39% of Investors Asked to Leave. The Fund Let Out 5%.

In the third quarter of this year, investors in one Blue Owl fund submitted requests to withdraw 39 percent of their money.

The fund let out 5 percent.

That was not a failure or an emergency measure. It was the contract working exactly as written. The fund’s rules cap quarterly redemptions at 5 percent of assets, and when more people ask to leave than that, everyone is scaled back proportionally.

Apollo did the same thing. Its Debt Solutions fund has capped redemptions at 5 percent for three consecutive quarters after withdrawal requests reached 14.7 percent. Earlier this year, when 11.2 percent of investors in that fund asked for their money, each one received roughly 45 cents for every dollar they had requested.

Across the major private credit funds in the first quarter of 2026, investors submitted approximately $20 billion in redemption requests. Around half was honoured.

And in February, Blue Owl stopped pretending. It scrapped the quarterly redemption window on its $1.6 billion OBDC II fund entirely and moved the vehicle into wind-down.

On the twelfth of October, the chair of the Financial Stability Board wrote to the G20 finance ministers and named private credit as one of the three greatest threats to global financial stability, alongside artificial intelligence risk and sovereign bond volatility.

The market is worth roughly $1.7 trillion. In 2012 it was $250 billion.

Here is the full file.

After 2008, regulators did what everyone agreed they should do. They made bank lending safer and more expensive. Capital requirements rose. Banks had to hold more equity against risky loans, mark positions to market, submit to stress tests, and open their books to supervisors.

The intended effect was achieved. Banks became considerably safer.

The unintended effect was that the lending did not stop. It moved.

Companies still needed to borrow. The loans the banks could no longer profitably make were picked up by investment funds instead. Those funds are not banks. They have no capital requirements, no primary regulator, no deposit insurance, no stress tests, and no obligation to mark their assets to any public price.

This is the same mechanism this newsletter has described in a dozen different settings. Regulation does not eliminate an activity. It relocates it to wherever the rules do not reach.

The activity in this case is corporate lending, and it now sits in a $1.7 trillion pool outside the banking system. The top ten managers control about $1.05 trillion of it, roughly 62 percent. Blackstone holds around $370 billion, Apollo around $350 billion, KKR about $110 billion, Blue Owl about $105 billion.

For most of the last decade this looked like a straightforward success. Returns were strong, defaults were low, and the funds grew at better than 15 percent a year.

Then rates rose, and the structure started to be tested.

This is the single most important thing to understand about private credit, and it is the reason regulators keep using the word opacity.

If you own a corporate bond, you know what it is worth. It trades. Hundreds of participants quote prices all day, and when the borrower’s prospects deteriorate, the price falls immediately and publicly. The market tells you something you may not want to hear, whether you ask or not.

A private credit loan does not trade. There is no screen, no quote, no secondary market of any depth. The loan was negotiated privately between one fund and one company, and it sits on the books until it matures.

So someone has to decide what it is worth.

That someone is the fund manager.

Now look at the incentive. The manager is paid a management fee calculated as a percentage of assets under management, plus performance fees based on returns. The number he reports as the value of the loans directly determines the number he is paid.

He is not committing fraud by marking a loan at par. He is making a judgement, using a model, with assumptions he selects, about an asset no one else prices. There is a very large range of defensible answers, and his compensation sits at one end of it.

The consequences of this show up in the default data.

Moody’s examined middle-market CLO defaults for the twelve months ending June 2026. The headline rate was 4 percent. When Moody’s included borrowers whose loans had been amended to allow payment in kind, meaning the borrower stopped paying cash interest and instead added it to the principal, and borrowers whose maturities had simply been extended, the true rate was 5.8 percent.

Understand what those two amendments actually are. A borrower who can no longer pay interest in cash is in trouble. A borrower who cannot repay at maturity is in trouble. In a public market, both would be priced as distressed immediately. In private credit, both can be handled by rewriting the document, and the loan continues to be reported as performing.

Morningstar DBRS reported this month that one in four actively rated private credit borrowers is under some form of operating pressure, and that covenant relief has hit a five-year high.

A default is an event that forces recognition of a loss. The entire architecture of private credit is unusually good at preventing that event from occurring on paper.

For most of its history, private credit was an institutional product. Pension funds and insurers committed capital for ten years and could not withdraw it. That structure matched the asset. The loans took years to mature, and the money was locked up for years.

Then the industry found a much larger pool of capital, and to reach it, the funds changed shape.

The vehicles sold to individual investors are called non-traded business development companies, or interval funds, or more generally semi-liquid vehicles. They hold the same illiquid loans. But they offer investors the ability to redeem quarterly, usually capped at 5 percent of fund assets per quarter.

Read that carefully, because it is the whole problem in one sentence.

The fund holds assets that cannot be sold quickly. It offers investors the right to leave quickly.

That is a liquidity mismatch, and it is the oldest structural flaw in finance. It is precisely what a bank is: deposits payable on demand, backed by loans that take years to repay. Banks survive this because they have deposit insurance, a central bank that will lend to them in a crisis, and capital rules designed around exactly this risk.

A private credit fund has none of those things. It has a gate.

And the gate creates a problem of its own, which the FSB identified explicitly in its May report. If you believe a fund may restrict withdrawals, the rational move is to request your money before everyone else does. Which makes the restriction more likely. Which gives everyone else the same incentive.

That dynamic has a name. It is a run. The 5 percent cap does not prevent it. It just means the run happens in slow motion, quarter by quarter, with the people who moved first getting more of their money back than the people who waited.

Blue Owl’s OTIC fund received requests for 39 percent of its assets in a single quarter. That is not an investor base rebalancing a portfolio. That is a queue.

Two further threads make this more than a story about one asset class.

The first is the banks, who were supposed to have been taken out of the risk. The FSB estimates bank exposure to private credit at between $220 billion and $500 billion, through lending to the funds themselves, credit facilities and financing lines. In March, JPMorgan marked down collateral on a private credit loan, and the New York Fed subsequently reviewed the exposures of JPMorgan, Morgan Stanley, Wells Fargo and Barclays.

The lending moved outside the banking system. The banks lent to the thing that moved outside the banking system.

The second is what the money was lent to. Private credit funds have heavy exposure to software companies, estimated at 20 to 25 percent of some portfolios, because software businesses have recurring revenue and few hard assets, which suits a lender charging a premium rate. That concentration was built on the assumption that software revenues are durable.

Carlyle warned this year that private credit financing of artificial intelligence infrastructure is dangerously concentrated and that $1 trillion of exposure could prove to be the industry’s largest mistake. Last week, CoreWeave’s 2031 bonds traded at yields near 16 percent, and the semiconductor index shed $1.6 trillion in six sessions.

A fund that lent heavily to software companies on the assumption of durable recurring revenue, in a year when AI is disrupting software business models, is holding a portfolio whose central assumption is being tested in public while the loans themselves are being valued in private.

Meanwhile the money keeps coming. Direct lending funds raised a record $129 billion in the first nine months of 2026, up 35 percent year on year, while actual lending volume in the third quarter fell 28 percent.

More capital arriving. Fewer good places to put it. The pressure to deploy does not improve lending standards.

Three conclusions.

An asset without a market price is not an asset without risk. It is an asset whose risk has not been measured yet. Investors are consistently drawn to private assets partly because the reported returns look smooth. Public markets fall 20 percent and you watch it happen in your account. Private credit marks do not move that way, and that stability is often described as a benefit. It is not a benefit. It is the absence of measurement. The volatility you cannot see is still there, and it arrives all at once, at the moment someone is finally forced to sell. Whenever an investment is marketed on the smoothness of its returns, check who is calculating them and what they are paid.

Read the redemption terms before the return projections. The most important document in any fund is not the performance history. It is the section explaining how and when you can get your money out, and under what conditions that right can be suspended. In private credit, these terms were disclosed in full, in writing, to every investor. The 5 percent quarterly cap was never a secret. Thousands of people read past it because the yield was attractive, and discovered what it meant in the specific quarter when they needed their money, which is also the quarter when everyone else needs theirs. Liquidity is only ever tested at the worst possible moment. That is the definition of liquidity risk.

Financial risk does not disappear when it is regulated. It relocates. This is the pattern underneath every issue of this newsletter. Basel III made bank lending safer, and the lending moved to funds with no capital rules. FATCA ended Swiss banking secrecy, and the money moved to South Dakota. Enforcement closed Nauru, and the structures moved elsewhere. The same reflex applies to what you are reading right now about private credit: whatever constraints follow this cycle, the activity will reappear somewhere with fewer rules and a less familiar name. The question to ask of any safe-looking system is never whether the risk has been eliminated. It is where the risk has gone, and who is holding it now.

39%. The share of one Blue Owl fund that investors asked to withdraw in a single quarter.

5%. The share the fund was contractually obliged to let out.

$1,700,000,000,000. The size of a market that has no public prices, no deposit insurance, no capital requirements and no primary regulator, and that was named this month by the Financial Stability Board as one of the three greatest threats to global financial stability.

In 2012 that market was $250 billion. The regulation that made banks safer did not remove the lending. It moved it somewhere nobody is required to look.

That is not a conspiracy theory. That is the redemption notice.

The Dark Money Letter is published every Wednesday and Sunday. → thedarkmoneyletter.com

Sources