue growth rates for large-scale cloud providers such as Amazon, Google, Microsoft, and Oracle accelerated to 48% in the second quarter, and Goldman Sachs expects it to further rise to 55% in the third quarter. Among them, the total backlog of cloud service contracts for Amazon, Google, and Microsoft has approached $1.7 trillion.

However, the rapid expansion of AI's capital expenditures has also come with a higher demand for financing, a considerable portion of which is being met through debt financing. In addition, there were two accounting factors that contributed to higher corporate profits in the first half of this year: the income generated from the valuation of private AI company equity held by large technology companies based on market value, as well as tariffs refunded by the U.S. Treasury Department.

Goldman Sachs estimates that large technology companies generated approximately $150 billion in revenue from the revaluation of equity in private AI companies in the second quarter, which is equivalent to 12% of the earnings per share of the S&P 500; meanwhile, the approximately $69 billion in tariffs refunded by the U.S. Treasury Department in the third quarter is equivalent to about 6% of the pre-tax profits of companies across the United States. Goldman Sachs has already excluded this revenue from its quarterly year-on-year comparisons and does not expect to see a similar level of revenue again in the third quarter.

Indices near historical highs, market breadth continues to narrow

In addition to the concentration of profits, market performance also exhibits a rare high degree of concentration. The top ten constituents of the S&P 500 currently account for approximately 40% of the index's market value and 37% of the expected earnings. Goldman Sachs data shows that the average realized correlation coefficient among S&P 500 constituents has dropped to 0.06, which is the lowest level in the firm's ten-year data, indicating that the overall performance of the index is becoming increasingly dependent on a few key companies and energy leaders.

Bank of America Merrill Lynch strategist Michael Hartnett also pointed out that currently, about 400 of the S&P 500 constituents have fallen below their 50-day moving average, and around 300 have fallen below their 200-day moving average. The market value concentration calculated by him, which includes AI Big 10 – Mag 7 plus Broadcom, AMD, and Micron – has reached 42%, exceeding the 40% during the "Nifty Fifty" era, the 41% during the bubble period of TMT, and the 36% of the 1920s. Among comparable historical cases, only the Japanese market's 44% and the 63% of U.S. railway stocks in 1881 are higher than this level.

However, investment banks remain relatively optimistic about the overall performance for the third quarter. JPMorgan Chase's market intelligence department adjusted its market outlook in its third-quarter report to "tactical bullish," believing that after a 52% profit growth in the second quarter, the market's expectation of around 29% profit growth for the third quarter may still be conservative. It is also anticipated that all 11 industries will achieve revenue and profit growth.

FactSet data shows that if expectations are met, the S&P 500 will achieve year-over-year revenue growth of over 10% for the third consecutive quarter, earnings growth of over 25%, and profit margins will also reach the second-highest level in history.

Goldman Sachs currently expects that the earnings per share of the S&P 500 index will reach $375 in 2026, a year-on-year increase of 36%; it is further projected to rise to $415 in 2027, with target levels of 8,000 points by the end of the year and 8,700 points over the next 12 months.

As the largest weight in the S&P 500 and also the second-largest contributor to earnings growth in the third quarter, NVIDIA will release its financial report in the third week of November. In the current market environment where earnings are highly concentrated, its performance and guidance on future AI capital expenditures and demand may become a key factor in determining whether this round of U.S. stock earnings growth can continue.