S&P 500 stock market profits are running at a pace that looks almost impossible to sustain. Earnings per share surged 51% year over year in the second quarter of 2026 and have climbed 26% across the past four quarters, putting corporate America well above its long-term earnings trend. Goldman Sachs agrees that something unusual is happening, but the bank’s conclusion is more nuanced than simply calling the market a bubble. Its strategists argue that U.S. companies are currently “over-earning,” helped by a combination of extraordinary artificial-intelligence spending, unusually strong semiconductor profitability and large investment gains flowing through the income statements of mega-cap technology companies. Those forces should weaken, but Goldman does not expect them to disappear in a way that causes profits to collapse. Instead, the bank sees earnings growth slowing sharply from today’s exceptional levels while remaining positive enough to support higher stock prices.
That distinction matters because the S&P 500 is now being valued on earnings that may look very different one or two years from now. If current profitability represents a temporary peak, then today’s seemingly reasonable forward valuation could be misleading. But if earnings merely decelerate from extraordinary growth toward more normal double-digit expansion, the market may be able to absorb that normalization without anything resembling a dot-com-style crash. Goldman currently forecasts S&P 500 earnings of roughly $415 per share in 2027 and $460 in 2028, implying growth of about 11% in both years. Its latest 12-month target for the index stands at 8,700, roughly 14% above the level used in its forecast. The debate, therefore, is no longer simply about whether AI has created a bubble. The more important question is how much of today’s profit surge is temporary — and how much remains once the most powerful tailwinds begin to fade.
S&P 500 Earnings Are Doing Something Usually Seen After Recessions
The scale of the current earnings acceleration helps explain why investors are becoming uneasy. Goldman says second-quarter S&P 500 EPS increased 51% from a year earlier, while trailing four-quarter earnings rose 26%. Historically, comparable bursts of profit growth have often appeared after major recessions, when companies rebound from deeply depressed earnings bases. That was the case after the 2008 financial crisis and again after the pandemic. This time, however, the U.S. economy has not simply emerged from a conventional downturn. Corporate earnings are instead being propelled by a powerful mix of AI investment, resilient consumer demand, elevated technology margins and unusually large gains from private investments held by some of the market’s biggest companies.
The strength has also been broader than analysts expected. FactSet data show that Wall Street actually raised aggregate third-quarter S&P 500 earnings estimates by 1.2% during July and August. Normally, analysts cut estimates as a quarter progresses and companies provide more detail about costs, demand and margins. Over the past decade, the average reduction during the first two months of a quarter was 2.1%. This time, the opposite happened. That tells investors that the current earnings cycle is not merely strong on paper; companies have repeatedly delivered enough operating momentum to force analysts to revise forecasts higher. The danger is that markets begin treating this unusual period as permanent. Goldman’s argument is that investors should expect a slowdown, but not necessarily assume that a slowdown means the earnings cycle has broken.
AI Spending Is Delivering an Earnings Boost That Cannot Accelerate Forever
The largest temporary force behind today’s profit boom is the explosion in artificial-intelligence infrastructure spending. Goldman estimates that AI-related capital investment is responsible for nearly half of S&P 500 earnings growth this year. Amazon, Alphabet, Meta and Microsoft are collectively on track to spend roughly $800 billion on capital expenditures in 2026, almost twice as much as they spent in 2025. That spending is flowing through the economy in multiple directions at once. Nvidia and other semiconductor companies sell accelerators, memory producers supply high-bandwidth chips, networking vendors connect enormous computing clusters, utilities provide the electricity, data-center operators build capacity and construction companies participate in the physical expansion of AI infrastructure.
Related articles
That creates an enormous earnings multiplier for the S&P 500, but it also creates a mathematical problem. Capital spending cannot keep doubling every year forever, particularly when the starting base has already become so large. Even if AI investment remains structurally high for many years, the rate of growth will almost certainly slow. Once that happens, the contribution from AI capex to aggregate corporate earnings growth will also decline. Goldman estimates that the current investment boom is adding roughly 11 percentage points to S&P 500 earnings growth this year, but expects that contribution to fade substantially and eventually become a marginal drag by 2028 as depreciation, financing costs and slower spending growth begin to offset the initial boost.
That does not necessarily mean the AI investment thesis fails. In fact, it could mean the opposite: AI may simply be transitioning from an extraordinary buildout phase into a more mature part of the economy. The problem for investors is that markets are often most enthusiastic when growth rates are accelerating, not when enormous businesses are merely growing at more normal levels. If hyperscaler spending remains massive but increases only 10% or 15% instead of 80% or 100%, the absolute dollar opportunity could still be huge while the earnings-growth contribution becomes much less dramatic. That is exactly the type of normalization Goldman believes investors should expect.
Semiconductor Margins May Be the Market’s Most Dangerous Hidden Number
Semiconductor profitability is the second major source of temporary earnings strength. Goldman estimates that gross margins across some areas of the chip industry are running close to 70%, with certain DRAM memory producers approaching 80%. Those are extraordinary economics for an industry that has historically been cyclical, capital-intensive and vulnerable to periods of oversupply. According to Goldman, margin expansion alone accounts for roughly one-quarter of semiconductor earnings growth this year, meaning a significant part of the sector’s profit surge is coming not merely from selling more chips, but from earning unusually high profits on each dollar of revenue.
That is where the risk becomes important for the broader index. High margins encourage investment. Semiconductor companies increase capacity, suppliers expand production and customers eventually gain more negotiating leverage as shortages ease. If that process pushes industry gross margins from roughly 70% today back toward the 15-year average near 55%, Goldman estimates the effect could reduce S&P 500 earnings by around 10% relative to current levels. That is a large number for one industry, even an industry as important as semiconductors, and it shows why today’s aggregate earnings can look deceptively strong.
This does not mean chipmakers suddenly become unprofitable. Nvidia, Broadcom, memory producers and networking suppliers could remain highly successful while their margins gradually normalize. The risk lies in valuation. If investors are applying premium multiples to earnings that were generated during a period of extraordinary scarcity and pricing power, those multiples can become more demanding even if stock prices do not move. A company can continue growing revenue while reported earnings slow simply because each new dollar of sales produces slightly less profit. At the index level, that kind of margin normalization could become one of the biggest reasons earnings growth slows from today’s 50%-plus pace toward something much more ordinary.
The S&P 500’s P/E Ratio Is Not Screaming 1999
One of Goldman’s more reassuring arguments is that the stock market does not look nearly as expensive when investors use forward earnings rather than heavily normalized historical profits. The S&P 500 trades at roughly 19 to 20 times expected earnings, according to recent estimates, which is above the very long-term historical average but nowhere near the levels associated with the dot-com bubble. A year ago, the forward multiple was closer to 23 times earnings, meaning the market has actually become cheaper on that basis even as the index remained elevated. That happened because earnings grew faster than share prices.
The technology sector shows an even clearer difference from the late 1990s. Information technology currently trades around 22.5 times forward earnings, compared with a peak close to 55 times earnings in early 2000. More importantly, today’s dominant AI companies generate enormous profits. Nvidia alone produced tens of billions of dollars in annual net income, while Microsoft, Alphabet, Meta and Amazon have highly profitable underlying businesses and balance sheets strong enough to finance enormous infrastructure programs. Many of the companies that attracted extreme valuations during the dot-com boom had little revenue and no sustainable profit.
That comparison does not prove the market is cheap or immune from a correction. It simply means the valuation debate is more complicated than “AI equals 1999.” Investors are paying premium multiples for companies that actually generate extraordinary cash flows. The key risk is not that the earnings are imaginary. It is that some of those earnings are being produced under conditions that may be unusually favorable. If growth and margins normalize, the market could still look expensive even without ever reaching the speculative extremes of the dot-com era.
A $150 Billion Accounting Boost Is About to Become Harder to Repeat
Goldman’s third warning is less visible because it does not come directly from selling products or services. Mega-cap technology companies generated more than $150 billion of “other income” from appreciating private investments during the second quarter, according to the bank’s analysis. Those gains represented approximately 12% of total S&P 500 quarterly earnings. In other words, a meaningful slice of the index’s extraordinary profit growth did not come from cloud subscriptions, advertising, software licenses or semiconductor shipments. It came from increases in the value of investment holdings carried on corporate balance sheets.
Those gains are legitimate under accounting rules, but they are not dependable. A private investment portfolio can appreciate dramatically in one quarter and contribute very little the next. It can also reverse. Goldman therefore expects this source of earnings growth to fade sharply in 2027, potentially creating an 8-percentage-point drag on year-over-year S&P 500 earnings comparisons. That does not mean companies suddenly lose cash or their core businesses weaken. It simply means the comparison becomes much harder because the previous year included a very large benefit that is unlikely to repeat at the same scale.
This accounting effect is another reason investors should be cautious about extrapolating recent EPS growth. When earnings rise 51%, it is natural to assume operating businesses are expanding at a similar rate. In reality, several temporary forces are contributing simultaneously. As those effects disappear, reported earnings growth could slow dramatically even if the underlying economy remains healthy. Goldman’s bullish case depends on investors accepting that deceleration rather than interpreting it as evidence that the entire profit cycle is collapsing.
Goldman’s 8,700 Target Only Needs Earnings to Keep Growing
Goldman’s outlook ultimately rests on a much less spectacular number than the 51% growth investors have just seen: roughly 11%. The bank expects S&P 500 earnings of around $375 in 2026, $415 in 2027 and $460 in 2028. If those forecasts prove accurate, the market can potentially advance without requiring investors to pay ever-higher valuation multiples. Earnings themselves would continue doing most of the work. That is important because the environment has become less favorable for multiple expansion.
The Federal Reserve has raised interest rates again, and the 10-year U.S. Treasury yield has returned to around 5%. At those levels, equities face genuine competition from fixed income. Investors can earn attractive yields from government bonds without accepting the volatility of stocks, which raises the hurdle rate for owning expensive companies. Higher rates also reduce the present value of distant future earnings, making speculative growth stocks particularly sensitive to changes in bond yields. The S&P 500 therefore cannot rely indefinitely on investors simply becoming willing to pay more for each dollar of earnings.
This is why Goldman’s forecast is fundamentally an earnings story rather than a valuation story. If profits continue growing around 10% to 11% annually, then an index target of 8,700 does not require a return to extreme multiples. The market could move higher while valuation remains broadly stable. But that also makes the earnings estimates more important than ever. If profit growth disappoints and valuation multiples do not expand, there is very little cushion.
The Biggest Risk Is Not a Bubble — It Is an Earnings Disappointment
This is where Goldman’s relatively constructive view still contains a serious warning for investors. The market does not need to resemble 1999 to suffer if earnings fall short. Suppose hyperscalers slow AI spending faster than expected, semiconductor margins normalize sharply, private investment gains disappear, higher oil prices squeeze corporate costs and tighter monetary policy begins weakening consumer demand. None of those developments alone would necessarily create a crisis. Together, however, they could make today’s earnings forecasts too optimistic.
That would immediately change the valuation picture. A forward P/E ratio of 19 or 20 times earnings looks manageable if profits actually grow toward $415 per share next year. If analysts are forced to cut that number materially, the same index level suddenly becomes more expensive. This is the subtle risk hidden inside the “over-earning” argument: earnings can disappoint without collapsing, and stocks can still fall because valuation was based on a level of profitability that proved difficult to sustain.
Recent market breadth also shows that investors are already becoming more selective. Even when headline indices remain resilient, many sectors and individual stocks have struggled under the pressure of higher rates, expensive energy and questions about the durability of AI demand. That matters because a market dominated by a relatively small number of highly profitable companies can appear healthier than the average stock underneath it. If earnings normalization spreads from semiconductors and mega-cap technology into other sectors, the S&P 500 may become increasingly dependent on whether its largest companies can continue carrying the index.
The S&P 500’s Next Move Depends on What Is Left After the Sugar Rush
Goldman’s description of the market as “over-earning” sounds bearish at first, but the underlying message is much more balanced. Yes, today’s S&P 500 profit growth is almost certainly unsustainable. AI capital spending cannot nearly double forever. Semiconductor margins are unlikely to remain permanently near extreme highs. Mega-cap investment gains cannot reliably add more than $150 billion every quarter. As those forces fade, earnings growth should slow sharply.
But slowing from extraordinary growth to roughly 11% is very different from an earnings recession. FactSet’s recent revisions show analysts are not yet seeing widespread deterioration; third-quarter forecasts increased when they would normally be cut. Corporate balance sheets remain strong, the largest technology companies are highly profitable and AI spending is still enormous even if its growth rate eventually moderates. That gives the market a path to keep advancing without requiring another speculative valuation boom.
The coming quarters will reveal whether that path is realistic. Investors should watch how quickly AI capex growth slows, whether semiconductor margins normalize gradually or abruptly, and whether underlying economic growth becomes a larger contributor to earnings as temporary accounting and investment gains fade. If those transitions happen smoothly, Goldman’s 8,700 target has a straightforward logic: profits keep rising, valuations remain broadly stable and the index grinds higher.
If they do not, today’s seemingly ordinary forward P/E ratio may prove deceptive because the earnings underneath it were temporarily too strong.
That is the real tension inside the S&P 500 stock market outlook. The market does not need another year of 51% earnings growth to justify higher prices. It does, however, need investors to believe that once the current profit surge fades, there is still enough genuine growth underneath it to support the next leg of the bull market.
And after corporate America has spent the past year making extraordinary earnings look almost normal, that may be a harder transition than it sounds.
Disclaimer
This article is for informational purposes only and does not constitute financial or investment advice. Readers should conduct their own research and, where appropriate, consult a qualified financial advisor before making investment decisions. This article was researched and drafted with the support of AI, then reviewed, fact-checked and edited by the editorial team before publication.
