This week we have the most hotly anticipated securitisation of the year (by me at least) and a potential deal of the year candidate (by vulgar publications which publish such rankings).
We refer of course to Deutsche Bank’s PayPal credit card takeout. It doesn’t give too much away about our internal processes that the Slack thread we have running is titled “OMG PayPal”.
Why do we care so much? And why should you care so much?
Number one, it is PayPal — it’s a massive company which has been the counterparty for some of the biggest asset-backed trades of the last five years. It’s a financial which is strategically moving to a balance sheet lite model, and that’s catnip to funds looking for assets.
We’ve written here and here and here about KKR’s Danube project on PayPal’s European BNPL book; it’s also a key milestone for KKR’s ABF strategy, and a large position in some of its public ABF-exposed funds.
The second reason we care is because the UK credit card deal now on screen is one of the most structurally innovative deals in post-crisis history. It solves a fiendishly difficult problem: how to deconsolidate — for accounting, regulatory and SRT purposes — a deal backed by rapidly revolving credit card receivables which require constant support and maintenance.
It’s a trade that people thought couldn’t be done, but here it is, selling bonds, presumably with robust legal and accounting opinions laying out how these assets get off PayPal’s balance sheet.
We’ve done a quick pass over the structure here, but the gist is — a normal master trust has a seller’s share, which solves most of the problems of a credit card master trust, handling refinancing uncertainty, sizing uncertainty, and refunds, among other things.
The route to deconsolidation here has been to create an investor-only master trust, so all of the things the seller’s share would usually address need to be addressed by other deal mechanics.
Any one of these would be a noteworthy innovation in its own right — together, we are in fully uncharted territory. This is, for a certain value of cool, one of the coolest deals ever; ground-up problem-solving threading needles across multiple regimes (US, EU, UK accounting and audit, UK and EU bank regulation, UK consumer credit, UK and EU securitisation regs).
Legal honours belong to Clifford Chance (arranger side) and AO Shearman for PayPal.
Someone involved described it as “pretty simple, really”…but only at the highest conceptual level. Once you get into the weeds, the details really matter.
The third reason we care about PayPal is because it’s a test of arranger Deutsche Bank. Deutsche beat out a bunch of the usual fund suspects to win the portfolio last year, but it was a very unusual process. PayPal didn’t hire external advisors, and it asked potential bidders to submit their deconsolidation analyses.
That’s obviously the kind of ask which an investment bank with a big FIG practice is well-placed to meet, versus a thinly-staffed hedge fund, but price was a factor too, and Deutsche outbid competitors to have the right to buy and structure the deconsolidation.
No structurer I have met is scared of a complex deal structure (like me, they mostly think it is cool and interesting) but obviously it’s not just about solving all of the problems detailed above. It’s also about actually making some money out of the thing.
Deutsche Bank has been firing up its principal finance business lately (helped by last year’s hire of Cenan Djenan) and there’s been one takeout so far in Europe (Moorfields Consumer Funding, backed by loans from Fintern / Abound). DB also has a forward flow with Keystone Property Finance, a flow with Funding Circle (DB was formerly a senior financier) and probably others as well.
We have neither the info nor the expertise to assess the economics of the PayPal takeout — but it’s by far the most ambitious of these transactions.
We’d assume that the structural innovation also meant a smaller potential audience for the equity package. Structured credit hedge funds aren’t scared of a complex deal structure, but there aren’t that many possible counterparties for a £2bn deal size, and it’s possible that structure knocks out some of the more real-money type buyers. Some thinly staffed funds might have also felt that the game wasn’t worth the candle; how much staff time is required to work up a competitive bid on something like this, versus a regular mortgage portfolio?
Anyway, the trade is still in market (Deutsche joined by Bank of America and Lloyds as JLMs) as Thames 2026-1, so it’s a bit early for the teams involved to take their victory lap, but we’ll be covering in depth when they do. Break a leg!
9fin news
The challenger bank that looked at an MFS facility this Feb
Apollo’s new CLO structure evolves
How DB and PayPal squared the circle
What’s the resale value on Chinese EVs?
Apollo launches €10bn credit platform
Pepper cooks up portfolio sale alternative
Data center CMBS and ABS could benefit from slowdown
Warm Homes scheme to fuel securitisation
Will the CFPB come roaring back?
Selling at the wrong time
It’s been quite a week for AI news. We wrote last week’s column partly in the glow of a post-midterms rally, only to find a pretty big equity puke by the time we came to file, and then a relief rally this week — exhausting stuff, and then came what even The Economist calls “the most expensive shag in history”.
At the time of writing, Jensen Huang had stepped down as Nvidia CEO, but the company was still worth more than $4trn, so some of the alarmism about its centrality to the whole AI complex is…a bit overdone, we think?
But what concerns us is that a number of banks seem keen to sell decent chunks of their data centre lending. 9fin reported this week that SG was selling some $300m of its exposure to the Stargate complex, the previous week saw a big BWIC from ING, and before that, there was a large trade out for some $200m of Project Jupiter.
These numbers are dwarfed by what the banks still own, of course, but how should we read it? There are multiple reasons why they might be selling. Simple limit relief is one — banks just being too full up on these assets. Year-end balance sheet is fast approaching, which might be a reason to trim, and the ECB is reportedly planning to look at data centre exposures soon.
Whatever happens with construction after the mid-terms, 2027 will bring more data centre lending to do — and much more than the volume of loans currently being shopped around. If banks are giving themselves relief from limits now, those limits will be biting again, soon enough.
It’s seems likely that the mooted loan sales will involve losses, potentially quite substantial ones. As infrastructure loans, they should be held at par in the banking book; crystallising a hit of several 10s of millions by offering these out, especially with the current AI backdrop, seems a move of desperation — and yet, surprisingly popular.
The SRT market offers another route, but that has problems, too. SRTs take a long time to set up, need regulatory approval, and might not really do the trick. In Europe, unlike in the US, they can be used for limit relief but their primary purpose is freeing up regulatory capital. The structure and efficiency of a typical SRT transaction relies on a relatively granular portfolio, which might not work — even if a bank has a lot of projects on its books, the chances are they are credit-linked to a handful of firms.
The market is also fairly concentrated on the investor side. Infrastructure SRTs are already something of a specialist niche, requiring infrastructure expertise and loan-by-loan underwrites; concentrated digital infrastructure deals even more so.
Other approaches are also possible. Banks looking to derisk these exposures could structure FOLOs — first out last out. These grew to prominence in the private credit unitranche market, and amount to a back-door tranching of a given loan position, through an agreement between lenders. The borrower faces a single counterparty, but this counterparty can derisk (or lever up) behind the scenes.
For obvious reasons, this activity is hard to observe — and, for the sizes involved, probably means talking to the same group of large credit funds which are already up to their eyeballs in digital infrastructure debt and digital infra SRTs.
Meanwhile, Europe’s data centre ABS market is only just starting to gather steam. From a first deal in 2025, to a handful this year (a 15G has just been filed for another Vantage Germany deal), perhaps 2027 is the year it finally takes off. We hear optimistic words to that effect (but that’s what they said this time last year too).
