March 17, 2026

The gates are closing.

Not metaphorically. Literally. In the last seven days, the largest names in private credit have locked their doors to investors trying to leave.

Cliffwater's $33 billion flagship fund: investors requested 14% back. The fund capped redemptions at 7%. Half of the money that wanted out is trapped inside.

BlackRock's $26 billion HPS lending fund: redemption requests hit 9.3%. The fund enforced its 5% quarterly cap. Roughly half denied.

Morgan Stanley's $7.6 billion North Haven fund: investors requested nearly 11%. The fund returned 45.8% of what was asked. More than half denied.

Blackstone's $82 billion BCRED: $6.5 billion in redemption requests (7.9% of the fund). Blackstone executives injected $400 million of their own capital to meet the requests. When the managers put their own money in to stop the run, that is not a sign of confidence. That is a sign of desperation dressed in a suit.

Blue Owl: paused quarterly redemptions entirely. Sold $1.4 billion in loan assets to North American pension funds at 99.7 cents on the dollar. The sale price tells you what the seller needed (liquidity, immediately) and what the buyer believed (those loans are not worth 99.7 cents; they are worth buying at that price because the seller is distressed).

These are not small firms with bad judgment. These are the five largest names in private credit. Together, these funds manage over $170 billion. The gating is not isolated. It is systemic.

For readers newer to private credit, a brief guide to the terms used in this issue:

Private credit / direct lending: Loans made by investment funds (not banks) to mid-size companies, typically at higher interest rates than bank loans. The $2.1 trillion market has grown rapidly since 2008 as banks pulled back from this lending.

BDC (Business Development Company): A publicly traded fund that makes private loans. BDC prices are set by the stock market (honest, real-time). This makes BDCs the window into what private credit is actually worth, because most private credit funds set their own values (see GP marks below).

GP marks: The values that fund managers (General Partners) assign to the loans in their own portfolios. These are the numbers investors see in their quarterly reports. The problem: GPs have an incentive to keep marks high, because their fees are calculated on reported value.

Gating: When a fund limits how much money investors can withdraw, typically to 5-7% per quarter. Designed as a stability mechanism; in practice, it means investors who want out cannot get out.

PIK (Payment-in-Kind): When a borrower cannot pay interest in cash, the interest is added to the loan balance instead. The loan stays "current" on paper, but the borrower has demonstrated it cannot service its debt. PIK masks distress as performance.

Divergence Overlay: A ScenarioPlan visual showing two trend lines: what governance structures report (based on GP marks and conventional metrics) vs. what is actually happening (based on market prices and honest data). The gap between the two lines IS the information suppression, measured in points.

FI (Fragility Index): A 1-10 composite score measuring an institution's vulnerability across four dimensions: correlation exposure, coupling density, assumption concentration, and adaptive capacity. Scores from 1 (antifragile) to 10 (critical fragility). The private credit institutional LP's reported FI is 4.7 ("moderate"); the actual FI is 7.5 ("fragile").

MPS (Machine Pressure Score): A 1-10 score measuring how close a "machine" (a self-reinforcing system of aligned incentives) is to breaking. Private credit's MPS is 7.5 and rising. Above 7.0 requires monthly monitoring.

Three things happened in the last 30 days that converted this from "stress" to "the confidence machine is activating."

First, the default data broke through the narrative ceiling. Fitch tracks two private credit default measures. The broader composite (covering their full tracked universe) hit 5.8% in January 2026, the highest since Fitch began tracking. The subset of their weakest-credit borrowers (the Privately Monitored Rating portfolio, smaller companies with sub-$25M earnings) hit 9.4% through January 2026, up from a record 9.2% for full-year 2025: nearly one in ten defaulted. Both are records. And both understate the real picture, because they count only formal defaults. The "true" stress rate, including companies that avoided formal default by toggling to PIK interest (paying lenders with more debt instead of cash), negotiating covenant waivers, or extending maturities under distress, is significantly higher. PIK usage now accounts for 8% of BDC investment income, up sharply. Selective defaults outpaced conventional defaults five to one in 2024 (S&P Global). Morgan Stanley, in a note published today (March 17), projects defaults will reach 8%, approaching COVID peak levels. Pimco, in a note published March 6, called for a "full-blown default cycle." When two of the largest fixed income managers in the world use the words "full-blown default cycle" and "approaching COVID peaks" in the same month, the narrative has changed.

Second, the marks are being challenged publicly. Glendon Capital Management, a distressed debt fund, informed its investors that private credit lenders including Blue Owl have "misrepresented" loss rates and are sitting on "larger losses than reported." Glendon documented specific cases where junior debt positions are marked at approximately 90 cents while senior debt in the same capital structure trades at 78 cents in public markets. Junior debt cannot be worth more than senior debt in the same company under any honest valuation methodology. The inversion is the proof. Blue Owl's response (that private marks are set at quarter-end and intraperiod public prices are not a valid comparator) is technically defensible and economically absurd. The marks are wrong, and now someone is saying it publicly.

Third, JPMorgan marked down the collateral. When the banks that provide leverage to private credit funds mark down the value of the loans held as collateral, the borrowing capacity of those funds shrinks. That is a margin call by another name. JPMorgan's decision to preemptively devalue software-related loans is the bank-to-private-credit transmission channel firing. The fund cannot borrow as much against the same portfolio. It must either post more collateral (which it may not have), sell assets (into a market where every fund is trying to sell), or reduce leverage (which reduces returns, which increases redemptions, which tightens the spiral).

This is the machine ScenarioPlan (our proprietary analytical model) mapped in TC #40 (Private Credit, v3.05, scored A). Five interlocking sub-machines: origination, mark opacity, maturity/refinancing, SaaSpocalypse transmission, and redemption/confidence.

The first four have been operating for two years. The fifth, the redemption/confidence machine, was described in TC #40 as "early-stage activation" with "Blue Owl redemption halt as the first major gating event."

That was seven days ago.

In seven days, the gating has spread from Blue Owl to Cliffwater to BlackRock to Morgan Stanley to Blackstone. Five of the largest funds in the industry, gating simultaneously. The confidence machine is no longer in early-stage activation. It is in full activation. The question is no longer whether the confidence machine will fire. It has fired. The question is whether the gating mechanisms (the 5% and 7% quarterly caps) can absorb the pressure, or whether the pressure overwhelms the gates.

The TC #40 Divergence Overlay showed a 2.8-point gap between what the GP marks show investors (FI 4.7, "moderate") and what the BDC parallel pricing reveals (FI 7.5, "fragile"). That gap has almost certainly widened in the last week. The GP marks have not moved. The market has.

Morgan Stanley's note today names the mechanism explicitly: "AI disruption will be a meaningful catalyst to drive defaults higher in direct lending." The connection between AI agents displacing software per-seat revenue (the SaaSpocalypse, TC #21) and private credit defaults (TC #40) is no longer a ScenarioPlan hypothesis. It is a Morgan Stanley research note published this morning.

The numbers: 15% to 25% of many private credit portfolios are concentrated in software firms. These are companies that borrowed against revenue growth assumptions that AI is now permanently invalidating. $46.9 billion in tech distressed debt. $25 billion in software loans trading below 80 cents. Apollo has already cut its software exposure from 20% to 10%. Blue Owl stock is down 41% year to date. Blackstone is down 31%.

The SaaSpocalypse is the catalyst. The underlying fragility (leverage, mark opacity, maturity concentration, extend-and-pretend masking cash flow distress through PIK) was already present. The SaaSpocalypse broke the revenue assumption for the most concentrated sector in the portfolio. But the stress is not limited to software. Healthcare services has the highest number of unique defaulters (Fitch). Consumer products default rates have more than doubled year-over-year to 12.8%. The software sector is where the break started; it is not where the break will end.

The Q2 redemption requests. Q1 requests were record-breaking across every major fund. If Q2 requests are equal or higher, the gating mechanisms will be tested beyond their design parameters. An interval fund that gates at 7% can survive one quarter of 14% requests. It cannot survive four consecutive quarters. The math forces one of two outcomes: either the requests decline (confidence returns) or the fund enters a multi-quarter drawdown where capital is returned only as loans mature or are sold at distressed prices. The second outcome is a slow-motion liquidation that the fund's marketing materials never mentioned.

The secondary market. Saba Capital (Boaz Weinstein) is launching tender offers to buy stakes in Blue Owl and other private credit vehicles. The secondary market is the pressure valve. If it functions (enough buyers at enough liquidity), the pressure dissipates without a cascade. If it cannot absorb the volume (which Raymond James' Haldea has warned is possible if "the floodgates completely open"), the gating becomes permanent and the confidence spiral accelerates.

The bank collateral channel. JPMorgan moved first. If Goldman Sachs, Bank of America, or Citigroup follow with their own collateral markdowns, the leverage available to private credit funds shrinks across the board, not just at JPMorgan-financed funds. That is the systemic channel: the banks provide the leverage that makes the returns possible; the banks can withdraw the leverage when the collateral deteriorates; and the withdrawal of leverage is the mechanism that converts a credit problem into a liquidity crisis.

The pension fund response. The perspective holder from TC #40 (institutional LP, state pension fund with 15-20% alternatives allocation) is the entity whose retirement beneficiaries are ultimately exposed. The pension fund's consultants recommended the private credit allocation. The pension fund's reports show the GP's marks, not the BDC's prices. The pension fund's investment committee is consuming information that the market has repriced. The Divergence Overlay gap is the distance between what the investment committee sees and what is actually happening to its beneficiaries' retirement capital. That distance widened this week.

What is risk (quantifiable): The maturity wall ($270B+ through 2028) is a calendar. The BDC maturity wall ($12.7B in 2026) is a calendar. The PIK rates (8% of income) are reported. The default rate (5.8% Fitch composite, 9.4% weakest-credit segment, 8% Morgan Stanley projection) is calculable within a range. The redemption request volumes (7-14% across major funds) are reported. The gating caps (5-7% quarterly) are contractual. These are calculable.

What is uncertainty (not quantifiable): Whether the redemption pressure abates or intensifies in Q2. Whether the secondary market can absorb the volume. Whether the bank collateral markdowns spread beyond JPMorgan. Whether a major private credit fund fails outright (no gating mechanism can survive indefinite excess redemptions). Whether the pension fund governance structures can process what is happening before the losses are locked in. Whether the confidence machine, once activated, can be deactivated without a systemic event. These are uncertain. The models that treat them as calculable are the models that produced the marks the market is now rejecting.

The private credit machine is breaking. Not "may break." Not "faces stress." Is breaking. The gates are the proof: you do not lock the doors when everyone wants to stay inside.

The break is slow (quarterly redemption cycles, not overnight depositor runs) but accelerating. The transmission channels are firing (SaaSpocalypse to software defaults to private credit, bank collateral markdowns to fund leverage, redemption requests to gating to confidence erosion). The information suppression mechanism (GP marks at 95-98 cents while the market prices at 78-90 cents) is losing credibility because Glendon named it publicly and JPMorgan marked it down operationally.

The question is not whether losses will be recognized. The question is whether the recognition is managed (gradual GP mark adjustments, orderly secondary market sales, pension fund allocation reductions over quarters) or unmanaged (a major fund failure, a cascade of gating across the industry, a bank pulling leverage from multiple funds simultaneously, a pension fund forced to write down its alternatives allocation in a single quarter).

The managed path is still available. It requires honest marks, transparent disclosure, and governance structures that consume the BDC prices rather than the GP marks. It requires pension fund investment committees to ask the question the Divergence Overlay makes visible: "What does our private credit allocation look like at market prices, not reported prices?"

The unmanaged path is the one where everyone waits for someone else to go first, and the gates get higher.

The gates are closing. The question is whether anyone opens them honestly before they are forced open by the math.

The Paranoidist publishes weekly, with flash issues when events warrant.

If this changed how you think about something you thought you understood, consider subscribing.

Paul Morin is the founder of DeepStrategy.ai, worries about the things that keep other people awake at night, and has more than three decades in entrepreneurship, finance, risk management, and insurance.

Researched, written, and edited in collaboration with Claude by Anthropic.