Newsletter · · Ashutosh Agarwal

Money Keeps Flowing to Credit's Safe End as the Alarms Go Quiet - Private Credit & Alternatives - Week of September 23, 2026

Private Credit & Alternatives for the week of September 23, 2026. Podcast synthesis on a quiet week after the summer's alarms: CLO ETFs pulling in roughly $2 billion a month (much of it retail), a hot CLO refinancing market, NAV lending maturing into a mainstream liquidity tool, a leveraged-finance conference where the private-credit worry had gone quiet, and Robinhood pushing private markets, including a BDC-wrapped fund of seed-stage startups, to 30 million retail accounts.

Private Credit & Alternatives

Week of September 23, 2026: Money Keeps Flowing to Credit's Safe End as the Alarms Go Quiet

Last week ended with a bang. A single loan inside a Blue Owl fund went from a full-value mark to nearly nothing in a matter of months, and for the first time the people who actually run the money (the co-CEOs of one of the largest credit firms in the world) went on the record describing the machinery of a downturn: collapsed recoveries, a bad loan vintage, a maturity wall dated to 2028-29. The argument had stopped being about whether the marks were real and become about how many more would come down.

This week, nobody printed a new 100-to-0. No named manager blew up. No insurance company or big alternatives boss picked up a microphone. After a summer of escalating alarms, the loudest voices went quiet, and that quiet is itself the story.

Because when the shouting stops, you can finally hear the plumbing. And the plumbing this week was working overtime, all of it pointed in the same direction: toward the safe, transparent, sellable end of credit, and toward getting more of it into the hands of ordinary investors. The refinancing machine that keeps loan funds cheap is running hot. The senior, "never-defaulted" slice of the market is pulling in roughly two billion dollars a month, much of it from retail. A whole side-industry has sprung up to lend money against private-equity portfolios so their investors don't have to wait for cash. And the biggest retail brokerage in America is now packaging late-stage startups, and even a fund of seed-stage companies, for its thirty million account holders.

None of that is a crisis. Most of it is the opposite: the market calmly rerouting money around the stressed corner rather than into it. But every one of these quiet moves is the same bet the summer's warnings were about: that illiquid things can be made to feel liquid, and that the safe end will stay safe. This week was a good week to look at where the money is actually going while the drama is off the front page.

TL;DR

  • The money kept pouring into credit's safe, transparent cousin, and it's increasingly retail money. On The NAVigator, Young Choi of King Street/Rockford Tower, which manages about $14 billion across 19 U.S. and 10 European CLOs, said exchange-traded funds that buy the safest slice of CLOs (bundles of hundreds of tradable corporate loans) are now taking in "around 1.8 billion or nearly 2 billion a month." He was blunt that this is not private credit: CLO loans are traded daily with "visible pricing," while in private credit "there's very little, if any, price discovery." His warning for the crowd rushing in: if private credit does blow up, CLOs could get thrown out "with the bathwater": painful, but a buying opportunity, not a solvency event.
  • At the year's big leveraged-finance gathering, the private-credit worry had gone quiet, a tell in itself. On At Any Rate, J.P. Morgan's credit strategists recapped their 10th annual conference in London (92 companies, over 1,000 investors). One strategist, comparing it to their Miami event earlier in the year, noted they "spent a lot less time talking about private credit and software refinancing." The mood was "cautiously optimistic" but nervier underneath: the investor question has shifted from "is high yield attractive?" to "where are we paid for assuming the risk?", and "a large or growing portion of the index spread is dominated by a relatively small number of stressed credits."
  • The industry's answer to the summer's liquidity scare is more ways to get money in and out, including a fund of startups for retail. On Origins, Robinhood's Sarah Pinto described two funds built to sell private markets to the roughly 88-90% of Americans who legally can't touch them: a $700 million fund holding 14 late-stage names (Databricks, OpenAI, Stripe, SpaceX), and a second fund, structured as a business development company (the same wrapper at the center of the summer's gating scare), that will buy stakes in about 150 early-stage Y Combinator startups a year and charge a hedge-fund-style "2-and-20." Both aimed at Robinhood's "30 million funded accounts."

What's New

1. The refinancing machine is running hot, and it quietly keeps loan funds cheap. Start with the least glamorous, most important gear in the whole system. On The CLO Investor Podcast (Sep 17), Chris Gilbert, co-head of securitized products at the bank Natixis (the same insider who last week reminded everyone that the CLO "safety" record was built almost entirely in a falling-rate world), walked through how these funds refresh themselves. A quick definition: a CLO (collateralized loan obligation) is a fund that pools hundreds of corporate loans and sells slices of the income, from the safest (rated AAA) to the riskiest (the "equity"). After a two-year lock-up, the equity owners can force a "refi" (reprice the fund's debt cheaper) or a "reset," which Gilbert called a chance to rebuild "a whole new cap stack. It's almost like a new CLO."

Two things made this worth your attention. First, the machine is busy right now: "We priced a reset for a middle market deal today," Gilbert said, "and we launched a reset for a broadly syndicated deal today as well. The market's active. The cost of the debt that we issue is lower today than it was looking back two years." In plain terms, CLOs can currently borrow more cheaply than when they were created, which means the funds holding leveraged loans are getting cheaper to run, not more expensive, even as everyone frets about credit. That is a real, if unglamorous, tailwind for the asset class.

Second, Gilbert laid bare a conflict of interest baked into the system, one that echoes the summer's whole "whose interests does this really serve?" theme. Fund managers, he explained, tend to push for the longer "reset" because it "extends the life of the deal and they earn management fees," while the equity investors who actually own the upside usually prefer the cheaper, shorter "refi." "It's a principal-agent problem," he said: the manager "incentivized by their fee to extend," the investor just wanting "to maximize the return." His rule of thumb for whether a refi is worth the legal and rating-agency bills: if the savings pay back the fixed cost "in less than six months, it's probably a great idea." He also flagged a bit of financial engineering called an "X note" (a top-rated slice that's really "financed equity," paid back fast out of early cash flows) and offered the line of the week on why the early years matter most: "credit's like fish, it doesn't get better with age."

2. Everyone's crowding into the transparent cousin, and roughly $2 billion a month is retail's. If private credit is the opaque, hard-to-sell end of the market, CLOs are its liquid, daily-priced relative, and the money is voting with its feet. On The NAVigator (Sep 18), Young Choi, a partner at King Street (a $30-billion credit manager with a 30-year track record) who runs its Rockford Tower CLO arm ("approximately $14 billion in AUM across 19 U.S. CLOs, 10 European deals"), described a stampede into the safest slice. "The growth of the CLO ETF market is really nothing short of remarkable," he said. These funds "started in early 2021," took until "early 2022 for it to eclipse the 1 billion AUM mark," and now the monthly inflows have "been averaging around 1.8 billion or nearly 2 billion a month." Almost all of it buys the AAA layer, and the flood of new buyers has actually pushed spreads tighter, good for the CLO equity, but a sign of just how crowded the "safe" trade has become. This lands right on top of last week's story of Janus Henderson's AAA CLO fund swelling to ~$31 billion, "at least half" of it retail: the senior end of credit is being bought from every direction at once.

Choi's most useful contribution was to insist that CLOs are not the thing everyone's worried about. He drew three sharp distinctions from private credit. The loans inside a CLO are "tradable. There are markets with visible pricing": "you can pull up a CLO and look at the assets, you can also go into the market and see where almost every one of those loans are trading." In private credit, by contrast, "there's very little, if any, price discovery," the exact opacity behind last week's 100-to-0 mark. CLOs report their holdings monthly and their borrowers' financials are obtainable; private credit's often aren't. And the track record is genuinely different: through the 2008 crisis and COVID, "you haven't had losses on AAAs or AAs," with maybe a single single-A tranche impaired in the market's entire history.

But he was refreshingly honest about what he is seeing beneath the calm surface, the same "everything looks fine until you look closely" warning the bears have made all summer. "If you look at the indices... everything seems relatively benign. But once you peel back the onion... there's quite a bit of dispersion. So there is a cohort of loans and credits that are performing fine. And then there is another cohort that actually isn't." And he conceded that if private credit does crack, CLOs could take collateral damage from being "conflated with private credit," the "baby out with the bathwater" scenario. His base case is that CLOs keep riding the wave; his hedge is that a private-credit scare would hand patient buyers a bargain rather than a wipeout.

3. At the year's big leveraged-finance conference, the private-credit worry had gone quiet. Here's the clearest "silence as data" signal of the week. On At Any Rate (Sep 17), five J.P. Morgan credit-research strategists recapped their 10th annual Leveraged Finance Conference in London, "92 companies presenting" to "over 1,000 clients." One strategist, Nelson Janssen, offered an almost throwaway observation that says a lot: compared with the firm's Miami conference earlier in the year, "you probably spent a lot less time talking about private credit and software refinancing." After a summer in which private credit was the only thing anyone wanted to discuss, the professionals who trade this stuff had, at least in London, largely moved on. Take that for what it's worth (this is sell-side research reading a room, not an all-clear), but a conspicuously quieter conversation is worth noting after months of noise.

What they were talking about rhymes with the bears' second point, that the average looks fine but the edges don't. The overall tone was "cautiously optimistic," with European high-yield bonds yielding "north of 6%, closing on 6.5%." But strategist Gareth Davis captured the shift in mood precisely: the question investors used to ask, "is high yield attractive?", "today has become, where are we paid for assuming the risk?" Dispersion (the growing gap between strong and weak borrowers) "is increasingly important," and, as strategist Daniel Lammy noted, "a large or growing portion of the index spread is dominated by a relatively small number of stressed credits. So it's becoming harder and harder to pick a portfolio that gives you the index yield without taking incrementally more risk." Davis, describing himself as "a grumpy credit guy," admitted he "came away a little bit more negative on the world ahead," precisely because "the benign economic backdrop... makes it relatively easy to be complacent."

They also kept the summer's other big thread alive: the AI borrowing wave crowding everyone else. The strategists counted "over $300 billion of issuance across markets year to date," "primarily U.S., primarily hyperscalers," with U.S. high-yield net volumes "up 60% year over year... driven by the AI financing boom." And the market's early verdict on that debt is lukewarm: its performance has been "at best middling, in some cases pretty bad" given how much is being sold at once, prompting a debate over whether investors need "much bigger new issue concessions" or simply "a pause in issuance." The read-through for private credit is the same as last week's: if even blue-chip AI paper is straining to find buyers, the weakest middle-market borrowers counting on a friendly refinancing window may find it narrower than they hoped.

4. NAV lending grows up, the liquidity workaround that keeps the exit drought from forcing marks. One reason this hasn't turned into a fire sale is a fast-maturing corner of finance most people never hear about. On Mission Matters Money (Sep 16), Louis Glanfield, a director at the fund-finance advisory Avarti Partners (named "Fund Finance Advisor of the Year 2026"), explained "NAV financing," where a private-equity fund borrows money against the value of the companies it already owns, rather than selling them. Why it's booming comes down to a plumbing problem: for "the last five years or so, the exit environment has been significantly slower," so "cash coming back to the underlying investors is either non-existent or very slow." Those investors ("pension funds," "insurers," "endowments") "made commitments several years in advance" and "want the liquidity." NAV lending "can sit in that gap."

Two shifts matter here. First, this used to be a product with "a bit of a negative connotation," a distress signal that a fund needed cash. Now, Glanfield said, "it's not necessarily a defensive option... very much it can be used as an offensive play," borrowing to fund "an attractive follow-on investment." And the lender base has broadened well beyond banks: "alternative lenders, insurance companies" have entered, making the market "a bit more competitive," with facilities now running "a three or a four or five year, maybe even seven year." Some "larger fund managers... are actually creating... lending strategies that are specifically going to lend to NAV," the surest sign a niche has gone mainstream.

The honest way to read this connects straight to last week's insider warning that "liability management exercises" merely "delay the day of reckoning." NAV lending is a genuinely useful tool that lets good funds hold good assets through a bad selling window. It is also, viewed less charitably, another way to borrow against illiquid holdings rather than test what they'd actually fetch: a liquidity bridge that works beautifully until the assets on the other side don't sell for what the model says. Which of those two things it is depends entirely on whether the exit market thaws.

5. Retail's on-ramp widens again, now all the way to seed-stage startups, via a BDC. The wealth channel isn't retreating from the summer's scare; it's accelerating. On Origins (Sep 21), Robinhood's Sarah Pinto laid out the pitch with genuine conviction: "Only about 10%, 12% of Americans are accredited. If you're not accredited, you can't invest in privates at all," which means ordinary people "can only watch companies like Anthropic or OpenAI... from the sidelines." Her fix is a fund type (the "40 Act" closed-end fund) that "the SEC allows anyone to invest in, including non-accredited investors," with "no minimums," that trades daily "like an ETF."

The scale is what's new. The first fund, RV1, raised through an IPO "in March of 2026," is now "a $700 million fund" holding "14 companies," the largest being "Databricks and OpenAI," alongside "Stripe... Ramp... Revolut... SpaceX... Airwallex." It charges "2% management fee and no carry," and "most of the capital was raised from retail," which Pinto rightly called "very, very novel." The second fund, RV2, is where the summer's warnings should start blinking. It is "a very diversified portfolio, very early stage companies" ("seed stage," meaning firms "that have existed for months, not years") built by investing in "probably 150 companies a year" out of Y Combinator's batches. Crucially, it "uses a vehicle called the BDC" (a business development company, the very structure whose retail versions spent the summer hitting redemption limits) and it charges a full "two and 20." Robinhood's reach is the whole point: "about 30 million funded accounts."

There is a real case for this: Pinto's argument that people learn by "trying it out for yourself," with the disclosure and daily pricing a regulated fund requires, is not nothing, and RV1's first-tier names and no-carry fee are a genuinely fairer deal than most private-market wrappers. But a fund of seed-stage startups is about as illiquid and lottery-like as investing gets, and putting it in a daily-traded wrapper aimed at thirty million retail accounts is exactly the mismatch (liquid on the outside, deeply illiquid on the inside) that the summer's gating episodes were about. The on-ramp isn't narrowing after the scare. It's widening, and reaching earlier-stage, riskier assets than before.

6. At the small end, operators are turning the stress into a sales pitch. Two smaller private lenders spent the week using the private-credit headlines to sell their own, differently-structured funds, a reminder that "not all private credit is created equal" has become the industry's favorite defense, and also its favorite marketing line. Treat both as operators with books to talk, but their framing is instructive.

On The Lifestyle Investor (Sep 17), serial entrepreneur Ben Rubenstein, who runs a $500 million asset-backed lending fund, offered the cleanest plain-English diagnosis of how the trouble built up. "Scale is the enemy of alpha," he said: when a fund has "so much money to deploy," it "make[s] more money on the management fees than... on the carry," so it's "incentivized just to deploy money, not necessarily to make money." Private credit "went from nothing a decade ago to trillions," and to place all that cash, funds "started to do a lot of direct lending, which is just lending to businesses based on no actual assets behind them." Then came "the SaaSpocalypse" (falling software-company values) and, because the loans had "no collateral" and investors "started all taking their money out at the same time," a dynamic that "it's like a bank run." He also pointed to outright fraud, naming "Tricolor and First Brands," where a borrower would "double pledge their assets" (pledging "the exact same cars" to two different lenders) or "manufacture invoices or alter invoices," with "nobody tracking." His own fund, by contrast, holds "$43 million of our own money," lends only against "US residential real estate... held in a bankruptcy remote SPV," and, he said, has had "zero loss" over seven years, with "six of our borrowers" winding down but the fund "fully repaid without even having to go to the collateral."

On The Infinite Wealth Podcast (Sep 22), real-estate lender Chris Lopez made the same argument from the collateral angle. Wall Street private credit, he said, is mostly loans "collateralized by the business" (private-equity buyouts) and "business collateral is usually worth a lot less than real estate collateral." That's why "a lot of those business loans and that Wall Street private credit, they're having a lot of pain right now... they're being revalued and a lot of them get marked down," hurt further because "AI is overcoming their moat" on software companies. His pitch: real estate "declines 15 to 35%" even in a crisis, while "businesses... they go to zero." How often, he asked, "does real estate go to zero?" It's a self-interested comparison (he's raising money for hard-money real-estate funds), but it captures why the market is fleeing toward anything with hard, sellable collateral behind it.

The Debate

If private credit cracks, does it stay contained, or splash onto everything next to it? This is the week's sharpest genuine disagreement, and it's mostly a debate within the bull camp. Choi's view on The NAVigator is that CLOs are structurally different (liquid, transparent, actively managed, with a spotless senior-tranche record), so a private-credit blowup would hit them only as sentiment "blowback," creating "a buying opportunity" rather than real losses. The J.P. Morgan strategists on At Any Rate implicitly agree the acute worry has cooled: they "spent a lot less time" on it than a few months ago. The uncomfortable counterpoint is that "contained" is exactly what everyone always says before a shock jumps the fence, and both sides concede the same underlying fact: beneath calm-looking indices, "quite a bit of dispersion," with spreads "dominated by a relatively small number of stressed credits." The honest synthesis: the senior, liquid parts of credit really are better protected, and the smart money is treating any contagion as a chance to buy, but "better protected" is not "immune," and the crowding into the safe end (two billion a month) is itself the kind of one-way trade that gets tested.

Is NAV lending prudent liquidity, or the exit drought's version of "delaying the day of reckoning"? Glanfield on Mission Matters Money makes the constructive case: with exits frozen for five years, borrowing against a good portfolio to return cash to pensions and endowments (or to fund a promising follow-on) is a sensible, now-mature tool, and the arrival of banks and insurers as lenders has made it cheaper and safer. The skeptical case writes itself from last week's insider testimony that debt reshuffles mostly "delay the day of reckoning": a NAV loan lets a fund avoid selling assets, which means avoiding the market test of what those assets are actually worth, the same instinct that keeps a loan marked at par until "reality overwhelms the model." Both are true at once. NAV lending is genuinely useful for a solvent fund riding out a bad selling window, and genuinely dangerous for one using leverage to postpone an unavoidable markdown. The tell will be whether exits thaw before these three-to-seven-year facilities come due.

Should a thirty-million-account brokerage be selling a BDC of seed-stage startups to non-accredited investors? Pinto's democratization argument on Origins is sincere and, for RV1, reasonably defensible: a diversified, daily-priced, fully-disclosed fund of already-huge private names at a 2%-and-no-carry fee is a fairer deal than the private-market products most retail investors have ever been offered. But RV2 is a different animal: a "two-and-20" fund of companies "that have existed for months, not years," wrapped in the same BDC structure whose retail cousins spent the summer hitting their redemption caps. The bull case is access and education; the bear case is that seed-stage venture is the most illiquid, most failure-prone thing in finance, and the wrapper's daily liquidity is a promise the underlying assets can't keep. This isn't a solvency risk to anyone systemic: the sums are small. It's the summer's core lesson being ignored in real time: the industry's response to a liquidity mismatch is to build more of it, for more people.

What the Silence Still Says

For an eighth straight week, the insurers holding the most opaque private-credit risk (Athene, Corebridge, Global Atlantic, F&G, Brookfield Reinsurance) did not speak in their own voice. This week they were joined in the silence by the big alternatives managers themselves: after last week's Brookfield co-CEO webcast and the Apollo pundit chorus, no Apollo, Ares, Blackstone, KKR or Blue Owl executive surfaced with anything material, and there was no fresh news on the 401(k)/retirement-access front that lit up earlier in the summer. Read it two ways. The generous reading: the acute phase has passed, the operators said their piece, and the market has moved on to plumbing. The cautious reading: the people with the most at stake tend to go quiet precisely when there's nothing reassuring to say and no reason to draw attention. A quieter week is not the same as a safer one: it's just a week without a new data point. The instruction from last week stands: watch the count of markdowns over the next two quarters, not the volume of the conversation.

The Names in Play

  • Natixis / the CLO refi-reset machine: via The CLO Investor Podcast, Chris Gilbert reports an "active" market pricing resets "today" at debt costs "lower... than it was looking back two years," a genuine, underappreciated tailwind for loan funds, offset by a manager-vs-investor fee conflict that pushes toward longer, costlier extensions.
  • King Street / Rockford Tower (private) / the CLO ETF complex: via The NAVigator, Young Choi pegs CLO-ETF inflows at "1.8 billion or nearly 2 billion a month," pushing AAA spreads tighter, and insists CLOs' liquidity and transparency make them a different animal from private credit, with contagion risk if the market conflates the two.
  • Robinhood (HOOD) / RV1 / RV2: via Origins, the retail on-ramp to privates: a $700M fund of 14 late-stage names (Databricks, OpenAI, Stripe, SpaceX) at 2% and no carry, plus a BDC-wrapped "2-and-20" fund of ~150 Y Combinator seed-stage startups a year, aimed at 30 million accounts.
  • Avarti Partners (private) / NAV lending: via Mission Matters Money, Louis Glanfield describes NAV financing maturing into a mainstream, "offensive" tool as banks and insurers pile in, the liquidity bridge letting stuck PE funds return cash without selling into a frozen exit market.
  • J.P. Morgan leveraged-finance research: via At Any Rate, strategists Steve Dulac, Gareth Davis, Daniel Lammy, Tarek Amid and Nelson Janssen: private-credit chatter notably quieter than earlier in the year, ">$300 billion" of AI-related issuance crowding markets, and spreads increasingly "dominated by a relatively small number of stressed credits."
  • The small-end operators, Ben Rubenstein (asset-backed fund) and Chris Lopez (real-estate lending): via The Lifestyle Investor and The Infinite Wealth Podcast, both using the "scale is the enemy of alpha" and "business collateral goes to zero" framing to sell hard-asset lending: self-interested, but a clean articulation of why money is fleeing toward collateral it can actually sell. Fraud names invoked: Tricolor and First Brands.
  • The silent ones: Apollo, Ares, Blackstone, KKR, Blue Owl, and the insurers (Athene, Corebridge, Global Atlantic, F&G, Brookfield Reinsurance) made no material appearance this week, a notable quiet after last week's chorus.

Read-Throughs

  • Watch the count, not the volume. Last week's parting instruction (track the number of near-total markdowns over the next two quarters) still governs, and this week produced no new ones. A single quiet week neither confirms nor refutes anything; don't mistake a slower news cycle for an improving credit cycle. The leading indicator remains a cluster of write-downs, and the mechanism behind them (collapsed recoveries, a 2028-29 maturity wall) is unchanged.
  • The safe end is where the crowd now is, which is its own kind of risk. Roughly $2 billion a month into AAA CLO ETFs, "at least half" of it retail, spreads compressing as a result: the flight to the transparent, senior, never-defaulted slice of credit is real and rational, but a one-directional trade into "safety" is exactly what gets repriced when everyone reaches for the same exit. The senior CLO record is genuinely strong; the crowding into it is genuinely new.
  • Know which "private credit" you own: the category is now too broad to be a risk label. This week reinforced the map: CLOs of tradable syndicated loans (liquid, priced daily, monthly reporting) are a different risk from direct lending against a software company's cash flows (opaque, modeled marks) is a different risk from asset-backed lending against real estate (hard collateral you can sell). The stress is concentrated in the middle one. If you hold "private credit," the single most useful question remains: which of these is it, and what actually sits behind the loan if the borrower fails?
  • NAV lending is the pressure valve to monitor. As long as private-equity exits stay frozen, expect more borrowing against portfolios to return cash to LPs. That relieves the immediate pressure that would otherwise force sales and markdowns, which is good news for stability and, simultaneously, a way of postponing the market test of what these assets are worth. If exits thaw, it was prudent bridging; if they don't before these facilities mature, it was the reckoning deferred. Either way, it's why you haven't seen more forced selling.
  • The retail wrapper keeps outrunning the underlying. AAA CLO ETFs, single-GP evergreen funds, and now a daily-traded fund of seed-stage startups: the industry's consistent answer to a liquidity mismatch is to build more liquid-looking wrappers around illiquid assets. Mostly harmless at today's sums, but the pool of ordinary investors holding genuinely illiquid credit and venture through liquid-feeling products keeps growing, and it's now reaching all the way down to companies "that have existed for months, not years."

What Changed

Last week the argument stopped being about whether the marks were real and became about how many more would come down, and, for the first time, the people telling you to worry were the ones who run the loans.

This week nothing broke. No new 100-to-0, no manager in trouble, and, notably, no big alternatives boss or insurer stepping up to a microphone at all. After a summer of escalating alarms, the loudest voices simply went quiet. At the year's major leveraged-finance conference, private credit was, by one strategist's account, a smaller part of the conversation than it had been just months earlier.

Into that quiet, the plumbing kept running, all of it pointed the same way. The CLO refinancing machine is active and borrowing cheaply. Two billion dollars a month is flowing into the safest, most transparent slice of credit, increasingly from retail. A maturing NAV-lending market is quietly letting stuck private-equity funds return cash without selling. And the biggest retail brokerage in America pushed its private-markets on-ramp further than ever, all the way to a "two-and-20" fund of seed-stage startups wrapped in a BDC, for thirty million accounts. Underneath, the AI borrowing wave kept crowding the credit markets, and the operators at the small end kept using the stress as a reason to buy hard collateral instead.

What changed, in a sentence: the drama receded, and in its place you could see the market's actual response to the summer's warning: not a retreat from illiquid credit, but a rerouting of money toward the parts that can be sold and the wrappers that feel liquid, with the underlying mismatch left in place. The bears still have their printed markdown, their collapsed recovery number, and a date on the calendar. The bulls have a quiet week, a cheap refinancing market, and a flood of new buyers at the safe end. Quiet is not the same as resolved. It just means the next data point hasn't printed yet.