Summary
- Dow Jones Market Data’s analysis attributes 34% of FactSet-expected S&P 500 third-quarter profit growth to Nvidia and Micron and roughly two-thirds to 10 companies, establishing descriptive concentration rather than a quantified causal decomposition.
- The Wall Street Journal’s AI and energy-price accounts identify two transmission channels, but neither the reported contribution shares nor the accompanying mechanism narrative estimates what expected growth would have been under an intervention.
- FactSet’s 30% forecast remains prospective, and the available reporting supplies neither realized earnings nor the company-level forecasts and revisions needed to link expected growth to actual results.
- Holding other company forecasts fixed, revisions at Nvidia, Micron and other leading contributors would move expected index profit growth more than the headline 30% rate implies, although no dollar effect can be calculated from the article.
Expected third-quarter S&P 500 earnings growth is forecast to be concentrated in a small group: FactSet expects a 30% year-over-year increase, Dow Jones Market Data attributes 34% of the expected increment to Nvidia and Micron, and 10 companies account for roughly two-thirds. The available evidence establishes where expected growth is concentrated, not which developments produced it. The Wall Street Journal reports a broad AI-related interpretation and a conflict-to-energy-price-to-oil-and-gas channel, but the reporting does not include realized earnings, company-level forecasts and revisions, or controls needed to estimate either channel’s effect. The result is forecast risk concentrated in companies whose revisions can move the index expectation before the underlying operating gains are known.
Concentration of expected growth
FactSet expects companies in the S&P 500 to report third-quarter earnings 30% above the year-earlier level. Dow Jones Market Data’s analysis, reported by The Wall Street Journal, found that Nvidia and Micron together account for 34% of the expected profit growth. The same analysis found that 10 companies account for roughly two-thirds of the increase, with Broadcom and Meta Platforms among the members. The reported shares imply that Nvidia and Micron supply 34% of the increment, the other members of the 10-company group supply roughly another third, and companies outside the group supply roughly one-third.
Those figures describe the expected year-over-year earnings increment, not shares of total index earnings or market value. FactSet’s 30% figure is a year-over-year growth rate and therefore represents a different quantity from the 34% and roughly two-thirds contributor shares. The source does not disclose FactSet’s weighting or Dow Jones Market Data’s contributor calculation, so the three figures should not be treated as arithmetically interchangeable without their respective methodologies. Aggregate concentration also does not provide a representative estimate for a typical S&P 500 company.
The source does not disclose the individual ranking of Nvidia and Micron, so their combined contribution does not establish that either company individually led the contributors. Dow Jones Market Data’s 10-company finding also belongs to the same analysis that produced the Nvidia–Micron figure, rather than constituting a separate measurement.
The report identifies two contributor channels. It attributes the earnings-growth narrative to a handful of AI giants, although it does not disclose company-level forecast assumptions connecting AI-related conditions to Nvidia or Micron, and it does not assign a specific AI mechanism to Broadcom or Meta Platforms. The report also identifies oil-and-gas companies within the contributor group that benefited from higher energy prices caused by the conflict in Iran. The Wall Street Journal separately reported that the national average price for a gallon of regular gasoline had changed 44.2% since the start of the Iran war. That observation is consistent with the reported energy-price channel, but it neither quantifies the channel’s contribution to expected profit nor isolates the conflict’s effect.
The raw Federal Reserve Economic Data S&P 500 series put the index at 7,757.17 on October 26, 2026, the report’s publication date. That market level cannot validate the operating forecast: equity prices can reflect sentiment about expectations as well as information about operating performance.
How the expected-earnings path is structured
The available evidence can be represented as a causal structure in which solid arrows mark source-attributed or analytically necessary paths, dashed arrows mark hypothesized confounders, and the contributor shares are arithmetic outputs rather than causes.
K: Broad macro and sector conditions (hypothesized confounders; unmeasured) - -→ A - -→ forecast earnings
F: Analyst information, priors, industry expectations and estimate revisions ─→ each forecast earnings change
A: AI-related business conditions* ─→ Ê Nvidia ─┐ └───────────────────────────────────────→ Ê Micron ─┤ W: Iran conflict ─→ P: Energy prices ─→ Ê oil/gas ──┤ Ê Broadcom ─────────────────────────────────────────┤ (+)→ G: R: Expected earnings changes, all other firms ──────┘ Expected aggregate earnings increase
(Ê Nvidia + Ê Micron) / G ─→ D34: 34% share (Sum of Êi for the ten firms) / G ─→ D10: roughly two-thirds share
The two AI edges represent the report’s broad AI-driven interpretation, not company-level forecast assumptions disclosed in the article. The report establishes Nvidia and Micron as major contributors, but it does not show how AI-related conditions enter either company’s forecast. The AI pathway is therefore a candidate transmission route that the source does not establish. By contrast, the conflict-to-energy-price-to-oil-and-gas path is described directly in the source, although its effect on aggregate expected earnings remains unidentified.
Each company-level forecast earnings change contributes arithmetically to the expected aggregate increase, while the remaining companies provide parallel contributions rather than a common cause. Analyst information, priors, industry expectations and estimate revisions affect the forecasts as a measurement ancestor; they are not operating causes of company profits. The 34% and roughly two-thirds contribution figures are computed outputs and have no independent causal role.
K is a placeholder for macroeconomic conditions, industry cycles, commodity-supply decisions and company-specific pricing or product mix. The source supplies no measurements for those variables. No direct arrows run from AI-related conditions or the conflict to aggregate earnings because the report describes their transmission through company earnings. No link runs from AI conditions to Broadcom or Meta because the source establishes their membership among the 10 largest contributors without assigning them a mechanism. No relationship between AI conditions and energy prices, or between the conflict and AI conditions, is reported. Those omitted links mark the limits of the evidence rather than evidence that such relationships are absent.
The aggregate measure G can also become a collider on a path connecting AI conditions and the conflict when upstream conditions are unobserved. Conditioning on aggregate growth to infer a relationship between the two channels could therefore create a noncausal association. The contribution shares themselves do not require conditioning on aggregate growth.
Why the causal effects remain unidentified
The available evidence answers a rung-one observational question: which company-level earnings changes make up the expected aggregate increment? It does not answer intervention questions about what expected growth would have been under a different AI-related business condition or energy-price path.
For the AI channel, an estimate of (P(G \mid do(A=a))) is not identified from the source. The backdoor criterion is not met. A valid adjustment set would need to block every common-cause path into A, contain no descendants of A, and show adequate overlap within its adjustment strata. Company earnings variables are mediators and descendants of A and cannot serve as backdoor controls. No measured macroeconomic, industry or company-exposure variables capable of satisfying the criterion are provided, and positivity cannot be checked. Frontdoor identification is also unavailable because no single observed mediator is shown to intercept every directed path from A to G with an unconfounded relationship to aggregate earnings, and the source does not disclose a complete set of company-level forecast assumptions.
For the energy channel, the more meaningful intervention is (P(G \mid do(P=p))), holding other conditions fixed. Its backdoor set would require observed controls for factors that jointly affect energy prices and aggregate earnings; the source provides none, and the absence of confounding between energy prices and firm earnings is not established. Frontdoor identification would require a candidate mediator to intercept every directed path from energy prices to aggregate earnings and satisfy the relevant confounding conditions. Oil-and-gas earnings would need to intercept all such paths, but energy prices may also affect other firms through costs, demand or inflation. The source does not establish that oil-and-gas earnings meet that interception condition.
The reported 44.2% gasoline-price change is neither an instrument nor a control and does not measure the energy channel’s contribution to S&P 500 earnings. Converting the contribution shares into identified effects would require company-level earnings estimates, prior-year baselines, relevant pre-existing conditions and a research design capable of separating the proposed channels.
A counterfactual asking what the forecast would have been without the conflict would require a structural model specifying how energy prices, production, demand and company profits otherwise would have differed. The available reporting supplies none of those inputs, and the gasoline-price observation does not fill that evidentiary gap.
An informative design would compare forecast revisions and realized profits around a plausibly exogenous differential shock affecting Nvidia or Micron relative to less-affected peers, accounting for pre-shock revisions, contemporaneous common shocks and differential pre-trends. Such evidence could help distinguish a firm-specific operating response from a sector-wide AI-demand path or a forecast-prior path, but it would not conclusively separate those explanations without additional assumptions about differential exposure. Whether a company-specific shock could affect Nvidia or Micron differently without also changing sector-wide expectations remains unverified; resolving that question would require firm-level exposure data and supply-chain analysis. The source provides no firm-level demand, pricing, cost, forecast-error or earnings data, no correlation estimates and no controls for common macroeconomic shocks.
Forecast formation is not earnings realization
The reported figures belong to the forecast-formation side of the earnings process rather than the realization process.
Prior-year realized index earnings (baseline) │ ┌────┴─────────────┐ ▼ ▼ Expected index Realized index earnings change earnings change ▲ ▲ Expected current Realized current index earnings index earnings ▲ ▲ Expected company Realized company earnings forecasts earnings
- index aggregation + index aggregation ▲ ▲ Forecast inputs and Actual operating assumptions developments
Expected index earnings change → expected growth (+30% forecast) Realized index earnings change → realized growth Forecast error = realized current index earnings − expected current index earnings
Analyst priors, industry expectations and company-specific forecasts feed expected company profits and, through index aggregation, expected index profit growth. Actual operating developments instead feed realized company profits and realized index earnings. The forecast error is the difference between realized current index earnings and expected current index earnings; the source supplies neither side needed to calculate that outcome.
The Wall Street Journal’s statement that profit growth was being driven by a handful of AI giants is a reported interpretation, not a disclosed causal test. The source does not contain the company-level assumptions or realized data needed to distinguish operating performance from changes in analysts’ expectations.
Consequences of forecast concentration
Holding all other company forecasts fixed, a downward revision at either Nvidia or Micron lowers expected index profit growth, while an upward revision raises it. The source does not provide enough dollar-level information to calculate the magnitude. Revisions among this limited set of companies are therefore more consequential for the index expectation than the headline 30% growth rate alone suggests. Energy-company contributions are similarly conditional on the energy-price path.
Walden Siew’s Wall Street Journal report said much of Wall Street’s optimism about the stock market rested on a surge in corporate earnings, and it said FactSet’s forecast was supporting investor confidence that strong earnings could continue. That market response does not establish that the anticipated earnings will materialize, particularly when the supporting forecast is prospective and concentrated.
Audi follows a separate prospective path
Audi announced a $4 billion first U.S. factory in Macon expected to create around 4,500 jobs. Chief Executive Gernot Döllner said domestic production would give Audi greater flexibility to navigate trade barriers such as tariffs, which the company said clipped about 1.2 billion euros, or $1.3 billion, from profits last year. The announced chain runs from plant investment to future production, employment and potential tariff and sales effects.
Planned investment and projected employment are not yet measures of realized output or profit. Audi’s Macon project also does not explain the concentration of expected S&P 500 third-quarter profit growth.
Disclosure, verification and framing
No independent corroboration was supplied with the article for the 30% earnings-growth figure, the 44.2% gasoline-price change, the 7,757.17 index level, Audi’s investment and employment figures, or the tariff-profit figures. Until corroboration is available, those figures remain externally unverified, while the supplied Wall Street Journal report remains the controlling account for them. The underlying Dow Jones Market Data contributor table and calculation method and FactSet’s weighting are unavailable, and the individual ranking of Nvidia and Micron remains unresolved.
The report’s framing attributes the earnings-growth narrative to a handful of AI giants and presents Wall Street optimism as resting on an earnings surge. Those are reported interpretations layered on a prospective, concentrated forecast. The 34% figure represents Nvidia and Micron jointly rather than an individual leadership ranking; the 10-company result comes from the same Dow Jones Market Data analysis; and the AI and energy channels are not ranked against one another by magnitude.
The source establishes where expected growth is concentrated, but not what either reported channel caused or what the forecast would have been under different AI or energy-price conditions. Distinct contributor channels also do not by themselves establish diversification of index risk because the source supplies no correlation estimates or controls for common macroeconomic shocks. The contributor table, calculation method, company-level estimates and estimate revisions remain the material missing disclosures for distinguishing operating changes, forecast formation and common economic conditions.
Analytical techniques used in this piece
This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.
- Causal DAG
- Maps cause and effect as an explicit directed graph, exposing confounders and mediators (Pearl).