S&P 500 Hits Another Record While Three Stocks Do Most of the Work
The S&P 500 closed Friday at 7,757.64, its 26th record high of 2026, capping a 3.6% week. It got there days after US payrolls unexpectedly fell by 23,000, and while Iran signalled that a deal to reopen the Strait of Hormuz is not close.
Six months into a war that has cut shipping through the strait to a trickle, America’s benchmark index keeps setting records. Whether the price makes sense depends almost entirely on which earnings number you trust.
Valuation is expensive, not insane
On the measure Wall Street uses most, the market does not look absurd. FactSet puts the forward 12-month price/earnings ratio for the S&P 500 at 20.0, above the five-year average of 19.9 and the ten-year average of 19.0, but below the 20.4 recorded at the end of June. Multiples have compressed slightly this summer, because forward earnings estimates rose faster than prices did, which is the healthiest way for an expensive equity market to de-rate.
Longer-term measures look worse. The trailing price/earnings ratio is 28.2 against a ten-year average of 23.5, and the cyclically adjusted P/E sits near 41, the second-highest reading in roughly 150 years of data. At a trailing earnings yield of about 3.6%, the index now yields less than the 10-year Treasury, which ended July around 4.74%.
Three stocks, a quarter of the gain
The S&P 500 returned 13.7% through 7 August. Nvidia contributed 1.49 percentage points of that, Micron Technology 1.13 and Apple 1.05.
Micron is the surprise. It carries roughly 1.5% of the index against Nvidia’s 6.9%, but a 208% run this year gave it almost the same impact, until a sharp pullback after its latest results. Both companies are levered to the same data-centre buildout from different points in the supply chain.
Concentration is still near a record. The ten largest companies reached nearly 41% of index weight at the end of 2025, roughly double the level of a decade ago, which is worth knowing for anyone treating an S&P 500 tracker as a diversified core. Breadth has improved a little this year, with the equal-weighted index ahead of the cap-weighted version by about two percentage points in early July.
Earnings look better than they are
Second-quarter profits are on track to grow 50.4%, the fastest since 2021. Strip out Alphabet and Amazon and growth falls to 32.0%.
The gap is not operating profit. Alphabet’s quarter included roughly $98 billion of other income, mostly unrealised gains on equity securities. Amazon’s included a $53.4 billion gain tied largely to its stake in Anthropic. Those are mark-to-market gains on artificial intelligence companies, being used to justify the valuations of artificial intelligence companies.
Energy told a similar story from another direction. Sector earnings rose 147% because oil averaged $92.55 in the quarter against $63.68 a year earlier. That reverses if the strait reopens.
Bulls are not ignoring any of this. Analysts still expect 27.4% earnings growth in the third quarter, company guidance has been unusually positive, and Goldman Sachs recently raised its year-end target to 8,000. Much of the AI spending behind those forecasts is already financed through bonds sold last year, which makes near-term chip and memory revenue more visible than in a normal cycle.
To summarize
US GDP grew at a 1.5% annualised pace last quarter, payrolls are shrinking, inflation is above target, and the Federal Reserve under new chair Kevin Warsh has signalled its next move could be a hike rather than a cut. Against that backdrop, the market is paying 20 times earnings, a third of which came from investment gains at two companies, with a quarter of this year’s advance delivered by three stocks.
None of that is a timing signal. Valuation has almost no predictive power over a twelve-month horizon. What it does say is that the cushion is thin, and an ordinary disappointment could move prices more than usual.
