Assessing AI-driven risks in European CLOs

Software has historically been viewed as one of the most defensive sectors in European leveraged finance, characterised by recurring revenues, strong cash flow generation and high barriers to entry. While, over the long term, the sector’s stability may still be a correct evaluation, the rapid development of large language models (LLMs) and related AI technologies is beginning to challenge some long‑held assumptions. This is driving greater dispersion across software credits and heightened scrutiny from European CLO managers.[1]

Against this backdrop, Janus Henderson has undertaken a detailed assessment of technology and software exposure within European CLOs. Using proprietary systems, we mapped technology exposure deal by deal and manager by manager, relative to both the European CLO universe and a representative loans index. We also identified market‑based signals of deterioration across specific sub-sectors and leveraged our extensive network and on‑the‑ground presence to engage with most European CLO managers.

Read more

In collaboration with our loans and corporate credit specialists, we framed our questions around how AI‑related risks are being assessed, priced and managed within CLOs and how these considerations are influencing underwriting, watchlists, and prospective portfolio actions. We asked CLO managers to distinguish whether recent volatility reflects idiosyncratic issuer-specific issues or a broader sector-wide dynamic, and to explain the top‑down rationale for any underweight or overweight exposure versus the relevant loans index.

We then assessed:

• the timing and conclusions of recent revenue and credit reassessments,
• how the forward-looking AI risks and competitive threats are monitored at the issuer level,
• the triggers that would prompt watchlist actions or de-risking, and
• tail-risk governance, by clarifying each portfolio’s avoidance (or disposal) of stressed issuers and the investment rationale, as well as expected recovery, for any remaining idiosyncratic exposures.

The prompt responses to our outreach suggest that AI‑related risks are firmly on CLO managers’ agendas. Across managers, AI is being assessed primarily through a credit lens rather than as a headline technology narrative. The primary concern is less about near‑term market disruption and more about AI’s potential to reshape competitive dynamics, compress terminal values, and widen dispersion between resilient and vulnerable business models.

In this paper, we summarise what we learned from the responses from CLO managers, highlighting where views are broadly aligned, where they diverge, and why AI is not viewed as a uniform, sector‑wide disruption. Instead, AI is seen as a differentiator across business models, with implications for valuations, refinancing risk, and long‑term credit quality. The variation in responses reveals meaningful differences in each manager’s views, underwriting assumptions, and intended portfolio actions. This reinforces the importance of active CLO manager selection, particularly as bifurcation in the loans market continues to unfold – this sits at the core of Janus Henderson’s active CLO management approach.

Footnotes 

[1] A CLO manager actively manages the portfolio of loans within a CLO.