AI Infrastructure Continues to Drive Earnings; Enterprise AI Productivity Benefits Still Await Validation

Overview

This report reviews second-quarter 2026 earnings of U.S. companies, assesses the impact of enterprise AI adoption on earnings, spending, software demand, and labor markets, and updates the AI productivity beneficiary and high-/low-labor-cost stock baskets. The core conclusion is that AI capital expenditures have already provided a highly significant earnings lift to infrastructure-related companies, but direct earnings contributions from AI deployment have not yet generated a broad, statistically verifiable signal across most enterprises.

Core views

S&P 500 earnings growth is broad-based and strong, but AI infrastructure remains the most significant incremental driver. Enterprise AI spending is accelerating and is being financed primarily in the near term through reallocating existing software, technology, and labor budgets; consequently, its impact on aggregate market earnings is limited, though it could alter the distribution of earnings across industries and companies. Software fundamentals have not yet shown a widespread AI substitution shock, and AI’s impact on the labor market is visible but remains confined to specific occupations and industries.

Analysis framework

The report evaluates AI spending, productivity, and industry earnings impacts by combining S&P 500 quarterly results, company earnings-call commentary, the GS IT Spending Survey, the Ramp AI Index, estimated revenue of model providers, company disclosures on employees and compensation, and occupation-based measures of AI automation exposure.

Methodology notes

• Earnings AttributionS&P 500 EPS Growth Attribution

  Distinguishes the earnings contribution of AI infrastructure companies from that of the rest of the market.

  Classifies hyperscale cloud providers and companies benefiting from their capital expenditures as AI infrastructure, while excluding certain private-investment-related other income, to determine AI infrastructure’s contribution to index EPS growth.

• Enterprise Spending AnalysisAI Inference Cost Estimate

  Estimates the share of AI inference costs in S&P 500 revenue and IT budgets.

  Estimates inference costs using revenue of major AI model providers, the split between enterprise and consumer spending, and the S&P 500 share of spending; the estimate excludes other implementation costs such as skilled labor and technology infrastructure.

• Stock ScreeningAI Productivity Beneficiary Screening

  Identifies potential beneficiary companies based on labor-cost sensitivity and AI automation exposure.

  Screens the Russell 1000 for companies ranking in the top 50% of their industries both in payroll AI automation exposure and labor costs as a share of sales; it also requires that companies link AI with productivity or efficiency on second-quarter earnings calls, while excluding constituents of AI infrastructure and AI disruption-risk baskets.

Asset mapping & comparison

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

• AI Infrastructure Stocks

  Direct beneficiaries of AI capital expenditures

  Strengths

  High near-term visibility into capital expenditures and earnings growth, with a significant contribution to S&P 500 earnings growth in the second quarter of 2026.

  Weaknesses

  Valuations and growth expectations may already be elevated, and are sensitive to hyperscale cloud-provider capital expenditure cycles.

  Comparison

  Compared with AI productivity beneficiaries, earnings realization is currently clearer and therefore more favored by investors.

  Risks

  Capital expenditure slowdown, supply-chain or competitive changes, and downward earnings revisions.

• Software Stocks

  Potentially pressured by AI budget reallocation, while also serving as providers of enterprise AI deployment tools

  Strengths

  Surveys and disclosed fundamentals have not yet shown widespread substitution; ARR growth among Goldman Sachs-covered software companies has modestly accelerated recently.

  Weaknesses

  The market remains concerned that enterprises may build internal AI tools to replace some external software applications.

  Comparison

  Relative to AI infrastructure, the software sector benefits more indirectly from accelerating AI spending and faces greater near-term uncertainty.

  Risks

  Customers shifting to in-house development, budget migration toward models and infrastructure, and intensifying AI-native competition.

• AI Productivity Beneficiary Basket

  Potential beneficiaries of lower labor costs from AI automation

  Strengths

  Labor costs represent a relatively high share of revenue and AI automation exposure is high, creating substantial margin upside if productivity gains materialize.

  Weaknesses

  To date, there is insufficient evidence of meaningful AI-driven earnings improvement, and historical performance has been broadly in line with the S&P 500.

  Comparison

  Lacks the near-term, quantifiable earnings catalysts of AI infrastructure stocks.

  Risks

  Automation exposure may also indicate business disruption risk rather than simply efficiency gains.

• High Labor Cost Stock Basket (GSTHHLAB)

  More sensitive to AI productivity improvements and labor-budget reallocation

  Strengths

  Offers greater potential cost-savings opportunities if AI materially improves efficiency.

  Weaknesses

  Risks from labor substitution, implementation costs, and business disruption are also greater.

  Comparison

  Median labor costs account for 31% of revenue, compared with 5% for the low-labor-cost basket and 14% for the S&P 500 median.

  Risks

  AI benefits materializing more slowly than expected, labor-market adjustment frictions, and cost savings failing to convert into profits.

Key data

• S&P 500 second-quarter 2026 year-over-year EPS growth31%After excluding certain private-investment-related “other income.”
• Year-over-year EPS growth of AI infrastructure companies54%Hyperscale cloud providers and companies benefiting from their capital expenditures.
• AI infrastructure contribution to S&P 500 EPS growthApproximately 50%Second quarter of 2026.
• Median EPS growth among S&P 500 companies14%Second quarter of 2026.
• Share of S&P 500 companies quantifying specific AI productivity impacts11%For example, coding or customer support.
• Share of S&P 500 companies quantifying AI’s earnings impact2%Similar to the share in the first quarter of 2026.
• AI inference costs as a share of S&P 500 revenueLess than 0.5%Estimated value.
• Monthly AI spending per employeeMedian increased from $5 at the start of the year to $12 in JulyRamp AI Index; spending distribution is highly dispersed.
• Sources of AI spendingApproximately two-thirds comes from reallocating existing budgetsIncluding 18% from software budgets and 11% from labor budgets.
• S&P 500 labor costsApproximately 12% of revenue, totaling about $2.1 trillionBased on company disclosures and 2025 estimates.

Impact & implications

Investors should continue to focus on AI infrastructure value chains with high earnings leverage and order visibility, but should not pursue productivity beneficiaries solely on the AI narrative. Accelerating enterprise AI spending is an important validation point for the next several quarters: if productivity improvements translate into margin expansion and EPS upgrades, companies with high labor costs and high automation exposure may benefit; if AI budgets primarily crowd out existing software and labor spending, the reallocation of revenue and profits within industries will be more prominent than expansion in aggregate market earnings.

Risks

• Productivity and earnings benefits from AI adoption may be lower than expected or take longer to materialize.
• A slowdown in AI infrastructure capital expenditure growth could weaken the current primary earnings driver.
• Enterprises shifting budgets from software or labor toward AI may reallocate revenue and profits within industries.
• High AI automation exposure may generate efficiency gains but may also increase business disruption risk.
• AI spending, inference costs, and automation exposure include estimates and survey data, creating methodology and sample biases.

What to watch

• Whether the proportion of companies quantifying AI’s impact on revenue, costs, margins, and EPS rises in subsequent earnings reports.
• The sustained growth rate of AI inference costs and AI spending per employee.
• Whether AI funding shifts from reallocating existing budgets to incremental budgets.
• Whether software companies’ ARR, renewals, and customer in-house-tool trends show signs of AI substitution.
• Changes in margins, headcount, and operating efficiency at companies with high labor costs and high automation exposure.
• Hyperscale cloud-provider capital expenditures and the orders, revenue, and earnings guidance of AI infrastructure companies.