Listen to the latest episode of Credit Exchange with Lisa Lee
Published in London & New York
10 Queen Street Place, London
1345 Avenue of the Americas, New York
Creditflux is an
company
© Creditflux Ltd 2026. All rights reserved. Available by subscription only.
Opinion Credit
AI will upend operational models in a very short space of time
by Duncan Sankey
Duncan Sankey
Portfolio director and head
of credit research
Cheyne Capital
With online travel firm equities and life insurer CDS already off by a third, the systemic risk of AI is real
I love a good dystopia movie and, in the latest research blog from Citrini (‘The 2028 Global Intelligence Crisis’), I think we have the screenplay for a new classic. While not dismissing entirely Citrini’s post-AI wasteland of disenfranchised investment professionals and software solutions account managers, it’s worth remembering that dystopian futurescapes are conditioned by their current and historical context (the GFC, COVID, a fissure in the established political order). An alternative playbook might see AI-related profits taxed and redistributed as universal basic income, job sharing and a reduced working week. Call me a hopeless optimist…
None of which is to deny Citrini’s tenet that AI will upend operational models in several industries in a very short space of time. The equities of IG-rated online travel agencies Expedia and Booking Holdings are off 34% and 26%, respectively, from their 2026 highs, precisely because agentic AI could render the companies otiose.
Customers should be able to construct and book entire itineraries using AI agents, enabling suppliers to cut out fee-heavy middlemen and capture the all-important data they currently amass. Maybe customers will value the safety blanket OTAs provide if things go wrong; maybe the OTAs’ own efforts to pair up with LLMs will create new moats. So far, investors remain unconvinced.
Software firms are in the crosshairs
Citrini also zeroes in on the precarity of software as a service (SaaS) in the agentic AI world. Financial services firms (and others) currently pay (through the nose) for these offerings, which provide them with the basic plumbing of their businesses (payments processing, payroll, treasury functions), guardrails (anti money laundering, fraud detection, transaction monitoring, reporting, cybersecurity), portfolio management, data analytics and customer interfaces.
While firms are unlikely to trust regulatory functions to AI agents (and regulators probably won’t let them), they could become rapid adopters in other fields. Agentic AI could provide the means to in-house many of these services and make material savings on fees. At the very least, it seems reasonable that AI will squeeze SaaS profits and force a number of players out of business.
From a credit perspective, this is where it gets interesting. 9fin estimated that the IT and communications sector accounted for between 20% and 25% of private credit deals (a market now worth over USD 2tn). It is reasonable to expect that many of these transactions were leveraged, in which case any earnings squeeze threatens their ability to service debt. Some of the most active investors in this market have been US life insurers, which have parked their growing annuities into private credit assets (of questionable liquidity). Indeed, private capital has been active establishing itself in the life insurance business.
Ratings shopping in private credit
In my last article, I alluded to a proliferation of rating agencies accredited as Nationally Recognized Statistical Rating Organizations (NRSROs) by the US National Association of Insurance Commissioners (NAIC). There are now 10 — compared to only three in 2000.
Rating agency newbies have enabled insurers to ratings-shop for private letter-ratings for their private credit investments. Given that ratings are a key determinant of the risk-based capital factor assigned to the asset, the higher the ratings, the better for the insurer in capital terms. The difference between an NAIC 2 (roughly triple B) and NAIC 3 (roughly double B) rating equates to an up-to-4.5x increase in the risk-based capital (RBC) factor to be applied. Managers have also used pooling and tranching to gussy-up the ratings of underlying private loans.
Sound familiar? If agentic AI unleashes a wave of downgrades and defaults through the SaaS sector, could it prompt additional capital needs among life insurers? Could regulators wise up to the dangers of rating agency proliferation and enforce a more stringent approach, again with implications for RBCs? (Interestingly, while Egan-Jones is still a US NRSRO, it has been dropped by the Bermuda Monetary Authority). Could this be another transmission mechanism through which private credit risk becomes systemic?
Maybe this is just another dystopian screenplay. However, judging by the 30% CDS spread widening in names like MetLife, Pru and Lincoln National in the week to 2 March, it is one that has got a lot of people thinking.
