Newsletter · · Ashutosh Agarwal

Blue Owl's Default Rate Hits a Five Year High as Big Manager Credit Strains - The Private Credit Boom (and Cracks) - Week of August 26, 2026

The Private Credit Boom (and Cracks) for the week of August 19 to August 26, 2026. Podcast synthesis on the first hard Q2 default figures (Blue Owl at 2.8%, a five-year high, with Ares, Blackstone and Golub at post-2021 worsts), Blue Owl's 80% ownership of the venture financing Meta's Hyperion data center, and the structural bear case on layered leverage and gated funds.

The Private Credit Boom (and Cracks)

Week of August 19 to August 26, 2026: Blue Owl's Default Rate Hits a Five Year High as Big Manager Credit Strains

Private credit, direct lending, BDCs, and the big managers, what the podcasts said this week, and why it matters for your book.

Quick vocabulary, in plain English (skip if you know it): Private credit means loans made by investment firms instead of banks. A BDC (business development company) is a publicly traded fund that makes these loans, you can buy it like a stock (Ares Capital, ticker ARCC, is the biggest). A non-accrual is a loan that has stopped paying, so the fund stops booking interest on it, the cleanest early sign of trouble. Gating is when a fund tells investors they can't pull their money out right now. ABF (asset-based finance) means lending against hard collateral, increasingly, the chips and buildings inside AI data centers.

TL;DR

  • The week turned on a single number. A Wall Street Journal analysis, picked up on two podcasts, put Blue Owl's fund default rate at 2.8% in the second quarter, its highest in at least five years, with Ares, Blackstone, Blue Owl and Golub all reporting their worst default levels since 2021.
  • The data-center money machine got a peek under the hood. New reporting detailed how Blue Owl owns 80% of the venture financing Meta's giant "Hyperion" data center, backed by $27 billion of bonds and a $12.3 billion Meta lease, a textbook example of private credit muscling into AI infrastructure.
  • Read the specifics with the source mix in mind. Every voice this week was a pundit or a journalist rather than a manager speaking to their own live numbers, so the figures below are best treated as "per the Journal," not as company disclosure.

What's new

1. The first hard Q2 default numbers finally surfaced, and they're five-year highs

The single most useful thing said all week came secondhand. On the Money Metals' Weekly Market Wrap Podcast (episode: "How the Debt Black Hole Is Warping the Economy: Private Credit Edition," Aug 19), host Mike Meharry, a precious-metals commentator rather than an insider, walked through a Wall Street Journal analysis of the biggest managers' second-quarter reports.

The headline: private credit funds run by Ares Management, Blackstone, Blue Owl and Golub Capital "reported loan defaults hitting the highest level since 2021." The most specific figure: "the default rate at Blue Owl Capital's fund hit 2.8% in Q2… the highest level in at least five years." Non-performing loans at other funds also hit five-year highs, above even the 2023 rate-hike scare. And zooming out, "According to Fitch Ratings, U.S. private credit default rate stood at 6% at the end of May," with 14 defaults in May alone.

Where's the pain? Per the same reporting: healthcare, businesses squeezed by higher oil prices, industrials and business services. One named casualty, Loparex, a plastic-film maker, "recently defaulted on its private loan." The worry that keeps managers up at night is software: in many funds, "20% or more of the outstanding debt is owed by software companies," the group most exposed to being undercut by AI.

Why it matters: These are the first concrete second-quarter default figures to reach the podcast feed, and they name names: Blue Owl (OWL), Ares (ARES), Blackstone (BX), Golub (public BDC: GBDC). A 2.8% fund default rate is still low in absolute terms, but "highest in five years" is the direction that moves BDC discounts and, eventually, dividends. The catch: this came through a gold-focused show relaying the WSJ, not from any manager's mouth. Treat it as "per the Journal," not gospel.

2. Blue Owl's role in the AI build-out got specific: 80% of Meta's Hyperion, $27B of bonds

The Banker Next Door (episode: "Big Tech AI spending has gone off the rails!," Aug 20), hosted by Dr. Joseph Bergquist, unpacked a second Wall Street Journal piece, this one on the roughly $3 trillion of AI spending Big Tech has committed to but kept off its balance sheets. Buried in it is the clearest private-credit story of the week.

Meta's Hyperion data center in Louisiana, "about 1,700 football fields" in size, is owned through a special venture that is 20% Meta, 80% Blue Owl. As the host put it, quoting the reporting: "neither Hyperion nor the $27 billion in debt that's financing its construction show up on Meta's balance sheet. Funds managed by the Wall Street firm Blue Owl Capital own the majority of a joint venture that… owns the campus." The holding company that owns Blue Owl's stake "raised the construction financing in a bond sale," so bondholders, not Meta, put up the cash. Meta's job is to pay rent: an initial four-year lease starting in 2029, renewable up to 20 years, with an "aggregate initial lease commitment… about $12.3 billion," and a promise to make bondholders whole if it walks away early. All told, Meta disclosed $347 billion of obligations for leases that haven't even started yet as of June.

Bergquist didn't hide his view of Blue Owl, calling it "one of our friends that has been in the private credit mess" whose "private credit funds have been getting bludgeoned with bad loans and redemption requests."

Why it matters: This is the ABF and data-center origination theme made concrete. Blue Owl (OWL) isn't just lending around the edges of the AI boom, it owns the majority of the vehicle that owns one of the largest data centers being built, with a Meta lease as the backstop that makes the $27B of bonds money-good. It's a fat, long-duration income stream if Meta pays for 20 years, and a very different-looking asset if AI demand disappoints. Direct read to OWL; the whole "who finances the data centers" trade sits underneath it.

3. The bear's structural case, laid out cleanly (if without numbers you can bank)

The Competent Investor (episode: "Chris MacIntosh: Pension Funds and Investors Trapped in the Looming Debt Implosion," Aug 21) featured macro-bear Chris MacIntosh of Capitalist Exploits. Fair warning: these are his assertions, not sourced figures, and the episode wanders into unrelated conspiracy territory later, but the credit portion is a tidy version of the whole bear thesis.

His frame: three bubbles all hinge on interest rates, private credit, the "AI Mag7 equity bubble," and the "big daddy," the sovereign-debt bubble. On private credit specifically, he argues there's "layered leverage… running up to 20 times leverage on… illiquid assets." The time bomb, in his telling, is the refinancing wall: "much of that private credit was financed at rates three, 400, 500 basis points lower than today," so as loans come due and reprice "400 basis points or more higher," funds that lack fresh cash are "forced to sell… they're not at par… In many instances, the equity tranche is completely wiped out and even the debt's impaired." The result, he claims: "hundreds of these funds that are all gated," investors locked out of their own money.

And he pointed the finger at who ends up holding it: pension funds and insurers, courtesy of rule changes that opened these products to them. "I've spoken with some pension fund managers who don't even know what they bought," he said, structured vehicles yielding 12% that look "liquid until it's not."

Why it matters: No tradable number here, but it's the cleanest statement of the bear case: leverage stacking, a repricing wall as cheap-rate loans roll, gated retail vehicles, and pension and insurance contagion. Useful as the frame for the debate, not as a data point.

The debate: cycle-starting crack, or idiosyncratic noise?

The bear steel-man: The numbers are finally turning the wrong way at the same time across every big manager, Ares, Blackstone, Blue Owl, Golub all at their worst since 2021, Blue Owl at a five-year high, Fitch's asset-class default rate at 6%. That's not one bad loan; that's a trend. Layer in a refinancing wall (loans written at rates 3 to 5 points below today's, now repricing higher), a big slug of exposure to AI-threatened software borrowers, and end-buyers (pensions, insurers, retail) who don't fully understand what they own, and you have the setup for a slow unwind that "happens slowly, then all at once."

The bull steel-man: The week's figures came to the feed through two journalists' recaps relayed by a gold commentator and a macro pundit, plus a self-described bear with no verifiable numbers. A 2.8% fund default rate is still low. Defaults sit "below levels seen during… the pandemic and the oil price crash in 2015," and, per the same WSJ piece, "losses could abate if interest rates decline and economic activity remains robust." Blue Owl owning 80% of a data center backed by a 20-year Meta lease isn't a crack; it's exactly the durable, collateral-backed origination the asset class was built to do.

The pull-quote that captures the week, from Money Metals, relaying the Journal:

"the default rate at Blue Owl Capital's fund hit 2.8 percent in Q2… the highest level in at least five years."

Low number, ominous label. That tension is the whole week.

Stocks in play

Blue Owl Capital (OWL), the name that came up most.

  • Bull: Owns 80% of the venture financing Meta's Hyperion, with a $12.3B Meta lease backstopping $27B of bonds, a marquee, long-duration, collateral-backed income stream, and proof it's winning the data-center origination race.
  • Bear: Its fund's default rate hit a five-year high (2.8% in Q2), and a commentator flagged its funds "getting bludgeoned with bad loans and redemption requests." Concentrated bets on huge AI projects cut both ways.
  • Next catalyst: Any disclosure on non-accruals, redemptions, or the Hyperion financing; the next earnings print and NAV mark.

Ares Management (ARES), Blackstone (BX), Golub (public BDC: GBDC), named in the WSJ default roundup.

  • Bull: Defaults are rising off a low base and remain below crisis levels; scale and diversification are supposed to blunt idiosyncratic hits.
  • Bear: All flagged at their worst default levels since 2021, a synchronized deterioration, not a one-off.
  • Next catalyst: Confirmation (or not) of the WSJ's Q2 default read in the managers' own reporting; watch-list and non-accrual trends next quarter.

Meta Platforms (META), the counterparty, not a lender.

  • Bull: Off-balance-sheet structure lets it build enormous AI capacity while Blue Owl and bondholders carry the construction debt.
  • Bear: $347B of not-yet-started lease obligations and a make-bondholders-whole guarantee is real leverage in disguise; if AI demand disappoints, the tab still comes due.
  • Next catalyst: Capex and free-cash-flow trajectory (already turning negative at peers) and any change to the Hyperion lease terms.

Read-throughs

  • Public BDCs (ARCC, BXSL, OBDC): The WSJ default data covers their parents' credit (Ares to ARCC, Blackstone to BXSL, Blue Owl to OBDC). If Q2 defaults really are at five-year highs across the platforms, expect it to show up in non-accruals and NAV marks, the two things that move BDC discounts and dividend safety.
  • Insurance and pension partners: MacIntosh's claim that pensions and insurers "don't even know what they bought," and that these vehicles are "liquid until it's not," is the recurring contagion worry, the channel through which private-credit stress would reach ordinary savers. Unverified, but the theme won't go away.
  • Data-center and ABF borrowers: The Meta and Blue Owl Hyperion structure is the template for how AI infrastructure gets financed off-balance-sheet. Great origination volume for the managers; a concentrated bet on AI demand holding up for two decades.

What changed vs last week

Last week (Aug 12 to 19) was busy and bear-heavy, driven by named critics and an activist: Boaz Weinstein and Saba on discount tenders for non-traded BDCs, Chris Whalen on a "slow-moving train wreck" and insurance contagion (777 Partners), Robin Wigglesworth's balanced-but-worried journalist take, and a concrete new Blackstone, Wellington and Vanguard retail product.

Here's what actually moved this week:

  1. A hard number replaced the hand-waving. Last week's bear case was largely about liquidity and discounts (BDCs trading 20% to 35% below value, redemptions rising). This week we got the underlying credit data the critics were circling: Blue Owl 2.8% Q2 defaults (five-year high); Ares, Blackstone, Blue Owl and Golub at post-2021 highs; Fitch 6% at end-May. That's a step from "it feels bad" to "here are the defaults," via the WSJ.
  2. The Meta and Blue Owl data-center story got fleshed out. Last week's issue noted a "Meta Hyperion SPV with KKR + Blue Owl." This week's reporting gave the mechanics, Blue Owl 80% of the JV, $27B of bonds, $12.3B Meta initial lease, $347B of Meta uncommenced leases, but named only Blue Owl as the JV majority owner, with no mention of KKR. Worth flagging the discrepancy rather than reconciling it from memory.
  3. The activist and discount-tender thread went unaddressed. No Weinstein or Saba update landed this week.
  4. The source mix narrowed to pundits and journalists. Last week carried a range of named critics; this week was two WSJ recaps and one macro-bear, so the figures rest on secondhand reporting rather than manager disclosure.