Goldman Sachs sees S&P 500 earnings slowing from unusually strong levels, not collapsing into an earnings bubble

Overview

Goldman Sachs examines whether exceptional recent S&P 500 profit growth constitutes an earnings bubble. It concludes that several temporary earnings supports will fade, causing a deceleration, but its base case remains continued growth supported by GDP, moderating energy pressure, and an emerging contribution from AI productivity.

Core views

S&P 500 earnings have risen far above their recent historical pattern: EPS grew 51% year-on-year in Q2 2026 and 26% across the past four quarters, versus an average four-quarter growth rate of 7% over the past 30 years. Earnings have also exceeded the level suggested by their historical relationship with US GDP growth. Goldman Sachs agrees that companies are currently “over-earning,” but distinguishes that condition from an earnings collapse. The firm expects growth to decelerate as unusual supports fade rather than reverse sharply. The first major support is the AI investment boom. US hyperscalers are projected to spend $800 billion on capex in 2026, up 94% from 2025. This spending benefits the AI-infrastructure complex—semiconductors, technology hardware, industrials, and utilities—which accounts for roughly half of consensus S&P 500 earnings growth this year. Goldman Sachs estimates AI capex contributes 11 percentage points to 2026 S&P 500 EPS growth. Capex is still expected to rise, with the firm's base case calling for $1.232 trillion in 2027 and $1.379 trillion in 2028, but its growth rate should slow while accumulated depreciation expense increases. Depreciation is estimated to subtract 5 percentage points from 2027 earnings growth, offsetting nearly half of the capex benefit, and to offset the remaining capex uplift by 2028. The report stresses that this outlook is highly sensitive to hyperscaler spending. In its base case, capex contributes 7 percentage points to S&P 500 EPS growth in 2027 and subtracts 1 percentage point in 2028. A $250 billion capex surprise in either direction in 2027 would move S&P 500 EPS growth by about 6 percentage points in the same direction. In a downside case where capex falls roughly 30% to $570 billion, the decline would offset earnings growth elsewhere in the index and reduce AI-infrastructure-complex earnings by about 40%, although those profits would remain about 25% above 2025 levels. The report also notes that fully expensing capex would make consensus 2027 S&P 500 EPS 22% lower than reported under current accounting rules, illustrating the unusually large current difference between reported and hypothetical fully expensed earnings. The second temporary driver is semiconductor profitability. Tight supply and strong demand have lifted chip prices and gross margins, particularly in memory, where S&P 500 memory-company gross margins are roughly 80%, more than twice their historical average. Gross-margin expansion accounts for about one-quarter of semiconductor earnings growth in 2026. Goldman Sachs expects memory supply-demand conditions to remain tight through 2027, with incremental supply unlikely to affect margins before 2028, but expects the pace of additional margin expansion to slow materially in 2027. Each percentage-point movement in S&P 500 semiconductor gross margins next year is estimated to shift aggregate S&P 500 EPS growth by about 1 percentage point. A more adverse outcome in which semiconductor gross margins fall from roughly 70% to their 15-year average of 55% would reduce S&P 500 earnings by roughly 10%. The third source of elevated earnings is GAAP “other income” from mega-cap technology companies' private-equity holdings. Unrealized gains totaled roughly $150 billion in Q2 2026 and lifted S&P 500 EPS by 12%. Goldman Sachs expects additional gains in the second half of 2026, but a much smaller contribution in 2027. If this income disappeared entirely in 2027, it would create an 8-percentage-point drag on S&P 500 earnings growth relative to 2026. Conversely, excluding this income from the inflated 2026 base, the firm's 2027 forecast implies 18% EPS growth. Goldman Sachs forecasts adjusted S&P 500 EPS of $415 in 2027 and $460 in 2028, both up 11% year-on-year. Its 2027 forecast is close to bottom-up consensus of $419 and above the median top-down strategist estimate of $403. The 2028 forecast is about 5% below bottom-up consensus, reflecting the firm's expectation that consensus estimates will resume their typical historical pattern of downward revisions. Outside the AI-capex beneficiaries, Goldman Sachs expects sales growth near nominal GDP, modest margin expansion, and AI productivity gains to produce 9% earnings growth in both 2027 and 2028; this is about 4–5% below consensus for that group. Macroeconomic assumptions support the base case. Goldman Sachs economists expect US real GDP growth of 2.1% in 2027 and 2.3% in 2028; its top-down model identifies economic growth as the main variable, explaining more than half of historical variation in S&P 500 EPS growth. Brent crude is expected to fall to $80 per barrel by end-2027, modestly weighing on Energy profits but supporting most other sectors. As AI-investment benefits fade, the report increasingly relies on AI productivity, embedding a 1-percentage-point EPS-growth boost in 2027 and a 2-percentage-point boost in 2028. Enterprise adoption is increasing but remains early-stage and concentrated, so its earnings effect should become clearer over coming quarters. On valuation, the report finds no near-term earnings-multiple bubble: the S&P 500 forward P/E has declined from 23x a year earlier to 19x, matching its 10-year average. However, valuation based on trend earnings remains elevated, and the cyclically adjusted P/E is among its highest historical readings, below the 1999–2000 peak but above 2021. Current profitability is also unusually high: S&P 500 ROE is a record 24%, while the current macro backdrop and a 19x forward P/E are more consistent with roughly 22% ROE. Goldman Sachs interprets this gap as market skepticism about whether today's profitability can persist. The firm expects continued equity upside to come from earnings rather than multiple expansion. Its 3-, 6-, and 12-month S&P 500 targets are 8,000, 8,300, and 8,700, implying returns of 5%, 9%, and 14%, respectively, while maintaining a forward P/E close to 19x. Modestly lower interest rates could support valuations, but uncertainty around the scale and distribution of long-term AI profits is expected to restrain a dramatic rerating.

Analysis framework

Goldman Sachs first compares recent S&P 500 EPS growth with long-run earnings and GDP relationships, then identifies three sources of above-trend earnings: AI capex, semiconductor margins, and private-investment gains. It applies scenario and sensitivity analysis to capex and margins, combines bottom-up and top-down EPS forecasts, and assesses valuation through forward and cyclically adjusted P/E multiples as well as the historical relationship between ROE and P/E.

Methodology notes

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI-capex transmission through the infrastructure complex

  The report traces hyperscaler capital spending through semiconductor, hardware, industrial, and utility earnings, and then to aggregate S&P 500 EPS.

• Industry AnalysisVolume-price decomposition

  Semiconductor profit growth decomposition

  The report separates chip earnings growth into revenue and gross-margin effects, highlighting that margin expansion accounts for about one-quarter of 2026 semiconductor earnings growth.

• Valuation methodsP/E and PEG Valuation

  Forward and cyclically adjusted P/E analysis

  Goldman Sachs compares the current 19x forward P/E with its 10-year average and contrasts it with valuations based on longer-run earnings to assess sustainability risk.

• Corporate Fundamentals and FinanceEarnings Quality Analysis

  Adjustment for capex accounting and other income

  The report tests reported EPS against hypothetical full capex expensing and isolates temporary private-investment gains to judge the durability of earnings.

Asset mapping & comparison

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

• S&P 500

  Primary market subject; aggregate earnings are supported by AI-related investment but face normalization of temporary profit drivers.

  Strengths

  Projected EPS growth of 11% in both 2027 and 2028, supported by GDP growth, lower energy pressure, and AI productivity gains.

  Weaknesses

  Current earnings are elevated relative to historical trend and GDP-linked patterns.

  Comparison

  The 2027 Goldman Sachs EPS forecast of $415 is near bottom-up consensus of $419 and above the median top-down estimate of $403; its 2028 forecast is about 5% below bottom-up consensus.

  Risks

  AI capex slowdown, semiconductor-margin compression, fading other income, input-cost pressure, and consumer effects from equity-market volatility.

• S&P 500 semiconductor sector

  AI-infrastructure beneficiary and major contributor to aggregate earnings growth.

  Strengths

  Tight memory supply-demand conditions are expected through 2027.

  Weaknesses

  The rate of gross-margin expansion is expected to slow in 2027.

  Comparison

  Memory-company gross margins are roughly 80%, more than double their historical average.

  Risks

  Slower AI infrastructure spending, added supply, or technological shifts could lower chip prices and margins.

Key data

• S&P 500 EPS growth51% year-on-year in Q2 2026; 26% over the past four quartersThe 30-year average four-quarter EPS growth rate is 7%.
• S&P 500 forward P/E19xDown from 23x a year earlier and equal to the 10-year average.
• Hyperscaler capex$800 billion in 2026Up 94% versus 2025; Goldman Sachs forecasts $1.232 trillion in 2027 and $1.379 trillion in 2028 in its base case.
• AI capex contribution to S&P 500 EPS growth11 percentage points in 2026; 7 percentage points in 2027; negative 1 percentage point in 2028Base-case contribution as capex growth slows and depreciation rises.
• Private-investment other incomeOver $150 billion in Q2 2026Mega-cap technology unrealized gains lifted S&P 500 EPS by 12%.
• S&P 500 adjusted EPS forecast$415 in 2027 and $460 in 2028Both years imply 11% annual growth.
• US real GDP growth forecast2.1% in 2027 and 2.3% in 2028A key macro input to Goldman Sachs' top-down earnings model.
• S&P 500 ROE24%A record high; the current 19x forward P/E is more consistent with roughly 22% ROE under the current macro backdrop.

Impact & implications

The report expects the earnings cycle to normalize rather than break: fading AI-capex momentum, slower semiconductor-margin expansion, and less other income should reduce headline growth, while GDP growth and AI productivity support continued gains. It argues that the market already reflects some skepticism about sustained profitability, leaving earnings delivery—not a substantial valuation rerating—as the central driver of projected index returns.

Risks

• Energy prices and interest rates are near-term macro risks to the earnings outlook.
• Persistent input-cost pressure could weaken earnings growth.
• Equity-market volatility could pressure consumer spending.
• A sharper-than-expected slowdown in AI infrastructure spending would reduce AI-complex profits and S&P 500 earnings growth.
• A decline in semiconductor prices and gross margins would materially weaken aggregate EPS growth.
• The scale, timing, and distribution of long-term AI productivity gains remain uncertain.

What to watch

• Hyperscaler capex plans and surprises versus expectations, as a $250 billion 2027 deviation would shift S&P 500 EPS growth by about 6 percentage points.
• Growth in hyperscaler depreciation expense as it increasingly offsets capex-driven earnings benefits.
• Semiconductor supply-demand conditions, chip pricing, and gross-margin trends.
• The size of mega-cap technology companies' other-income gains in the second half of 2026 and their expected decline in 2027.
• Evidence that enterprise AI adoption is producing broader productivity gains.
• US GDP growth, Brent crude oil, interest rates, input costs, and consumer resilience.