UBS believes AI disruption risk remains underestimated in leveraged loans

Overview

This report is the third edition of UBS Global Strategy's quantitative research on AI disruption risk, focused on US/EU leveraged loans. It extends the previously used bottom-up stress-testing framework for HY to leveraged loans and concludes that US and EU loan portfolios have similar exposure to AI disruption risk, with default rates potentially reaching around 7-8% in tail scenarios. As a result, current leveraged loan spreads still do not adequately compensate for the risk.

Core views

There are three core views. First, the bottom-up framework increases confidence in tail default rate estimates: US/EU LL default rates could reach around 7-8% in an aggressive AI disruption scenario, close to UBS's roughly 9% top-down estimate. Second, risk exposure in European and US leveraged loans is more similar, leading UBS to raise its 2026 base-case default forecast for EU LL to 3.0% and lower that for US LL to 3.75%. Third, the leveraged loan market still underestimates AI disruption risk, especially in sectors outside Technology such as Services, Media/Telecom, Gaming/Leisure, and Retail.

Analysis framework

The report uses a bottom-up classification at the sector and issuer level, dividing loans into high, medium, and low AI disruption risk categories, then combines this with a credit quality matrix to estimate incremental default intensity and adds contagion effects to derive market-level default rates. It then cross-checks the estimated incremental default risk against spread changes since the start of the year, regional AI adoption rates, sector risk distribution, and 2026 spread forecasts.

Methodology notes

• Credit stress testBottom-up AI disruption risk framework

  Classifies high, medium, and low AI disruption risk by issuer and sector, and estimates incremental default intensity.

  This framework is used to complement top-down default rate forecasts, with the goal of identifying which loan issuers and sectors are more likely to experience additional default pressure under an aggressive AI disruption scenario.

• Scenario analysisAggressive AI disruption scenario

  Assumes accelerated AI adoption and shocks to revenues, margins, or business models in highly exposed sectors.

  Under this scenario, the report estimates market default rates for US/EU LL at around 7-8% after incorporating contagion effects, an increase of about 3.5-4.5 percentage points from current levels.

• Market pricing testComparison of incremental default risk versus spread changes

  Compares whether sector-level incremental default risk matches YTD spread widening.

  The report finds that Technology has already widened meaningfully, while sectors with similar risk such as Media/Telecom and Services have shown insufficient spread response, leading to the conclusion that risk remains underestimated.

Asset mapping & comparison

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

• US LL

  Core stress-test target

  Strengths

  Deep market depth; some sectors such as Technology have already partly repriced AI risk.

  Weaknesses

  More high-risk sectors with broader distribution; nine sectors contain high-risk issuers, and higher AI adoption may lead to faster risk transmission.

  Comparison

  Compared with EU LL, the overall tail default rate estimate is similar, but contagion risk and timing risk are higher in the US.

  Risks

  AI disruption could pressure business models, tighten refinancing conditions, drive further spread widening, and increase default rates.

• EU LL

  Core stress-test target

  Strengths

  Risk concentration is somewhat higher and AI adoption is slower; UBS still expects relative outperformance versus US LL over the medium term.

  Weaknesses

  High overlap with US LL issuers means overall high/medium-risk exposure is not materially lower, and current spreads may still be insufficient.

  Comparison

  UBS raised its end-2026 EU LL spread forecast to 590bp and its base-case default forecast to 3.0%.

  Risks

  If AI adoption accelerates or financial conditions tighten, European loan portfolios could still face delayed but meaningful default pressure.

• US HY / EU HY

  Reference asset classes

  Strengths

  The bottom-up framework previously supported a HY tail default rate estimate of about 4-5%, and EU HY is used as a relatively preferred credit asset.

  Weaknesses

  HY is also exposed to macro, financing, and sector disruption risks.

  Comparison

  The report reiterates Long EU HY vs. EU LL, arguing that EU LL offers insufficient relative compensation.

  Risks

  If AI shocks spread or market liquidity deteriorates, HY spreads could also widen.

• Services、Media/Telecom、Gaming/Leisure、Retail

  High AI disruption risk sectors

  Strengths

  Some sectors have not yet repriced materially and may offer short or avoidance opportunities in relative value trades.

  Weaknesses

  They have relatively high shares of high or medium AI disruption risk, while spread compensation remains relatively insufficient.

  Comparison

  Technology has already widened significantly, while similarly risky sectors such as Media/Telecom and Services have reacted less.

  Risks

  They may later see catch-up downside, upward revisions to expected defaults, and wider liquidity discounts.

• Manufacturing、Food & Drug、Metals/Minerals、Energy

  Relatively low AI disruption risk sectors

  Strengths

  The report considers Food & Drug, Metals/Minerals, and Energy relatively insulated, while Manufacturing is used as a relative long.

  Weaknesses

  They still face general credit, rate, macro, and geopolitical risks.

  Comparison

  AI disruption exposure is lower relative to high-risk sectors such as Services.

  Risks

  If the macro environment deteriorates, low AI risk does not mean low credit risk.

Key data

• US/EU HY base-case default rate forecast by end-20261.8% / 1.0%Versus current levels of about 1.2% / 1.6%.
• US/EU LL base-case default rate forecast by end-20263.75% / 3.0%UBS lowered US LL from 4.0% to 3.75% and raised EU LL from 2.8% to 3.0%.
• US/EU LL default rate under an aggressive AI disruption scenarioaround 7-8%After incorporating contagion effects, close to UBS's roughly 9% top-down tail estimate.
• US/EU LL incremental default riskabout +3.5 to +4.5 percentage pointsThe bottom-up framework shows incremental default intensity of 4.1% for US LL and 3.6% for EU LL.
• Current US/EU LL spreads498bp / 515bpThe report says they have widened only 43bp / 41bp since the start of the year, and have narrowed somewhat since the start of the month.
• EU IG/HY/LL spread forecasts by end-202685bp / 285bp / 590bpUBS raised its year-end EU LL spread forecast to 590bp, implying about another 75bp of widening from current levels.
• Share of high-risk loans in US/EU LLabout 4-5%Medium-risk loans account for about 25-30%, with broadly similar overall exposure across the two regions.
• Overlap of US and European loan issuersabout 50%The high overlap is an important reason why AI risk exposure is similar across the two regions.
• YTD spread change in US Technology loansabout +260bpBy comparison, US Media/Telecom is about 0bp and US Services about +54bp, indicating insufficient repricing in non-tech high-risk sectors.
• Comparison of corporate AI adoption ratesUS higher than EU, with a larger gap among small firmsCharts show US AI adoption is significantly higher than EU for Small, Medium, and All categories, while Large firms are similar across the two regions.

Impact & implications

The investment implication is that spreads on leveraged loans, especially in sectors with high AI disruption risk, may continue to widen. UBS believes US and EU risk exposure is similar, but US sector risk is more dispersed and AI adoption is faster, so risk transmission may occur earlier; Europe may still outperform the US over the short to medium term. Strategically, the report prefers Long US LL Manufacturing vs. US LL Services, and Long EU HY vs. EU LL.

Risks

• AI model capabilities may improve faster than expected, pressuring revenue and margins in high-risk sectors.
• Redemptions or liquidity pressure in private credit could amplify contagion effects in the loan market.
• A tighter refinancing window and worsening financial conditions could push default rates higher.
• If current spreads continue to tighten, compensation for tail risk will decline further.
• Higher US AI adoption rates could cause US LL risk to surface earlier.
• Geopolitics, policy shocks, interest rates, FX, and market volatility could affect multi-asset portfolio returns.

What to watch

• Whether US/EU LL spreads move toward UBS's end-2026 forecast path, especially around EU LL 590bp and US LL 610bp.
• Whether Services, Media/Telecom, Gaming/Leisure, and Retail undergo repricing similar to Technology.
• Progress in AI models, corporate AI adoption rates, and signs of automation substitution across sectors.
• Private credit redemptions, loan fund flows, refinancing deal activity, and secondary-market liquidity.
• Whether actual US/EU LL default rates move toward the base-case forecasts of 3.75% / 3.0%.
• Whether the spread gap between high-risk and low-risk sectors continues to widen.