Four months ago, when a couple of bankruptcies in the auto sector riled regional lenders and stirred private credit fears, I lamented the necessity of documenting what I called “embryonic tail-risk realization.”

In simple terms, I don’t enjoy making mountains of molehills, and that’s what the October 2025 “roach” panic was: An exercise in dramatizing and otherwise exaggerating the read-across from a pair of idiosyncratic frauds, for lack of a more polite way to describe the goings-on at First Brands and Tricolor.

Idiosyncratic fraud isn’t systemic, pretty much by definition. Jamie Dimon used the old roach analogy to suggest that First Brands and Tricolor were indicative of an infestation. There’s never just one roach, the old adage tells us, and we all know that coming out of periods defined by easy money, bad lending gets exposed. The worst loans, as they say, are made during the best of times.

But, as I was keen to note in October, those weren’t the roaches you were looking for. “I don’t think these hiccups are material in any sense, let alone in any big-picture sense,” I wrote, on October 18, of scattered losses tied to the above-mentioned frauds. A couple of days later, the market forgot all about First Brands and Tricolor.

Fast forward to February and I’m getting similar vibes — i.e., “embryonic tail-risk realization” vibes — from the unfolding Blue Owl saga and the “loan-ageddon” narrative more generally. This time, I’m not inclined to call the concerns entirely unfounded.

Although Blue Owl’s garnering all the headlines, this isn’t just a Blue Owl story. And “loan-ageddon” doesn’t center around any fraud (every business has fraud, but that’s not the central theme here).

Rather, long-simmering concerns around the exponential expansion of private credit have bubbled over in recent weeks amid i) loan quality worries tied to agentic AI’s potential to disrupt the businesses of software companies, and ii) generalized angst around over-lending for AI infrastructure development.

Blue Owl’s knee-deep in the AI buildout, having led or participated in tens of billions worth of deals in recent months. As discussed here, the firm puts up cash and lines up construction loans for data centers which it then owns on the understanding that large companies, including so-called “hyper-scalers” like Oracle, will rent the facilities.

Those deals are assumed to be as safe as safe gets. After all, what’s safer than a long-term lease with the largest, most reputable companies in the world? That’s a rhetorical question. A lot could go wrong there. The technology could, for example, develop along a different arc that doesn’t lean so heavily on space and compute.

In a worst case scenario for a firm like Blue Owl, the AI frenzy could peter out such that demand for physical infrastructure leases falls well short of forecasts, even as the technology becomes ubiquitous enough to bankrupt some software businesses.

So, again, we’re witnessing a match on dry kindling moment for private credit, and Blue Owl’s among the largest players. The kindling is the $1.3 trillion US private credit market itself. The match is AI-related ambiguity, including pressing questions about the advisability of financing data centers and jitters around an existential disruption moment for indebted SaaS businesses.

Late last year, amid escalating redemptions, Blue Owl decided to merge one of its non-traded funds, Blue Owl Capital Corp. II, with one of its listed portfolios. Viewed in isolation, a merger between a private fund and a publicly-traded sister vehicle isn’t alarming. Indeed, that’s often the end game for smaller vehicles: They raise capital, invest it for a while, then provide investors with a liquidity event, in this case a merger with Blue Owl’s (much larger) publicly-traded BDC.

But the merger announcement didn’t go over well for a number of reasons, not least of which was that investors in the unlisted vehicle were staring at a potentially large paper loss stemming from the NAV discount to shares of the listed fund. That’s a kind of design flaw in the industry, and it’s not specific to Blue Owl.

The publicly-traded shares fell further after the merger announcement, in part because Blue Owl said redemptions from the non-traded portfolio were restricted until the merger closed, a stipulation which spooked investors in the public fund.

On November 19, Blue Owl abandoned the merger plan. The simple figure above, which shows the reaction in the publicly-traded BDC, goes a ways towards explaining why. The firm cited “current market conditions.”

It’s fair to say “market conditions” have deteriorated further since then, and redemption requests haven’t let up enough (if they’ve let up at all) to relieve the pressure. So, it wasn’t terribly surprising when Blue Owl this week gated the same non-public fund.

Late Wednesday, the firm said it would stop offering regular, quarterly redemptions in favor of payouts from, in part, the sale of portfolio loans. Investors in that fund thus lost some of their agency: They can no longer decide when to redeem.

The mid-week revelation triggered more losses for shares of the publicly-traded parent (distinct from shares of Blue Owl’s publicly-traded BDCs), which are now down nearly 30% for 2026 and 60% from the January 2025 highs.

Those shares are down five weeks in a row, eight in 11, 12 in 17, 15 in 22 and… well, you get the idea.

“We’re not halting redemptions, we’re just changing the form” Craig Packer, the firm’s co-president, told CNBC’s audience, referring to Blue Owl Capital Corp. II. “If anything, we’re accelerating redemptions.”

I don’t love that damage control messaging strategy. Intentional or not, it can come across to some observers — and, I imagine, to a lot of investors — as obfuscatory, indignant and a bit supercilious.

When you stop tendering for shares on a predictable quarterly schedule, you’re halting redemptions in spirit, even if you simultaneously announce an initiative which results in the return of more capital than investors could’ve accessed under the preexisting arrangement, which is what Blue Owl did.

Specifically, Blue Owl sold 34% of Blue Owl Capital Corp. II’s investment commitments as part of a larger divestment in which the firm offloaded $1.4 billion of direct lending investments to pension and insurance funds.

The share of that sale from Blue Owl Capital Corp. II — $600 million — will fund a capital distribution to investors. Hence Packer’s contention that the firm’s actually “accelerating” redemptions.

The press release made no mention of market conditions, and indeed suggested buyers were beating down Blue Owl’s door to get their hands on these commitments. “We saw strong demand,” Packer said, adding that “interest far exceed[ed] the value of the investments we ultimately chose to sell.”

I don’t doubt that, nor am I trying to cast aspersions, but the optics aren’t great. If they were, the parent’s stock wouldn’t have fallen 13% on Thursday and Friday.

Blue Owl made a big deal of the fact that the direct lending investments it offloaded were sold more or less at par. The implication: The marks for the portfolio aren’t overstated.

Maybe that’s true, but what do you sell when you need liquidity? Your best assets. Your other assets might be good too, but they’re probably not as good, otherwise “best” has no meaning. That raises this question: What, exactly, does the quality composition of Blue Owl’s portfolio look like now, post-sale?

The firm said this week that, “The largest industry represented” in the sale was “internet software and services at 13%,” a share it called “generally consistent with the industry composition of Blue Owl’s overall direct lending strategy.”

So, we can say 13 cents of every dollar across Blue Owl’s direct lending is tied to software, a sector which, as noted above, is under a lot of scrutiny right now in the face of what some say is an existential threat from agentic AI.

Blue Owl cited the 99.7% of par it received for the entire $1.4 billion loan sale as evidence of “confidence in the quality and valuations of software investments.” Setting aside that we don’t know what the price was for the software investments specifically and assuming something very close to par for those investments, all that says is that there was confidence in those software loans.

Perhaps we can extrapolate from that to say something reassuring about the rest of the software lending across Blue Owl’s platform. It’s hard to know. My guess is that we can — extrapolate, I mean — at least for now. But, again, that sector’s under a lot of pressure. Will Blue Owl be able to sell those investments at or near par six months from now? Who knows.

Not helping matters from a PR perspective, some of the portfolio loans Blue Owl sold where reportedly purchased by Kuvare Asset Management, a Chicago-based boutique specializing in management services for the insurance industry. Why does that matter? Well, because guess who owns Kuvare Asset Management? Spoiler alert: Blue Owl, which bought the firm in 2024.

In the above-mentioned interview with CNBC, Packer said there’s nothing concerning about any of this — the near-par prices Blue Owl received for the sales are indicative of underlying asset quality, he insisted.

When pressed about selling loans to what David Faber called “a wholly-owned subsidiary,” Packer appeared to dispute that characterization of Kuvare, but ultimately refused to confirm the identities of the loan buyers.

When Jim Cramer said “It’s very key,” referring to the Kuvare connection, Packer said, “It’s not key.” “We sold it to four institutions, they all bought the same amount, at the same time, at the same price,” he told Cramer. “The fact that one of the four might be part of our insurance business — how is it reasonable that that would undermine the other 75% of the sales?”

---