The S&P 500's 50.4% Earnings Growth Is Mostly Two Companies. FactSet's Own Note Says So.
The number moving through market commentary this month is 50.4%. That is the blended year-over-year earnings growth rate for the S&P 500 in the second quarter of 2026. FactSet spells out what it would mean if it holds: "If 50.4% is the actual growth rate for the quarter, it will mark the highest earnings growth rate reported by the index since Q2 2021 (91.6%)." It is a real figure, published by a serious data provider, and it is being repeated accurately. The problem is not the number. The problem is what people take it to mean.
The provider that publishes it, FactSet, also publishes the qualifier. In the same document sits this sentence: "Excluding Alphabet and Amazon.com, the blended earnings growth rate for the S&P 500 for Q2 2026 would fall to 32.0% from 50.4%." That is 18.4 percentage points of the index's earnings growth — more than a third of it — attributable to two companies out of five hundred. FactSet did not bury this. It sits in the report's revisions section, under a heading that says it outright: "Alphabet and Amazon.com Have Led Increase in Earnings since June 30."
A note on which edition this is, because the figures move weekly. As of Saturday morning, August 15, the current FactSet Earnings Insight is the August 7, 2026 edition. FactSet's canonical landing page for the report, factset.com/earningsinsight, 302-redirects to the file EarningsInsight_080726.pdf — the August 7 document. No August 14 edition was published to that address or to FactSet's Insight site as of this writing. Every FactSet figure below is the August 7 vintage, and readers checking next week should expect all of them to have shifted.
FactSet is explicit about the attribution, and the exact wording matters. The report states: "Combined, these two companies account for about 71% of the increase in dollar-level earnings for the index since June 30." The accompanying web summary adds: "Since June 30, the positive EPS surprises reported by Alphabet and Amazon.com have been the largest contributors to the increase in the overall earnings growth rate for the index over this period." That is a claim about the *increase* since quarter-end, not about the level of index earnings — a distinction worth holding onto.
Before going further it is worth explaining what a blended rate is, because the mechanics are the whole reason the statistic behaves the way it does. FactSet defines the blend as a figure that "combines actual results for companies that have reported and estimated results for companies that have yet to report." With 88% of the index reported as of August 7, most of the 50.4% is now history and the rest is still forecast — though 88% is a count of companies, not a share of earnings dollars, and the blend is dollar-weighted, so the two do not map onto each other. On June 30, before anyone had reported, the figure was 100% forecast. The blend is not a projection that gets refined; it is a running tally in which estimates are progressively swapped out for actuals.
That is why the number climbs during a season. Companies as a class beat the estimates set for them, so each swap of estimate-for-actual tends to nudge the blend up. On June 30 the rate stood at 23.1%, per the August 7 edition. A week before that edition it was 47.4%. It is now 50.4%. From quarter-end to today, the blended rate has improved by 27.3 percentage points.
How much improvement is normal? FactSet has published its own answer, though not in the August 7 document — in its July 10 quarter preview. There it wrote that "from the end of the quarter through the end of the earnings season, the earnings growth rate has increased by 6.2 percentage points on average (over the past ten years) due to the number and magnitude of positive earnings surprises." The five-year figure is 6.4 percentage points. Anyone citing 6.2 as FactSet's decade norm should cite it to the July preview; the August 7 update does not restate it.
That preview also produced a forecast, and the arithmetic behind it is instructive. FactSet wrote: "If this average increase is applied to the estimated earnings growth rate at the end of Q2 (June 30) of 23.2%, the actual earnings growth rate for the quarter would be 29.4% (23.2% + 6.2% = 29.4%)." The headline conclusion was that "the index will likely report earnings growth above 29% for Q2." One small inconsistency to flag honestly: the July 10 preview gives the June 30 starting point as 23.2%, while the August 7 update gives it as 23.1%. The gap is a tenth of a point and changes nothing material, but it is FactSet's own documents disagreeing, and readers deserve to know which one they are reading.
Set the two against each other. A typical season improves the rate by about 6.2 points. This season has improved it by 27.3 — roughly 4.4 times the decade norm. That is not a slightly strong quarter. It is a distributional outlier, and an outlier is a thing to explain rather than repeat.
The explanation is in the surprise magnitude, not the surprise rate. On the rate, this quarter is good but not strange: 86% of reporting companies beat EPS estimates, against a five-year average of 78% and a ten-year average of 76%. Seventy-six percent beat on revenue. On magnitude, it is a different universe. In aggregate, companies are reporting earnings 29.2% above estimates, against a five-year average of 7.0% and a ten-year average of 7.4%. Revenues came in 3.2% above estimates, against averages of 1.9% over five years and 1.6% over ten. Slightly more companies beat than usual; a couple of them beat by an amount with no recent precedent.
The filings show what produced it. Alphabet's second-quarter release reports revenues of $119.8 billion against $96.4 billion a year earlier, operating income of $40.8 billion against $31.3 billion, and diluted EPS of $9.11 against $2.31. The gap between the operating line and the EPS line is other income, net, which came to $98.0 billion versus $2.7 billion a year ago. Alphabet's own disclosure: "For Q2 2026, the net effect of the gain on equity securities of $99.0 billion increased the provision for income tax, net income, and diluted net income per common share by $21.9 billion, $77.1 billion, and $6.26, respectively." Subtracting the company's own stated per-share effect from its own reported diluted EPS leaves $2.85.
Amazon's release follows the same shape. Net sales of $200.6 billion, operating income of $27.5 billion, net income of $62.6 billion against $18.2 billion a year earlier, and diluted EPS of $5.75 against $1.68. The company states: "Second quarter 2026 net income includes non-operating pre-tax other income of $53.4 billion, primarily from our investments in Anthropic." That single non-operating item is roughly 1.9 times the quarter's entire operating income.
Both items are marks on privately held stakes, not cash from selling goods or services. Both companies had genuinely strong operating quarters underneath — Alphabet's operating income rose about 30% year over year, Amazon's revenue crossed $200 billion — and neither of those facts is diminished by noting that the reported bottom lines were dominated by something else entirely. What it means for the index statistic is that a large share of the S&P 500's 50.4% growth rate reflects revaluations of two stakes in private artificial-intelligence companies, and that those revaluations can move in the other direction in a future quarter with equal force.
Be careful with the revenue line, because the exclusion FactSet publishes there is a different one. Blended Q2 revenue growth is 15.0%, up from 12.2% at June 30. But FactSet does not publish an ex-Alphabet-and-Amazon revenue figure. What it publishes is a sector cut: "If these two sectors were excluded, the blended revenue growth rate for the S&P 500 for Q2 would fall to 9.7% from 15.0%" — the two sectors being Information Technology and Energy. Anyone pairing the 32.0% ex-two-company earnings figure with a 9.7% revenue figure is fusing two different exclusions. They are not comparable.
The same caution applies to margins. FactSet puts the blended Q2 net profit margin at 16.9%, against 12.9% a year ago and a five-year average of 12.4%. A net margin is net income over revenue, so the investment gains sit in the numerator of that ratio too. Breadth elsewhere is genuinely broad — ten of eleven sectors are reporting year-over-year earnings growth, led by Energy at 147.0% — but breadth of direction is not the same as breadth of magnitude, and the index-level number is a dollar-weighted sum, not a count.
A second provider counts it differently and arrives somewhere else, which is the most useful check available. Zacks, in a scorecard by Sheraz Mian dated August 12, reports: "For the 451 S&P 500 companies reporting Q2 results (representing 90.2% of total index membership), aggregate earnings grew +41.6% year-over-year on +14.7% higher revenues." That is an actual-only figure for companies that have reported — not a blended index-wide rate — which is one reason it sits below FactSet's 50.4%. Zacks then runs its own exclusion, and notably picks a different pair: "Excluding Micron and Alphabet, Q2 earnings for the remaining 449 reporting index members rose +21.5% (compared to +41.6% unadjusted) on +13.6% higher revenue (compared to +14.7% unadjusted)." Two providers, two methods, two different sets of two companies — and both find that removing two names cuts the growth rate roughly in half.
Here is what is not known, and it is a fair amount. FactSet's August 7 document does not spell out which line items sit inside its "actual" EPS for any individual company, so the link between the investment gains disclosed in Alphabet's and Amazon's filings and FactSet's measured EPS surprises is an inference from magnitude and from FactSet's own attribution, not something FactSet states line by line. Twelve percent of the index has not reported, so the 50.4% and the 32.0% will both move. LSEG publishes a competing S&P 500 earnings dashboard, including an August 14 edition — more current than FactSet's — but its pages returned repeated redirect errors on every attempt to read them, so no LSEG figure appears in this piece. A widely shared social-media post citing a 24.7% blended rate conflicts with FactSet's published documents and is not used here. And nothing in this article tells you whether stripping out mark-to-market gains is the right way to think about a technology company's earnings; reasonable analysts disagree, and the disagreement is about accounting philosophy, not arithmetic.
Looking forward, FactSet's August 7 edition reports that 75 S&P 500 companies have issued Q3 2026 EPS guidance, of which 25 have issued negative guidance and 50 have issued positive guidance. The estimated Q3 growth rate is 27.4% and the CY2026 estimate is 30.0% — figures that are, at this stage, entirely estimate and no actual, the exact opposite of the blend's composition today. The forward 12-month P/E ratio stands at 20.0, above its five-year average of 19.9 and above its ten-year average of 19.0. Worth noting: a forward P/E uses forward estimates in the denominator, so it is not directly affected by the trailing gains discussed above — but the estimates themselves are set by analysts watching the same reported numbers.
Markets closed Friday, August 14, with the S&P 500 at 7,785.76, down 0.17%; the Nasdaq Composite at 26,729.16, down 0.28%; the Dow Jones Industrial Average at 53,732.41, down 0.20%; and the Russell 2000 at 3,068.42, up 0.51% for a record close. None of that is a verdict on the earnings data. The point of this piece is narrower and, I would argue, more durable: 50.4% is an accurate description of a dollar-weighted sum that two companies happen to dominate this quarter. If you want a number that describes the experience of the median S&P 500 company, 50.4% is not it — and the organization that publishes 50.4% is the same one telling you so.
