Exploring the risks posed by AI to technology and software loans
While the proliferation of artificial intelligence (AI) solutions is expected to eventually impact most major industries, investor concern is presently hyper focused on the threat posed to software companies.
Autonomous agents are becoming more capable of executing tasks directly within business applications, which could undermine the traditional per-seat subscription model of many software providers. At the same time, AI is lowering barriers to entry by reducing the cost and complexity of software development, making it easier for new competitors – or even corporate customers themselves – to build customized tools.
Where does the risk reside?
Software companies have leaned more heavily on leveraged loans and private credit for financing than on corporate markets. Business Development Companies (BDCs) have been drawn to software businesses for leveraged buyouts because their recurring subscription revenues, strong margins, and asset-light operating models have been well suited to supporting higher debt loads.
Given the higher technology and software exposure of leveraged loans or collateralized loan obligations (CLOs), investors may be tempted to eschew these asset classes altogether. But we think the issue requires a more measured approach and warrants a closer look at loan-level metrics.
As outlined below, our method involves isolating and quantifying the risk, seeking to mitigate risk through active management, and selecting the investment vehicles that we believe may offer structural protections against defaults.
- Isolate and quantify the risk
Technology and technology-adjacent loans represent around 22% of the U.S. CLO loan universe and 17% of the European CLO loan universe. However, in our opinion, sector classification alone overstates the risk because not all issuers are directly exposed to AI disruption.
Drawing on our strong corporate credit research capabilities, we assess each loan individually and have determined that only 11% of U.S. CLO loans and 9% of European CLO loans are directly exposed to AI disruption – roughly half the headline technology exposure.
We go a step further and separate the loans with direct AI exposure into three risk buckets using real-time pricing data – which we consider to be a highly efficient information aggregator – to determine whether the market perceives the loan to be lower risk, higher risk, or stressed.
Low-beta loans in the U.S. and Europe are pricing very close to par (100) at 99.2 and 99.4, respectively, indicating the market sees no credit concern in those loans. The high-beta risk cohort, priced at 94.0 and 96.5, reflects some uncertainty but is still priced well above distressed levels.
The stressed portion, at just 3% of the index, is where real credit risk resides. Nonetheless, the stressed cohort is much smaller than headline exposure might suggest.
- Seek to mitigate risk through active management.
Once we have appropriately isolated and quantified the risk, we seek to mitigate the risk through active management. The two salient points here are to a) assess the maturity profile of the stressed loans to quantify imminent refinancing risk, and b) identify variations in direct AI exposure by individual CLO deal.
As it relates to a), a potential concern with AI-disrupted borrowers is refinancing risk. If a loan matures too soon for a borrower facing structural revenue uncertainty, it may be forced to refinance at materially wider spreads or fail to refinance altogether.
In the U.S., stressed technology loans maturing within three years represent just 1.4% of the loan universe, while in Europe that number is 0.7%. No stressed technology loans mature within one year.
This healthy maturity profile provides time for credit situations to develop or resolve before refinancing pressure builds, and therefore no imminent maturity wall exists to force crystallization of losses.
Regarding point b), the figures above describe the universe as a single portfolio, yet individual CLO deals are very different.
After analyzing over 2,750 CLO transactions across U.S. and European markets, we found that exposure to stressed technology loans varied significantly by deal and by CLO manager. In our analysis, we identified and measured each CLO’s exposure to stressed technology loans and then ranked the deals from lowest to highest exposure and grouped them into quartiles.
Exposure to stressed technology loans in the most-exposed quartile was ~300% higher in the U.S. and ~600% higher in Europe compared to the least-exposed quartile.
With over 170 CLO managers operating across both markets and individual managers running deals of varying composition and vintage, the exposure is ultimately determined by credit selection at the deal level.
- Select the investment vehicles we believe may offer structural protection
In our view, investors seeking floating-rate exposure should consider CLOs over leveraged loans due to the structural protections and higher credit quality within CLOs, as well as the ability to select one’s desired level of credit risk.
Whereas CLOs employ a waterfall structure to provide credit enhancement, leveraged loans participate in default-related losses from the first dollar, as there is no tranche hierarchy inherent in direct loan exposure.
This credit enhancement feature within CLOs has proven resilient over time, particularly for higher-rated tranches. That’s because losses from underlying loans hit the equity tranche first, moving upward through B and BB tiers before touching the BBB, A, AA, and AAA tranches.
Due to the heightened level of uncertainty within the technology sector, we prefer exposure to the highest-rated (AAA) tranches of CLOs where structural protections are at their maximum. AA and A rated CLOs remain a compelling option as well, as they exhibit a very high degree of credit enhancement while offering some additional credit spread over the AAA tranche.
For investors in BB and B CLO tranches, the risks are more direct. A negative turn in AI-exposed credits could drive tranche rating downgrades and spread repricing further down the capital stack and we do not believe the BB and B tranches are adequately compensating investors for this risk. Therefore, we believe investors should consider moving up in quality to the BBB tranche, or potentially into the A, AA, and AAA tranches.
In summary
As technology author and consultant Geoffrey Moore has observed, “the most common misunderstanding of disruptive innovations is to overestimate their impact in the short term and underestimate it in the long term.” In light of the data presented – we believe the headline hype around software loans is overblown in the short term, while longer-term impacts remain to be seen.
Floating-rate bond exposure remains an essential component of a diversified fixed income allocation, with CLOs being the preferred investment vehicle due to their strong credit ratings, structural protections, and historical resilience.
A combination of deep loan-level analysis, active selection, and a bias toward higher-rated CLO tranches remains our preferred approach to navigating the risk of AI disruption within software loans. Selecting deals with disciplined credit exposure and avoiding those with concentrated risk in AI-sensitive sectors is where we see the clearest opportunity to add value for investors.
John Kerschner, CFA is Global Head of Securitised Products, Portfolio Manager
