UBS Says AI Disruption Is Raising the Floor for Private Credit Default Risk

Overview

UBS updated its view on AI disruption and private credit risk in this report. The report argues that AI is permanently reshaping software growth, competition, pricing power, and credit risk, and that this propagates through software and business services exposure into private credit portfolios such as BDCs. Although the high-risk exposure is lower than the estimate under the rougher framework from early February, medium-risk exposure remains large enough to push default rates higher in a late-cycle credit environment.

Core views

The key views are: first, the software sector faces a structural AI substitution risk, especially in workflow applications, rule-based applications, customer support, RPA, project management, and content creation; second, vertical SaaS, ERP financial systems, and data and infrastructure are relatively more defensive; third, overall BDC technology exposure looks more low-to-medium risk than previously feared, but still has about 11% high-risk exposure and about 45% medium-risk exposure; fourth, the incremental impact of AI disruption on private credit default rates is around 3-4 percentage points and could lift this year's default rate into the mid-single digits, with a tail scenario rising to around 10% by early 2027.

Analysis framework

The report uses a two-step approach: first, it maps BDC technology exposure into UBS's software team's high-, medium-, and low-AI-disruption risk buckets; second, it applies bottom-up revisions based on the defensive narratives of individual issuers and compares the result with the earlier two-tier, subsector-based risk framework. For the business services portfolio, because industry mapping precision is lower, the report relies more heavily on issuer-level classification judgments.

Methodology notes

• Theme strategy and credit risk mappingUBS AI Disruption Framework

  AI disruption risk tiers

  Software companies are grouped into high, medium, and low risk based on AI substitution pressure and business defensibility. Low risk includes vertical SaaS and ERP financial systems; medium risk includes CRM, ITSM, HCM, and MarTech, which have contextual data, governance, or delivery infrastructure but are still affected by automation; high risk includes rule-based, repetitive applications.

• Private credit portfolio analysisBDC portfolio issuer mapping

  BDC exposure mapping

  Technology and services issuers in the BDC portfolio are reclassified by AI disruption risk to estimate the incremental default pressure on the private credit portfolio.

• Scenario analysislate-cycle default framework

  Late-cycle default framework

  The report assumes a credit environment that is weaker than a benign cycle but not yet under recessionary stress, and notes that default outcomes under different macro assumptions could be 50% lower or 100% higher than the base case.

Asset mapping & comparison

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

• Software sector

  The main asset under pressure from AI disruption

  Strengths

  Vertical SaaS, ERP financial systems, and businesses with high regulatory complexity and high switching costs are more defensive.

  Weaknesses

  Application software and rule-based workflows are easier to substitute with AI, which pressures growth, pricing power, and margins.

  Comparison

  Compared with semiconductors and hardware, software coverage sentiment is weaker, and capital and IT budgets are shifting toward AI experimentation and deployment.

  Risks

  AI-native competition, customer budget migration, insufficient personnel cost adjustment, and deteriorating credit quality.

• Private credit and BDC portfolios

  Indirectly affected through technology and business services exposure

  Strengths

  High-risk exposure is lower than the rough estimate under the earlier two-tier framework, and the portfolio is not fully exposed to the most vulnerable software assets.

  Weaknesses

  Medium-risk exposure remains large, and the analysis covers only about 55% of the BDC portfolio; healthcare and other sectors are not fully included.

  Comparison

  The updated framework is more granular than the February assumption, but it still shows that AI will have a material incremental impact on default rates.

  Risks

  Late-cycle credit conditions, rising default rates, valuation and liquidity pressure, and issuer-level classification error.

• Semiconductors and hardware

  A relative beneficiary of AI budget migration

  Strengths

  Higher spending on AI compute, model performance, and infrastructure benefits the related hardware chain.

  Weaknesses

  The report does not provide valuation or ratings for specific semiconductor or hardware companies.

  Comparison

  Relative to application software, semiconductors and hardware are more easily seen as beneficiaries of AI capex.

  Risks

  Capex cycle volatility, elevated valuations, supply-chain risk, and macro demand swings.

Key data

• BDC technology exposureabout 24%Technology accounts for about 24% of BDC holdings.
• High-risk share of technology exposureabout 11%Under the updated framework, about 11% of BDC technology exposure falls into the high AI disruption risk bucket.
• Medium-risk share of technology exposureabout 45%Medium risk remains the main source of pressure.
• High-risk exposure in business serviceshigh single digitsThe business services portfolio is more affected by workflow automation than by direct substitution.
• Medium-risk exposure in business servicesabout 25-30%This remains important in a late-cycle environment.
• Incremental default pressure from AIabout 3-4 percentage pointsThe updated bottom-up analysis estimates the marginal rise in default rates from AI disruption over the next year.
• Private credit default-rate pressure this yearabout +5% to 9-10%The report sees the risk tilted to the upside.
• Tail scenarioabout +10% by early 2027If the credit cycle deteriorates, the increase in defaults could widen materially.

Impact & implications

The investment implication is that the market should not view AI only as an equity growth theme, but also incorporate it into credit risk pricing. For private credit portfolios with high software exposure, AI may lift the default floor by slowing revenue growth, compressing margins, shifting budgets, and increasing refinancing pressure. By contrast, semiconductors, hardware, data, and infrastructure are more likely to benefit from the migration of AI capex, while traditional application software and borrowers tied to workflow automation are more vulnerable.

Risks

• Macro and credit cycle assumptions are the largest uncertainty, and default outcomes could be materially lower or higher than the base case.
• AI disruption risk classification relies on issuer-level judgment and may overclassify or underclassify risk.
• The report covers only about 55% of the BDC portfolio, and healthcare and other sectors are not fully included in the analysis.
• If company management teams do not actively invest in AI-native workflows, compute, and cost restructuring, credit quality could deteriorate further.
• Market risk, credit risk, interest rate risk, foreign exchange risk, geopolitical risk, and policy shocks may affect multi-asset returns.

What to watch

• The pace of penetration of AI platforms such as OpenAI and Anthropic into enterprise workflows.
• Whether software companies can maintain margins through AI-native products, cost restructuring, and workforce optimization.
• Default, downgrade, and refinancing trends among technology and business services issuers in the BDC portfolio.
• Whether private credit default rates continue rising from the base-case mid-single digits toward the 9-10% or higher tail scenario.
• The persistence of IT budget migration from traditional software to AI deployment, semiconductors, and hardware infrastructure.