Pebbles II

10,210 Holdings. 424 Companies in Multiple Funds. 33 Cents on the Dollar.

On February 9, Apollo’s co-president was asked about stress in private credit. He called it “pebbles.” Forty-two days later, investors who asked for their money back got 45 cents on the dollar.

Private Credit Stress

The first Pebbles piece went out on March 26. Here is what happened in the six days since.

I said FSK was the most vulnerable BDC in the system. The day after publication, FSK traded below $10 — the strike on my April puts — for the first time since the Moody’s downgrade. Moody’s confirmed what the data already showed: FSK’s non-accrual rate hit 5.5% at amortized cost, one of the highest among rated BDCs. PIK income is 14.7% of FSK’s total investment income. The peer median is 6.3%. FSK is running at more than double the industry average on the metric that Fitch says drives 60% of defaults.

I said Blue Owl was broken. OWL announced $1.4 billion in forced liquidations. Short interest reached 17.9%.

I said FSOC would eventually have to respond. On March 25 — the day before publication — FSOC voted to publish new guidance on nonbank financial company designations, specifically targeting the structure where the same firm originates, manages, and values the same assets. That is the exact structure I described.

What I got wrong: I said OWL was down 66 percent year to date. The stock hit 42 percent down year to date at its low and is currently down roughly 39 percent, with a 65 percent decline from its 2024 highs. I conflated the timeframes. Corrected here.

What hasn’t happened yet: the cascade. FSK stabilized near $10. BIZD is flat. The BDC sector has not crashed. The thesis requires Q1 10-Q filings in May to confirm or deny whether the PIK-to-default pipeline is accelerating. Six days is not enough time to resolve a structural thesis. The catalysts are on the calendar, not in the rearview mirror.

Nothing in the last six days contradicts the thesis. Nothing has confirmed it either. That is why I went deeper.

I wanted to understand how the contagion actually works. Not the narrative version — not “markets are volatile” or “credit cycles turn.” The mechanical version. Which companies are failing. Which funds hold them. And what happens when the same company sits inside five, six, seven funds at once and one of them marks it down.

So I did something I haven’t seen anyone else do. I downloaded the actual SEC filings — the 10-Ks, the Schedule of Investments — from eighteen BDC funds. Not summaries. Not fund fact sheets. The raw HTML filings from EDGAR, each one hundreds of pages long, formatted in nested tables that were clearly never designed to be read by anyone other than auditors and compliance attorneys. I extracted every single holding: company name, fair value, cost basis, interest rate, seniority, and two flags that matter more than anything else in private credit right now — whether the borrower is on PIK, and whether it’s in non-accrual.

This took weeks. The filings don’t use consistent naming conventions. The same company appears under different legal entity names in different funds. Abbreviations vary. Parent and subsidiary relationships aren’t flagged. I had to normalize over 10,000 entries by hand and by algorithm before the cross-fund map became legible.

The result: 10,210 individual holdings across 4,024 unique portfolio companies. The largest company-level map of the private credit system that I’m aware of outside the funds themselves.

I want to be clear about what this is and what it isn’t. This is a map drawn from public filings — not proprietary data, not leaked documents, not insider information. Every number in this analysis comes from documents that any investor could access on EDGAR. The difference is that nobody did. The filings are public. The work to connect them is not trivial. And the picture that emerges when you do connect them is one that no single fund’s disclosure reveals on its own.

What I found is worse than the fund-level analysis suggested.

Apollo’s own BDC — the fund managed by the firm whose co-president dismissed the stress as pebbles — has 46 companies in non-accrual. That’s the most of any fund in the entire universe I analyzed. Not second most. Not tied for most. The most, by a wide margin. The firm that told investors the stress was manageable is managing the portfolio with the most defaults.

FS KKR Capital has 24.4% of its entire portfolio on payment-in-kind. That means nearly one in four of FSK’s borrowers cannot pay cash interest. They are paying with IOUs — adding the interest to the principal, increasing the debt load on companies that already can’t service what they owe.

This is not a fringe concern. Fitch reported that PIK interest deferrals drove 60% of private credit defaults in the twelve-month period through January 2026. Not credit deterioration broadly. Not economic weakness. PIK deferrals specifically — the mechanism I’m describing here — caused more than half of all defaults. The private credit default rate hit 5.8% on a trailing-twelve-month basis through January 2026, up from 5.6% in December. The direction is clear.

Bloomberg reported in October 2025 that private credit’s rising pile of “bad PIK” points to default woes. TCW’s analysis describes PIK as creating a “shadow default rate” that masks underlying borrower stress. S&P Global and Fitch have both flagged the systemic risk at the industry level, not just at individual names.

FSK has 83 PIK holdings. At Fitch’s observed 60% conversion rate, the implied default pipeline from FSK’s PIK book alone is staggering. Even using a more conservative estimate — the 15 to 25 percent range cited by industry analysts over a 24-month horizon — the math says 12 to 20 of those become defaults over the next two years. And FSK is one fund.

I want to pause on what that means. One in four borrowers in FSK’s portfolio cannot pay cash interest right now. More than half of all private credit defaults are PIK-driven. The default rate is 5.8% and rising. And the firm managing this portfolio — KKR — is simultaneously raising the next vintage of private credit funds, promising institutional investors “attractive risk-adjusted returns.”

The returns look attractive because the PIK income gets booked as revenue. The risk gets deferred to the next quarter. As long as the defaults haven’t arrived yet, the numbers look fine. This is the shadow default rate that TCW described. It is not a metaphor. It is an accounting structure that converts future losses into present income.

The contagion mechanism.

The private credit industry presents itself as diversified. You own FSK. I own ARCC. She owns GBDC. We each own a different fund with a different manager and a different portfolio. Except we don’t.

After normalizing the company names across all eighteen funds — the same borrower appears as “HS Purchaser, LLC” in one filing and “Help/Systems Holdings” in another — I identified 424 companies held by two or more funds simultaneously. Thirty-nine companies are held by three or more. Four companies sit inside five or more funds at the same time.

The diversification that investors believe they own is, in many cases, an illusion created by different fund wrappers around the same underlying loans.

Take Spotless Brands. It’s a car wash chain — 220-plus locations across 13 states, roughly 3,900 employees, backed by Access Holdings. The kind of company most investors have never heard of. It just upsized a $958 million credit facility through Carlyle, which suggests growth, not distress. The company is currently in expansion mode.

And yet Spotless Brands sits inside seven BDC funds simultaneously: BXSL, CGBD, FSK, GBDC, GSBD, OBDC, and ORCC. Seven different fund managers hold the same car wash company. Seven different investor bases are exposed to the same credit. If Spotless deteriorates — if the consumer pulls back, if unit economics shift, if the debt load from that $958 million facility becomes unserviceable — seven fund managers have to write it down in their next quarterly report. Seven sets of investors see their NAV decline. Seven analyst models get revised.

From one company.

A car wash chain that most BDC investors have never visited, whose name they’ve never seen, whose debt they didn’t know they owned. Spotless Brands is not in distress today. It’s growing. It just upsized its credit facility. The point is not that Spotless is about to fail. The point is that when any company held by seven funds simultaneously does fail, the blast radius is seven times wider than the market expects. And there are 424 companies in the overlap network. Four of them are in five or more funds. The system is wired for contagion. It just needs a spark.

Kaseya — an IT management company that was carrying $4 billion in debt against $219 million in revenue — appeared in six funds in our most recent filing data. A caveat here: Kaseya refinanced $3.3 billion into a broadly syndicated loan in February 2025, which may have eliminated $1.9 billion from BDC books. If Kaseya still appears in our parsed data, it could reflect the lag between actual portfolio changes and the filing dates of the 10-Ks I extracted. I flag this because accuracy matters more than drama. The point is not one company — it’s the structure. Fullsteam Operations and Smarsh are in five funds each. These aren’t obscure names buried in a footnote. These are the nodes of a contagion network that nobody outside the funds is mapping.

But the most revealing finding isn’t the overlap. It’s the overlap that isn’t overlap at all.

GBDC and GSBD — Golub Capital BDC and Goldman Sachs BDC — share 192 identical companies. Their return correlation is 0.79. These are not two diversified funds. They are functionally the same portfolio — two firms co-lending on the same deals, charging two management fees, reporting to two sets of investors who believe they own something different. The same is true of CGBD and ORCC — Carlyle and Blue Owl — which share 128 identical companies.

An investor who owns GBDC and GSBD for “diversification” owns the same risk twice and pays two fees for the privilege. The advisor who recommended both funds as part of a “diversified income allocation” sold the illusion. The fee structure rewards it. The filings enable it, because nobody reads the Schedule of Investments line by line and cross-references across funds.

Nobody except me.

Think about what this means for institutional portfolio construction. A pension fund allocates to three BDC funds for diversification. Their consultant models uncorrelated return streams. But if those three funds share 100 or 200 of the same underlying companies, the “diversification” is a fiction. The portfolio is concentrated in ways that no allocation model captures, because the allocation model uses fund-level data and the concentration exists at the company level. The risk is invisible to the tools designed to measure it.

This is not a failure of due diligence. It is a failure of disclosure architecture. The SEC requires each BDC to file its own Schedule of Investments. Nobody requires the cross-fund comparison. The data exists in eighteen separate documents that were never designed to be read together.

Then there’s Pluralsight.

Pluralsight was a technology education platform — the kind of company that private equity loves to acquire, load with debt, and extract dividends from. Vista Equity Partners bought it in 2021. By August 2024, $803 million of Pluralsight’s debt was placed on non-accrual across multiple BDCs. The lenders — Blue Owl, Ares, Goldman Sachs, Oaktree — took 100% ownership of the company through a debt-for-equity swap. Vista’s equity was marked to zero. Four billion dollars of Vista’s investment, gone.

This is the poster child for the PIK-to-default pipeline. A company acquired by one of the most sophisticated private equity firms in the world. Debt structured by the largest private credit platforms. Held across multiple BDC funds. And when it failed, the losses cascaded across every fund that held it, the PE sponsor was wiped out, and the lenders ended up owning a company they never wanted to operate.

Pluralsight is not an exception. It is a preview. The PIK-to-non-accrual-to-lender-ownership pipeline is now the dominant resolution mechanism in private credit. Fitch’s data confirms it: 60% of defaults are PIK-driven. The lenders end up owning the companies. The companies were not built to be owned by lenders. The lenders were not built to be operators. And the investors who bought the fund for 10% yield end up owning equity in a distressed technology company that nobody wants to buy.

This is what “flexible capital solutions” looks like at the end of the cycle.

These aren’t hedge fund investors taking calculated bets. The pension funds buying BDC exposure and lending through the same platforms are fiduciaries investing on behalf of people who have never heard of Spotless Brands or Pluralsight. Teachers in Ontario whose retirement runs through CPP. Firefighters in the Netherlands whose pension allocated to private credit for “safe income.” Australian workers whose super fund chose direct lending as a bond alternative. When Spotless Brands defaults across seven funds, the markdown hits retirement accounts in Toronto, Amsterdam, and Sydney — accounts whose owners don’t know what a BDC is, have never read a Schedule of Investments, and have no mechanism to get out before the quarterly marks arrive.

The chain looks like this: a car wash company in Texas borrows from a fund in New York, which is owned by a platform in Connecticut, which manages money for a pension in Ontario, which pays benefits to a retired teacher in Thunder Bay. Five links. Five jurisdictions. Zero transparency from end to end. The teacher sees a quarterly pension statement. The statement shows a return. The return includes a markdown she can’t trace to a car wash she’s never visited in a state she’s never been to.

The distance between the borrower and the beneficiary is the feature that makes private credit opaque. It’s also what makes the contagion invisible until it isn’t.

The PIK pipeline.

There are 1,020 holdings across the eighteen funds currently on payment-in-kind. This is the leading indicator that almost nobody tracks at the system level.

PIK means the borrower has negotiated — or been forced into — an arrangement where they don’t pay cash interest. Instead, the interest gets added to the principal. The debt grows. The company’s balance sheet deteriorates. And the fund reports the holding at a value that assumes eventual repayment of a debt that is getting larger every quarter.

It’s a game of musical chairs played with compound interest. As long as the music plays — as long as rates stay manageable and revenue holds — the PIK borrower survives. When the music stops, the debt has grown so large that even a modest decline in revenue triggers a covenant breach. The borrower doesn’t gradually weaken. It snaps.

Fitch’s data makes the severity clear: PIK interest deferrals drove 60% of private credit defaults in the twelve-month period through January 2026. Not poor management. Not competitive disruption. The mechanical accumulation of deferred interest — the thing the fund managers describe as “flexible capital solutions” — is the primary default trigger. The industry’s signature product feature is its primary failure mode.

And when those defaults arrive, the recovery is worse than most investors expect.

Bloomberg reported in March 2024 that private credit direct loans recover 33 cents on the dollar after default. Syndicated loans — the traditional leveraged loan market that private credit was supposed to replace — recover 52 cents. Private credit, despite its senior position in the capital structure, despite the covenant protections, despite the direct lender relationship, recovers 36% less than the market it claims to be superior to.

The reason is structural. Syndicated loans have liquid secondary markets. When a borrower defaults, the debt trades, price discovery happens, and workouts proceed with market discipline. Private credit loans have no secondary market. The lender is often the only holder. Workouts happen behind closed doors with no price transparency. And the lender — who is also the fund manager, who earns fees on AUM — has every incentive to delay the markdown as long as possible. By the time the write-down arrives, the recovery value has already deteriorated.

Thirty-three cents. On senior secured debt. With covenants.

Applied to the current 1,020 PIK holdings, the historical conversion rates imply 150 to 250 additional defaults entering the system over the next two years. These defaults will not arrive all at once. They will arrive in clusters — because the same companies are held by multiple funds, and when one fund marks a company down, the auditors at the other funds are forced to revisit their own marks.

Seventeen companies in the database are already in both states simultaneously — on PIK and in non-accrual. These are companies actively defaulting while their interest obligations continue to compound. They include names like Pluralsight — where Vista lost $4 billion and the Blue Owl-led lender group took ownership — held by GBDC and GSBD. ATX Networks, on PIK in FSK, GBDC, and GSBD — three funds exposed to the same deteriorating credit.

The contagion doesn’t require a crisis. It requires one quarterly report. One auditor at one fund deciding that one shared company needs to be written down. The other funds can disagree — for one quarter. Maybe two. But the audit trail is public. The markdown is in the filing. And the other funds’ auditors will ask the question: “Why is your mark different from theirs?”

That question is the fuse.

Not every fund is exposed. BlackRock TCP Capital (TCPC) has only 3 PIK holdings out of 131 — a 2.3 percent PIK rate versus FSK’s 24.4 percent. Zero non-accruals. In the cross-fund overlap analysis, 95 of TCPC’s 97 unique companies appear in no other fund in the database. TCPC is the most isolated, least contagion-exposed BDC in the entire universe I parsed.

Good underwriting still exists. It’s just the exception. The rule is what FSK and Apollo’s fund look like — heavy PIK books, significant non-accruals, and deep overlap with the rest of the system. TCPC proves that the problem is not private credit as an asset class. The problem is how most of the industry practices it — with concentrated sector bets, overlapping co-investments, PIK as a default management tool, and fee structures that reward asset growth over credit discipline.

The difference between 2.3% PIK and 24.4% PIK is not a rounding error. It is a philosophy. And right now, FSK’s philosophy is the industry standard.

Your pension fund owns the same borrowers.

The Bank of Canada’s governor, Tiff Macklem, said it publicly on March 4: “Risks have not disappeared — they’ve migrated. Private credit is more opaque. Leverage can build quietly and then unwind very quickly.”

He’s right.

CPP Investments — Canada’s largest pension fund, managing C$777 billion in retirement savings for 22 million Canadians — owns Antares Capital, one of the largest middle-market direct lenders in the United States with approximately $90 billion in assets under management. Antares competes directly with FSK and ARCC for the same borrowers. When I cross-referenced Antares’ known portfolio against the BDC filings, I found 70 overlapping companies. Finastra. Athenahealth. Global Medical Response. Riverbed. Netsmart. Inovalon. Seventy companies that sit in both the Canadian pension system and the US BDC system that is currently gating.

OMERS — the Ontario Municipal Employees Retirement System — has 14 percent of its portfolio allocated to private credit. The four largest Canadian pension funds collectively manage approximately one trillion Canadian dollars. They have been increasing their private credit allocations for years. And the Bank of Canada governor is now warning publicly that the risks have migrated into exactly the structures they hold.

OSFI — Canada’s banking regulator — flagged “growth of private credit and its interconnectedness with regulated institutions” as a top-four risk. Thirty-six percent of Canadian mortgages are renewing by end of 2026, many at rates two to three times higher than their pandemic-era originations. The housing-credit nexus is tightening at the same time that private credit stress is building.

The Canadian pension system is not a bystander in the private credit story. It is a participant. Through Antares alone, CPP is lending to the same companies that are defaulting in the BDC system. When I found 70 overlapping names, I was not looking at two separate pools of risk. I was looking at the same pool, accessed from two different doors. The defaults in one system are, mechanically, the defaults in the other. The only difference is reporting lag.

But Canada is not the only country exposed. It’s not even the most exposed. It’s just the one with the governor willing to say it out loud.

Japan’s life insurance industry is quietly building one of the largest foreign allocations to US private credit in the world. Sumitomo Life is allocating $1.9 billion to private credit this fiscal year. Nippon Life committed $3.25 billion to TCW Alternative Credit Strategies and is increasing its ownership stake. Dai-ichi Life is shifting its portfolio toward private assets under new regulatory rules. Japanese life insurers collectively manage trillions of dollars in assets. Even at 1 to 3 percent allocation to private credit, the implied exposure is $26 to $78 billion. These institutions need yield after FX hedging costs eat into their traditional bond returns — private credit promises 8 to 10 percent. The same private credit that is currently gating. Nobody in Tokyo is reading FSK’s Schedule of Investments.

Australia is further along the curve. AustralianSuper — the country’s largest superannuation fund — is tripling its private credit allocation from A$5 billion to A$15 billion. It allocated $2.3 billion to a single fund managed by Churchill Asset Management. Australian super funds offer daily member switching. The underlying private credit has 3-to-7-year lockups. Nearly 18 percent of pension funds surveyed do not have enough liquidity to withstand an adverse scenario. When a member switches out of a “balanced” option that holds illiquid credit, the fund has to sell something liquid to meet the redemption. The mismatch is structural and it is growing — by a factor of three. Daily liquidity on illiquid assets. That is not a risk model. That is a lie.

South Korea’s National Pension Service committed a record 1.5 trillion won — $1.1 billion — to alternative investments in 2024 and is targeting over 20 percent of its portfolio in alternatives by 2026. NPS manages $793 billion. They opened a New York City office specifically for US credit. The won has been depreciating against the dollar, which means the private credit returns look better on paper in won terms — until the risk-off trade reverses the currency and the losses compound.

The United Kingdom never fully recovered from 2022. After the gilt crisis nearly collapsed the pension system, UK schemes reallocated into private credit as a “safer” alternative to leveraged LDI strategies. USS allocates roughly 20 percent to private credit. Railpen roughly 18 percent. BT Pension Scheme roughly 15 percent. Nearly 18 percent of UK pension fund managers say they lack sufficient liquidity to handle adverse scenarios given their private credit exposure. The Bank of England’s Financial Stability Report has flagged the risk. Parliament held hearings. Nobody changed anything.

The Netherlands may be the most politically combustible of all. ABP — Europe’s largest pension fund, with EUR530 billion in assets managed by APG — cut its US Treasury holdings by $12 billion in 2025 while APG has been expanding into US credit markets, including infrastructure debt and direct lending. PFZW, with EUR305 billion managed by PGGM, follows a similar strategy. De Nederlandsche Bank, the Dutch central bank, has warned about liquidity and valuation risks in private credit. Dutch pension assets total hundreds of billions of euros, a meaningful fraction of which is now allocated to US credit strategies.

Here is what makes the Netherlands different from every other country in this analysis: on January 1, 2026, the Dutch pension system began its historic transition from defined benefit to defined contribution. Every loss is now individually visible to every participant for the first time. A private credit markdown that would have been absorbed quietly inside a DB fund is now a line item on a personal retirement statement. The political visibility is unprecedented. A Dutch nurse who has never heard of Spotless Brands will see the loss on her phone.

I ran TonalityIQ on all six major Canadian banks. CIBC scored the highest on hedging language related to private credit and commercial real estate — the most linguistically defensive about credit exposure in the group. The market hasn’t noticed.

Brookfield Asset Management’s TonalityIQ ratio is 10.86 — the highest of any company in our 636-ticker universe. Higher than any SaaS company, any bank, any tech firm. Brookfield manages over $900 billion in assets across real estate, infrastructure, and private credit globally. When the management team at that scale scores higher on reassurance language than any other public company we track, the signal is not about one fund. It’s about the entire alternative asset class.

The bigger the manager, the louder the reassurance, the worse the underlying data. Brookfield’s score is not just high. It is the highest we have ever measured — across SaaS, banking, tech, energy, every sector we track. When the management team running a $900 billion platform scores higher on reassurance language than every other public company in our universe, the signal is not about one fund. It is performance masquerading as confidence.

On March 25, 2026, the Financial Stability Oversight Council met and discussed private credit. The public readout was mild — “monitoring market developments.” Bureaucratic language designed to communicate nothing. But the fact that it made the agenda is significant. FSOC does not convene to discuss things that are fine. It convenes to discuss things that might not be. The last time private credit appeared on an FSOC agenda at this level of prominence was never. This is new territory.

The confirmed commitments I traced across these regions total over $10 billion from named institutions alone. The full foreign institutional exposure to US private credit spans five continents and hundreds of billions of dollars in retirement assets.

The gating is already real. $4.6 billion is confirmed trapped in gated redemptions across the private credit system. Investors submitted $10 to $13 billion in redemption requests across twelve or more funds in Q1 alone. The gap between what investors want to withdraw and what funds will allow them to withdraw is widening every quarter. And these are the investors sophisticated enough to have requested redemptions early. The retail BDC investor — the one who bought FSK for the dividend yield and holds it in an IRA — hasn’t even started selling yet. That wave arrives when the Q1 10-Qs hit and the financial media translates the filings into headlines.

The “$100 billion” figure that circulated in some reports refers to market capitalization evaporated from publicly traded private equity and credit managers — the stock price decline at Apollo, Blue Owl, Ares, and others. That number is real, but it describes equity market sentiment, not gated capital. The $4.6 billion in actual trapped redemptions is smaller but more dangerous, because it represents real money that real investors cannot access. It is the difference between a stock price declining and a bank not letting you withdraw.

The full analysis — three scenarios for how the PIK-to-default pipeline transmits, the trades, and the catalyst timeline — continues below for paid subscribers.

---