AI and Momentum Drive U.S. Equities in a “One Big Trade” Setup, but Narrowing Breadth Raises Reversal Risk

Overview

This report discusses the concentrated rally in U.S. equities recently driven jointly by the AI trade and the Momentum factor. Goldman Sachs points out that the market increasingly looks like a single big trade: the S&P 500 is up 10% year to date, but TMT has contributed the vast majority of returns, with NVDA alone accounting for about 20% of the index's total YTD return. At the same time, narrowing breadth, the rapid rise in the Momentum factor, and hedge fund Momentum exposure near a five-year high mean investors need to assess AI fundamentals, the macro backdrop, and factor crowding simultaneously.

Core views

The key view is that, first, the AI trade has become a key driver of current U.S. equity performance and the composition of the Momentum factor; second, historically, Momentum rallies of more than 20% over three months often continue for another month, but tend to fade over the following two to three months; third, when such Momentum rallies occur near market highs, subsequent three-to-six-month S&P 500 returns are usually weaker than average; fourth, the difference between this cycle and the late 1990s or 2021 is that the recent rally has at least been partly supported by upward earnings revisions; fifth, investors should continue to focus on earnings growth and positive earnings estimate revisions while using low-Momentum stocks, consumer staples, health care, real estate, or low-AI-sensitivity portfolios to reduce reliance on a single AI trade.

Analysis framework

The report combines historical factor backtests, decomposition of sector and industry return contributions, earnings estimate revision analysis, correlation analysis, and stock screening. The Momentum factor is measured by GSMEFMOM, an equal-weight long-short portfolio constructed by ranking S&P 500 constituents into high and low buckets based on trailing 12-month returns; AI exposure is measured using the GS AI long-short baskets, GSTMTAIP and GSTMTAIR; economic growth exposure is proxied by the performance of an equal-weight cyclical basket relative to a defensive basket. Stock screening emphasizes positive EPS revisions, low sensitivity to the AI trade, and low sensitivity to U.S. economic growth pricing.

Methodology notes

• Factor researchHistorical analogs for the Momentum factor

  Compare the current sharp three-month rise in Momentum with similar events since 1980

  The report examines 11 separate events since 1980 in which three-month Momentum rose by 20% or more, finding that Momentum usually rises further in the following month and then weakens over the next two to three months; when the event occurs near an S&P 500 high, subsequent market returns tend to be weaker.

• Theme basket analysisGS AI basket pair

  Use GSTMTAIP and GSTMTAIR to measure AI trade exposure

  The report compares the long and short compositions of the current Momentum factor with AI beneficiary/loser baskets, noting that long sectors such as semiconductors, technology hardware, capital goods, and energy are tied to AI infrastructure buildout, while short sectors such as software and business services are affected by concerns about AI disruption.

• Earnings revision analysisEPS revision factor

  Compare the relative performance of stocks with the strongest and weakest earnings estimate revisions

  Goldman Sachs uses an industry-neutral long-short EPS revision factor to measure the return difference between stocks in the S&P 500 with the strongest and weakest two-year-forward EPS consensus revisions; the factor has generated an 8% long-short return year to date.

• Stock screeningInsensitive Portfolio

  Screen for stocks with low sensitivity to AI and U.S. economic growth pricing, plus positive EPS revisions

  This portfolio screens Russell 1000 stocks with at least 5% EPS growth in 2026 and 2027, recent upward EPS estimate revisions, and the lowest stock-price sensitivity to AI trade and U.S. economic growth pricing, aiming to find opportunities that do not depend entirely on the AI view.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• S&P 500

  Core research object and market benchmark

  Strengths

  The index is up 10% year to date, EPS estimates have been revised up 8% in both 2026 and 2027, and recent EPS revision breadth across all sectors has turned positive.

  Weaknesses

  Returns are highly concentrated, with TMT contributing 85%; excluding TMT, the index is up only 3%, and market breadth has declined.

  Comparison

  Compared with 1999-2000, breadth has not yet narrowed to that degree; similar to May 2023, narrowing breadth may bring volatility but does not necessarily end the bull market.

  Risks

  If AI expectations or the macro backdrop deteriorate, markets near highs have historically faced weaker returns in the following months.

• Momentum factor (GSMEFMOM)

  The main factor vehicle for the current AI trade

  Strengths

  Up 25% over the past three months; history suggests that similar rallies can still continue in the short term.

  Weaknesses

  The current gain already exceeds the historical average continuation magnitude, and hedge fund Momentum exposure is near a five-year high.

  Comparison

  Since 1980, after 11 similar three-month surges, Momentum has on average risen for another month before typically fading over the following two to three months.

  Risks

  A deterioration in AI investment expectations, higher macro volatility, or laggard stocks catching up could all trigger a Momentum reversal.

• AI-related stocks and the GS AI basket pair (GSTMTAIP vs. GSTMTAIR)

  The thematic core driving the market and Momentum

  Strengths

  Semiconductors, technology hardware, and capital goods tied to AI infrastructure benefit from rising AI capital spending expectations.

  Weaknesses

  Many stocks are already tightly linked to the AI trade, reducing diversification benefits.

  Comparison

  The recent co-movement between the Momentum factor and the GS AI long-short basket pair suggests that the two are highly coupled in the market.

  Risks

  A downward revision to AI capex expectations, premature discounting of long-term growth, or a reversal in fund flows could magnify drawdowns.

• Consumer Staples

  A defensive sector with low sensitivity to AI and Momentum

  Strengths

  The report identifies this as one of the sectors with the lowest exposure to AI and Momentum recently, making it useful for improving portfolio diversification.

  Weaknesses

  Its defensive nature may lag growth and cyclical sectors when risk appetite is strong.

  Comparison

  Energy and Consumer Discretionary also show relatively low AI exposure, but both are more closely tied to the geopolitical outlook.

  Risks

  If the macro outlook improves materially, relative performance of defensive sectors may come under pressure.

• Low-Momentum stocks

  A Momentum reversal hedge

  Strengths

  Historically, when Momentum sharply reverses, prior laggards often outperform on a relative basis and frequently rise in absolute terms as well.

  Weaknesses

  If Momentum remains strong, low-Momentum stocks may continue to lag in the short term.

  Comparison

  The report identifies 25 stocks currently on the short side of the Momentum factor with positive recent EPS revisions for 2026 and 2027.

  Risks

  If earnings revisions fail to materialize or the market continues to prefer high-Momentum names, the protection may arrive late.

• Insensitive Portfolio

  A stock basket that does not rely on the AI view

  Strengths

  The median stock in the portfolio has AI and U.S. economic growth pricing explain only 13% of its past-year daily return variation, below the Russell 1000 median of 30%, while also having positive EPS revisions.

  Weaknesses

  The portfolio is not risk-free; it is only relatively less sensitive, and individual stocks still depend on sector fundamentals and valuations.

  Comparison

  The table shows a median YTD return of 19% for the portfolio, above the Russell 1000 median of 1%, with the same 12-month forward P/E of 17x.

  Risks

  The screen is based on historical sensitivities, so the low-sensitivity characteristics may break down if market drivers change.

Key data

• S&P 500 YTD return10%As of the Thursday referenced in the report, the 2026 YTD return was 10%.
• TMT contribution to S&P 500 YTD return85%TMT includes Information Technology, Communication Services, AMZN, and TSLA.
• S&P 500 YTD return excluding TMT3%Shows that market gains are highly concentrated.
• NVDA contribution to total S&P 500 YTD return20%NVDA accounts for about 9% of the S&P 500's market-cap weight.
• S&P 500 one-month new high count14 timesNew highs have occurred while the share of constituents above the 200-day moving average has declined.
• Median S&P 500 constituent distance from 52-week high-13%Reflects narrowing market breadth.
• Momentum factor 3-month return25%One of the strongest three-month rallies in history.
• Number of historical comparable Momentum rallies11 timesIndependent events since 1980 in which three-month returns were 20% or higher.
• Average one-month continuation return after similar rallies6%The report notes that the current gain already exceeds this historical average continuation magnitude.
• YTD revision in S&P 500 2026 EPS consensus+8%Bottom-up consensus EPS expectations have been revised higher.
• YTD revision in S&P 500 2027 EPS consensus+8%Mainly driven by expectations for AI capital spending and energy prices.
• 2027 EPS outlook excluding AI infrastructure and energyessentially flatIndicates that earnings revisions are concentrated in a few themes, but the breadth of revisions has recently improved.
• YTD return of the EPS revision factor8%The industry-neutral long-short factor compares the strongest and weakest EPS revisions.
• Share of return sensitivity explained for the Insensitive Portfolio13%AI and U.S. economic growth pricing explain 13% of the variation in the median stock's daily returns over the past year, below the Russell 1000 median of 30%.

Impact & implications

The implication for asset allocation is that U.S. equity gains are still supported by AI capital spending, earnings estimate revisions, and fund flows, but concentration, crowding, and deteriorating breadth make portfolios more vulnerable to changes in AI expectations, macro volatility, or a drop in risk appetite. If the AI investment outlook deteriorates, Momentum may suffer a catch-down drawdown; if the macro outlook improves materially, lagging stocks may catch up, likewise causing a Momentum reversal. Therefore, portfolio construction should avoid betting solely on AI and Momentum as the main themes, and should prioritize stocks or sectors with supported revisions, reasonable valuations, and lower thematic sensitivity.

Risks

• A downward revision to AI capital spending or AI earnings prospects could trigger synchronized pullbacks in Momentum and related stocks.
• Continued narrowing of market breadth could make the index increasingly vulnerable to a few large-cap technology names.
• Hedge fund Momentum exposure is near a five-year high, so crowded positioning could amplify volatility.
• If the macro backdrop deteriorates and lifts equity or bond volatility, investors may reduce risk exposure.
• If the macro backdrop improves materially, laggard stocks catching up could also reverse the Momentum long-short structure.
• Some sectors' price performance has already moved ahead of or diverged from recent EPS revisions, suggesting valuations may embed more optimistic long-term growth assumptions.
• The report's baskets and screening portfolios are affected by liquidity, borrow constraints, and market conditions, and may not be fully tradable as shown by historical performance.

What to watch

• Whether the share of S&P 500 constituents above the 200-day moving average and overall market breadth continue to decline.
• Whether NVDA, semiconductors, technology hardware, and AI infrastructure-related capital goods contribute even more heavily to index returns.
• Whether S&P 500 EPS consensus estimates for 2026 and 2027 continue to be revised upward, especially the trend excluding AI infrastructure and energy.
• Whether the Momentum factor GSMEFMOM shows signs of peaking or reversing over the next one to three months.
• The relative performance of the GS AI basket pair GSTMTAIP versus GSTMTAIR, and whether its correlation with the Momentum factor declines.
• Whether hedge fund net Momentum exposure and gross leverage fall from five-year highs.
• Whether low-AI-sensitivity sectors such as Consumer Staples, Health Care, and Real Estate can provide defensive diversification.
• Whether low-Momentum stocks with positive EPS revisions can generate absolute returns during market volatility or a Momentum pullback.