As AI adoption heats up, the stronger signal for corporate credit markets is on the supply side

Overview

This report assesses the impact of AI adoption on U.S. and European investment-grade non-financial corporates from a corporate credit research perspective. Goldman Sachs screened earnings calls from 500 U.S. and 175 European investment-grade non-financial corporates, and further reviewed the latest disclosures of 42 large U.S. issuers and 46 large European issuers. The conclusion is that AI mentions are rising rapidly and converging across regions, but corporate disclosures still mainly focus on productivity, with few quantified earnings contributions; for credit investors, the more observable signal at present is not spread divergence, but rather that AI investment and data center construction may expand corporate debt financing needs.

Core views

First, AI-related mentions are rising simultaneously among U.S. and European corporates, with the greatest concentration in the technology and communication services sectors. Second, management teams' main AI narrative remains productivity improvement, rather than large-scale layoffs or already realized earnings contributions. Third, healthcare and biopharma and utilities lean more toward a productivity-benefit narrative, while software, communication services, and real estate are more often pressed on AI-driven business model disruption. Fourth, software spending has not shown a significant surge among non-hyperscale cloud providers and non-software companies; companies are more often building capabilities in key differentiated scenarios while continuing to rely on external vendors. Fifth, the credit implication of AI investment is more on the supply side and may push up USD/EUR IG corporate bond issuance volumes.

Analysis framework

The report combines macro AI adoption research, corporate earnings-call text screening, company disclosure review, and corporate bond supply estimation to first assess how AI narratives are spreading across regions and industries, and then evaluate their effects on productivity, labor, software spending, and corporate financing demand. The sample includes 500 U.S. and 175 European investment-grade non-financial corporates, along with a more detailed review of a subsample of large issuers representing about 50% of the outstanding amount of the USD IG and EUR IG non-financial corporate bond indices.

Methodology notes

• Text Signal AnalysisEarnings Call AI Mention Screening

  Measure the frequency and industry distribution of AI- and software-related language in corporate earnings calls.

  This method is used to observe the speed of AI theme diffusion among U.S. and European investment-grade corporates and to identify narrative differences across sectors such as technology, communication services, real estate, healthcare and biopharma, and utilities.

• Company Disclosure ReviewLarge Investment-Grade Issuer Disclosure Sample Analysis

  Review disclosures, presentation materials, and earnings calls from 42 large U.S. and 46 large European non-financial investment-grade issuers.

  This subsample is used to assess how companies are deploying AI, whether they quantify productivity or earnings impacts, whether they mention labor impacts, and whether they are building internal software capabilities or increasing software spending.

• Credit Supply AnalysisAI-Related Corporate Bond Supply Estimation

  Estimate corporate bond financing by hyperscale cloud providers, data center construction, and AI-related non-hyperscale issuers.

  The report maps AI investment demand to corporate bond issuance supply, arguing that general corporate purpose financing and centralized liquidity pools make tracking more difficult, but the upside risk on the supply side still merits attention.

Asset mapping & comparison

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

• USD IG non-financial corporate bond index

  AI investment may affect the supply side through additional issuance volume.

  Strengths

  Large U.S. companies show high AI mention intensity, with deeper AI integration in technology and communication services, and related investment and financing signals appearing earlier.

  Weaknesses

  Productivity benefits are rarely quantified at the earnings level, making it difficult to translate them into a clear spread advantage at present.

  Comparison

  European companies' AI mention intensity has already converged toward the U.S., and regional differences alone are insufficient to support a judgment of significant spread divergence.

  Risks

  If AI-related issuance continues to expand, additional supply may pressure the credit market's absorption capacity and spreads.

• EUR IG non-financial corporate bond index

  European companies' AI narrative is catching up with the U.S., and the supply side may likewise be affected by AI investment financing demand.

  Strengths

  About 67% of large European companies primarily describe AI as a productivity-enhancing tool, and the pace of disclosure diffusion is relatively fast.

  Weaknesses

  The European sample has fewer reporting companies in some quarters, so changes in AI mention intensity may be affected by sample coverage.

  Comparison

  Compared with the U.S., Europe has converged in AI mention intensity, but has lower rates of internal software capability building and increased software spending.

  Risks

  If AI investment is financed through corporate bonds, EUR IG supply forecasts may rise; however, precise identification is constrained by disclosure limits around general financing purposes.

• Corporate bonds of hyperscale cloud providers and data center-related companies

  They are the most visible and direct source of credit supply tied to AI buildout.

  Strengths

  Hyperscale cloud providers have already issued about $170 billion this year, and demand for data center construction and AI capex is clear.

  Weaknesses

  Capex is massive, cash flow usage is high, and follow-on financing demand may continue to rise.

  Comparison

  Compared with other sectors, these are easier to identify as AI-related financing, but they still do not capture all AI investment.

  Risks

  If capex continues to exceed expectations, corporate bond supply and refinancing demand may increase further.

• Credit in the software sector and software vendors' business models

  AI is driving discussion around substitution, internal capability building, and changes in software purchasing models.

  Strengths

  Companies still rely on external vendors for general software needs, and IT executive surveys show more customers prefer 'build less, buy more.'

  Weaknesses

  Some large issuers are beginning to build proprietary AI software capabilities for high-value use cases, which may pressure growth expectations for some vendors.

  Comparison

  In earnings calls by non-hyperscale and non-software companies, software mention intensity has not risen significantly.

  Risks

  The software sector faces the risk that AI reshapes the application-layer value chain, reallocates customer budgets, and changes valuation expectations.

• Healthcare and biopharma, and utility issuers

  These sectors most consistently link AI with productivity improvement.

  Strengths

  Healthcare and biopharma benefit from drug discovery, clinical development, patient risk stratification, and administrative automation; utilities benefit from grid optimization, outage prediction, predictive maintenance, and power demand from data centers.

  Weaknesses

  Actual earnings contributions still lack broad quantification, and the productivity narrative has not yet fully translated into verifiable financial gains.

  Comparison

  Compared with software, communication services, and real estate, these sectors are more often seen as beneficiaries of AI productivity rather than being disrupted by it.

  Risks

  Regulation, execution costs, data governance, and capex needs may weaken the positive credit impact of productivity gains.

• Real estate and communication services issuers

  These sectors are more often questioned about second-order AI effects and business model risks.

  Strengths

  Some companies can improve efficiency through automation, customer interaction, and network management.

  Weaknesses

  Real estate faces concerns about AI-driven labor structure changes and declining office demand; communication services face questions about the resilience of content economics, advertising, and bundled distribution models.

  Comparison

  Compared with healthcare and biopharma and utilities, AI discussion in these two sectors is more focused on risk and disruption.

  Risks

  If AI changes office demand, advertising allocation, or the economics of content production, related credit fundamentals may come under pressure.

Key data

• U.S. corporate AI adoption rate19.5%, expected to rise to 22.7% over the next six monthsGoldman Sachs' macro team believes AI adoption is still rising but remains at an early stage.
• AI adoption rate among large companies36.1% for organizations with more than 250 employees, 30.6% for mid-sized companiesLarge companies still lead, while mid-sized companies are catching up.
• AI-related job changesAI-exposed industries are losing about 11,000 jobs per month, while data center-related construction is adding about 9,000 jobs per monthLabor impacts remain uneven, and the overall effect is relatively modest.
• Evidence of generative AI productivity improvementAbout 23% on average in academic research, about 33% in company case studiesDeployed use cases show clear improvement, but this has not yet been widely quantified at the earnings level.
• Corporate disclosure on AI productivityAbout 54% of companies discuss AI productivity, 11% quantify returns from specific use cases, and 2% quantify earnings impactThe productivity narrative is strong, but earnings evidence remains limited.
• Expected AI capex by the four major hyperscale cloud providersAbout $5.3 trillion in total from 2025 to 2030Hardware and data center capex are the most visible sources of AI investment in credit markets.
• Estimated global non-hardware AI spendingMay exceed $1 trillionIncluding internal software, data infrastructure, and organizational transformation, likely absorbed mainly through internal labor costs.
• Earnings call screening sample500 U.S. investment-grade non-financial corporates, 175 European investment-grade non-financial corporatesUsed to compare the evolution of AI and software mentions across regions and industries.
• Large issuer review sample42 U.S. companies, 46 European companiesRepresents about 50% of the outstanding amount of the USD IG and EUR IG non-financial corporate bond indices after excluding international issuers and hyperscale cloud providers.
• Change in AI mention intensityFrom less than 0.5 times per company per quarter in early 2022 to about 4 times recentlyEurope had largely converged toward the U.S. level by late 2025.
• Share of large companies using a productivity narrativeAbout 60% in the U.S., about 67% in EuropeThe mainstream framing is that AI improves productivity, usually without explicitly mentioning workforce impact.
• Statements about labor impactAbout 19% of companies mention doing more work with fewer people; explicit layoff language is 7% in the U.S. and 2% in EuropeSlower hiring is more common than direct layoffs.
• AI and other topics9.5% of U.S. companies and 6.5% of European companies mention other factors such as security or regulation; fewer than 5% do not discuss AIAI has already broadly entered large-company disclosures.
• Internal software capability building9.5% of U.S. companies and 6.5% of European companies mention internally building AI-enabled software or proprietary capabilitiesMainly for high-value or differentiated use cases, while continuing to rely on external vendors for general software needs.
• Increase in software spending24% of U.S. companies and 15% of European companies discuss higher software spendingDriven by healthcare and biopharma in the U.S. and by industrial companies in Europe.
• IT executive survey56% expect less in-house building and more software purchasing; 17% expect more internal developmentBroadly consistent with the large-issuer disclosure signal of selective in-house building and external buying for general needs.
• Corporate bond issuance by hyperscale cloud providers this yearAbout $170 billionAlready issued in the global syndicated corporate credit market.
• Non-hyperscale AI-related supplyAbout $200 billionFrom non-hyperscale issuers in the USD and EUR markets directly related to AI buildout and data center financing.
• Estimated global AI-related syndicated corporate bond issuanceAt least about $370 billionIncludes hyperscale cloud providers and industries related to AI buildout and data centers, and may still underestimate broader AI spending by corporate borrowers.

Impact & implications

The report's implication for investment-grade corporate credit markets is more on the supply side than the spread side. AI adoption and investment are still spreading, but productivity and earnings contributions have not yet been widely quantified, so there is not enough evidence to show that U.S. and European investment-grade credit spreads will diverge significantly because of differences in AI adoption. By contrast, data center construction, hyperscale cloud provider capex, non-hardware AI investment, and broader AI deployment across industries may increase corporate financing demand and create upside risk to 2026 total supply forecasts for USD and EUR IG. Because many investment-grade corporates issue debt for 'general corporate purposes' and liquidity pools are centralized, tracking AI-related financing is relatively difficult.

Risks

• AI productivity gains have not yet been widely quantified, and management narratives may be ahead of actual financial impact.
• AI-related bond supply may be obscured by financing labeled as 'general corporate purposes,' causing the market to underestimate true financing demand.
• Centralized corporate cash and liquidity pools make the specific funding sources and uses of AI investment difficult to trace.
• The massive capex scale of hyperscale cloud providers and data center construction may increase leverage, free cash flow pressure, or refinancing needs.
• The software, communication services, and real estate sectors may face AI-driven business model re-rating risks.
• Security, regulation, data governance, and execution complexity may slow AI adoption and weaken productivity improvements.
• Current evidence is insufficient to support significant divergence in U.S. and European credit spreads, and trading regional AI differences too early may create misjudgment risk.

What to watch

• Whether quarterly AI mention intensity among U.S. and European companies continues to rise, and whether Europe remains close to the U.S. level.
• Whether large companies begin quantifying AI's impact on specific use cases, cost savings, revenue growth, or earnings per share.
• Whether issuance related to AI, data centers, and cloud capex in the USD and EUR IG markets continues to exceed expectations.
• Whether the stated use of corporate bond proceeds shifts from 'general corporate purposes' to more explicit AI, data center, or software investment language.
• Whether the AI productivity narrative in healthcare and biopharma and utilities translates into verifiable margin improvement.
• Whether AI disruption risk intensifies further in earnings-call Q&A for the software, communication services, and real estate sectors.
• Whether companies increase internal software development, step up external procurement, or adopt a hybrid model.
• Whether Goldman Sachs will in the future incorporate a higher qualitative upward adjustment into its total supply forecasts for USD and EUR IG.