The week AI's selloff reached the plumbing; JetBlue cracks

Archived reading, published Sat, 24 Oct 2026 13:39:07 UTC (37 hours ago). This is not the current state of the meter.

See the live reading →

Listen · 19 min

CRASH-O-METER

IT'S ON

how close are we

Fuel90

how much tinder is stacked up — moves slowly

Spark80

how close a spark is — moves fast

Crash Lab monitors the progress of the AI bubble and scans for signs of the crash. This page is updated daily around 10:00am ET.

Status: Held at 85. The week finished one point above where it started. The ignition half, how close a spark is, holds at 80. Friday's readings eased rather than built: the VIX, the market's gauge of how much turbulence traders expect over the next month, closed at 19.45 from 20.2, which puts it back under the 20 line that counts as calm (it passed 80 in March 2020); the Nasdaq 100 lost only 0.2%; the interest rate the government pays to borrow for ten years held just under 5% (4.99%); and the S&P 500 sits 2.5% below its record. The newest reading we have on the debt of companies too shaky to be considered safe is from October 22, when the extra interest they pay compared with the government widened to 3.4 percentage points, up 0.13 on the day and the widest of the month. That confirms Thursday's stress reached the riskiest borrowers without coming anywhere near the 8 percentage points that would trip our trigger. JetBlue's lenders organizing and SpaceX's stalled loan vehicle are sparks we had already partly counted, through JetBlue's bond price and the fact that the riskiest borrowers cannot sell new debt, and Thursday's hedge fund losses are a measurement of a day already on the books. The fragility half, how much tinder is stacked up, holds at 90. SpaceX's $40bn is money not yet lent, Walter's $2.31bn and the 146 stalled projects S&P counted measure loops and risks that were already standing, the Bank of Japan's stress test is a scenario rather than a position, and nothing was closed out by selling.

Reporting from 23 Oct to 23 Oct

Crash points

Eight places it could go, scored 0–10. Tap one for the explainer.

What’s moving the needle

What changed in the past 24 hours, and what each one says about the plumbing.

Hidden leverage and shadow banking23 Octhigh

JetBlue's lenders hire lawyers as its bonds fall to 42 cents on the dollar

A borrower with nothing to do with AI has gone from shaky to lenders-hiring-lawyers in three weeks, and that is what it looks like from the borrower's end when the market stops lending.

A group of the JetBlue lenders who stand first in line to be repaid has hired the investment bank PJT Partners and the law firm Akin, the Wall Street Journal reported, which is what lenders do when they think they may not be paid back on schedule. The airline's bonds have lost half their value in three weeks (its $2bn of notes due 2031 traded at 42 cents on the dollar on Thursday, down from about 80 earlier this month), and the shares have lost about 58% since the start of August. A bond at 42 cents is not a rough patch. It is the market saying it expects to get less than half its money back.

JetBlue is not an AI company, and that is why it matters to us. It is a plain story of a business burning cash, expensive fuel on thin profit margins, and it shows how a company that was merely weak in the summer becomes a candidate for restructuring once lenders pull back. The airline carried roughly $8.5bn of debt and aircraft-lease obligations in June, much of it secured on its planes and on the TrueBlue loyalty program. About two-thirds of its property, equipment and intangible assets are already pledged to lenders, who keep them if JetBlue cannot pay. There are unpledged assets left, but every new secured loan (the last was a $500m facility backed by aircraft, in April) takes security away from the lenders who have none, which is why the unsecured notes fall fastest.

The thing we would watch is the card processors. When you buy a ticket, the processor holds the cash until you have flown, and when an airline weakens the processor can demand that the airline pledge something against that cash. JetBlue said it faced no such demand at the end of June. If that changes, cash leaves the company at exactly the moment it needs it. Debt investors have been nervous since Spirit, a competitor, shut down earlier this year after repeated bankruptcies. Management updates its outlook on Tuesday's third-quarter earnings call.

Set against the week in credit, where the extra interest paid by companies too shaky to be considered safe reached its widest of the month on October 22 (3.4 percentage points over the government), JetBlue is the first borrower outside AI to go from expensive to distressed. For the odds of a crash this is the spark side rather than the tinder side, and we had already counted part of it through the bond price. What would change our view is the answer to one question: whether this is one airline's problem or the first of several.

The AI capex bubble23 Octmedium

SpaceX's $40bn chip loan, kept off its own books, has stalled

Keeping the debt in a separate company is how AI hardware gets paid for without hurting anyone's credit rating, and the most creditworthy borrower using the trick has just found that the market will not name a price.

SpaceX is trying to borrow $40bn without putting the debt on its own balance sheet, the list of what it owns and owes, and the selloff in AI-related debt has stalled the talks over what it would pay, Bloomberg reports. The structure works like this: a separate company, set up to hold the deal off SpaceX's books, borrows the money and buys Nvidia chips, and SpaceX uses them. Goldman Sachs is arranging it; the investors are led by Apollo, whose Atlas SP unit would manage the separate company; Pimco is considering joining. The stated reason for keeping it off SpaceX's books is to protect the company's credit rating, which sits in the top tier that lets a company borrow cheaply.

This is the template the whole boom now runs on. Anthropic's chip financing with Blackstone, reported on October 17, had the same shape, and the riskiest slice of it paid 10% interest. The chips sit in a box that is not the customer; the customer's lease payments pay the box's debt; the firms that grade how likely a borrower is to pay you back look at the customer and see a clean balance sheet. The debt exists all the same. It is held by whoever bought the box's paper, and for Apollo that often means an insurance company's own balance sheet.

What the week did was raise the price of the box. Bloomberg's sources say borrowing costs across the sector have risen and the two sides cannot agree a level, which is the polite way of saying the buyers want more interest than a borrower protecting a top-tier rating is willing to pay. If SpaceX cannot place $40bn at a price it likes, the smaller computing-rental companies underneath it, known in the industry as neoclouds, will not either; their bonds already pay buyers around 15%.

What is not reported: how long the loan runs, what interest rate is on the table, and what happens to the chips if SpaceX stops paying. Those terms decide whether this is a loan or a bet on the resale value of hardware that goes out of date quickly. Note also that the lead investor and the box's manager are the same firm, so the party deciding what the risk is worth is also the party running it.

For the odds of a crash: nothing has been added to the pile yet, because the $40bn has not been lent. What has changed is that the pile's main construction method has, for the moment, stopped working. That is the fragility side, not a spark, and a deal that fails at this level would tell us more than a dozen smaller ones.

Hidden leverage and shadow banking23 Octmedium

Hedge funds had their worst day since Covid, and lost on both sides of the bet

When the bets meant to protect a fund fail at the same time as its main bets, and the most heavily borrowed firms lose together, selling stops being a choice and becomes mechanical. That is how a selloff in one sector becomes a selloff in everything.

Thursday was the worst one-day loss across the hedge fund industry since the Covid crash, by the estimates of the Goldman Sachs desk that lends these funds money and holds their positions, reported on Bloomberg Television. Hedge funds are the investment firms that manage money for the wealthy and can bet in ways ordinary funds cannot, and more than two-thirds of them lost money on the day. Funds that pick stocks by hand, funds that pick them by computer and funds focused on technology all fell. The multi-strategy platforms, the giant firms built so that one desk's bad day cannot hurt the whole firm, fell almost 2%, their second-worst day since April last year. Those firms are designed not to have days like that.

The reason this is not simply "the Nasdaq fell" is in the bets against stocks. A hedge fund bets a share will fall by borrowing the share and selling it, and those bets are supposed to make money when the market drops. On Thursday the most popular of those bets rose while the ordinary holdings fell, so both sides lost at once. That pattern usually has one cause: somebody large is shrinking every bet it has, buying back what it sold and selling what it owns regardless of what it thinks anything is worth, and the relationships everyone else relies on stop holding.

The multi-strategy platforms matter because that is where the borrowed money is. They run dozens of teams across shares, government bonds, corporate debt and commodities, with bank desks lending them several times their own capital, under strict firm-wide limits on how much they can lose. A 2% day at the whole-firm level trips those limits, and the response is the same shrinking of bets that caused the damage in the first place. Bloomberg's reporters note that the same firms had been losing money in government bonds, corporate debt and convertible bonds for two months, so the cushion was already thin.

Goldman's own figures from October 20 showed hedge funds' net exposure, their holdings minus their bets against, at a one-year low, so some of the cutting had happened before Thursday. The industry still runs nearly $5.5tn. What nobody outside can see is the gross figure, the holdings and the bets against added together, and gross is what forces selling. Friday was calmer (the Nasdaq 100 closed down 0.2%), which suggests the shrinking paused rather than finished. On the crash question this is a measurement of a day already in the price rather than a new load. But it tells us Thursday's selling was partly forced, and forced selling is what turns one bad day into several.

Bond market dysfunction23 Octhigh

Moody's spares France; the National Assembly does not spare the budget

France's credit rating now rests on a budget its parliament is rejecting piece by piece, so the pressure on the country's debt is paused rather than resolved.

Moody's, one of the three firms that grade how likely a borrower is to pay you back, kept France's grade where it was on Friday, with a warning that the next move is more likely down than up, Bloomberg reports. The grade is Aa3, one step above where the other two firms, S&P and Fitch, put France in 2025, and seven steps above the line below which a borrower is no longer considered safe. It was the week's most anticipated non-event. The interest rate France pays to borrow for ten years passed 5% on Thursday, its highest in 24 years, and it has risen more this month than any other rich country's.

Moody's reasoning is worth reading because it describes a loop. "Slower growth and a higher deficit have pushed France's debt burden somewhat higher than previously expected," it said. "Bond yields have risen more rapidly than we expected, while widespread social protests have erupted which could constrain fiscal consolidation efforts." In other words, France is borrowing more, paying more to do it, and facing street protests against the cuts meant to fix that. The gap between what the government spends and what it takes in is set to reach 5.4% of the country's output this year. Prime Minister Sébastien Lecornu's budget aims for 5% next year, but he has no majority in parliament, and on Friday the National Assembly rejected the revenue half of the 2027 social-security budget, sending the original draft on to the Senate. The main budget is debated next week, after another weekend of protests the government is bracing for.

The rating holds because it assumes the budget passes. Borrowing costs have risen because the market does not. Higher rates raise the interest bill, which widens the gap the rating depends on closing. Moody's has in effect said it will downgrade if the cuts fail; the Assembly has just shown what it thinks of the cuts.

For the crash odds, France matters as the first large government where this year's global level of interest rates met a political system that cannot answer it. The rush to safety that lifted US government bonds this week did not reach French ones. A reprieve from a rating firm changes nothing about that arithmetic. It only delays the day the arithmetic gets written down, so this is the same tinder with a slightly longer wait, and we count it as neither up nor down.

The dollar, gold and reserve status23 Octhigh

US government bonds in, gold out: where money hid this week

A supposed safe haven that falls in a week when investors are selling everything risky is not being sheltered in. It is being sold to cover losses elsewhere, and that tells you where the borrowed money was.

The Nasdaq 100 ended the week with its worst two-week fall since April last year, 7.3% below the record it set earlier this month by Bloomberg's count. Meanwhile money poured into bond funds that trade on the stock exchange like a share: roughly $58bn in October, about 38% of everything that went into such funds, according to Bloomberg Intelligence. The interest rate the US government pays to borrow for ten years fell below 5% for the first time in a month, and the largest fund holding government IOUs that do not come due for twenty or thirty years rose for a third straight week. "We're seeing a flight to quality bid of Treasuries inverted to the AI trade," said Neuberger Berman's Jeffrey Blazek. "It's completely a plot twist." In other words, investors are running into US government bonds as they run out of AI shares, which is the opposite of what they did for most of the year.

It is a twist. For most of this year the standard position was to own big tech and bet against government bonds, on the theory that AI spending would push up inflation and interest rates with it. That is why October's earlier selloffs hit shares and government bonds together. This week they came apart. US government bonds behaved as a safe place to hide for the first time since the summer, and the S&P 500 ended the week down only 0.5%. Blackstone's president says there is no bubble, and the Treasury Secretary, who runs the department that borrows the money the government spends, sees no problem that could spread through the system. The flows say investors are buying protection against one anyway.

Gold did not get the money. It fell below $4,000 an ounce on Friday for the first time since July, the FT reports, down more than a fifth from its January peak of $5,400 and 3.3% on the week, "despite the risk-off backdrop", as Berenberg put it, meaning in a week when investors were selling anything risky. The last time gold was this weak while government bond rates fell this hard, by Berenberg's reckoning, was December 2008. Gold-mining shares had their worst week in seven months. The usual explanation, that gold suffers when people expect interest rates to rise, does not fit a week in which those expectations fell. The FT's explanation does: gold became a speculative bet last year, and speculative bets get sold to pay for losses elsewhere.

So the map: US government bonds in, gold out, French bonds nowhere. And the Bank of Japan, in Friday's report on its financial system, ran a test in which US shares fall 50% on an AI bust and concluded its banks would survive it. That a central bank now writes that scenario down is the week's quietest fact. On the odds of a crash, the week leans toward less likely in one respect, since the safest asset is doing its job again. The gold sale cuts the other way, because it shows borrowed positions being closed, and that is how a loss in one place travels to another.

Hidden leverage and shadow banking23 Octmedium

When his own insurers stopped lending to him, $2.3bn arrived anyway

It is a rare documented look at insurance companies' money being lent to people connected to them through middlemen, the kind of exposure nobody puts a price on until a subpoena arrives.

Mark Walter, who led the purchase of the Los Angeles Dodgers, was served with subpoenas in a federal criminal investigation in February, and his own insurance companies abruptly stopped lending to him. Within weeks, $2.31bn arrived from somewhere else: Amistad Financial Group, an insurance holding company on whose board sits Magic Johnson, Walter's business protégé, who also owns a minority stake. Bloomberg pieced the loans together from business records; they had not been reported before. About $1.26bn went to Walter's companies between March and June. Another $1.06bn went to middlemen that routinely pass loans through to him.

The investigation is being run by the Department of Justice and the SEC, the agency that polices American markets, side by side. It concerns Walter's insurance companies and the earlier deals in which Amistad acted as the middleman for him to borrow from his own insurers. Walter holds a single-digit voting stake in Amistad and does not control it, according to a person familiar with its operations. Some of Amistad's insurance units are run by Johnson's former business manager. Even before this year, Amistad's insurance units had lent billions to companies tied to Walter.

Why this belongs here: it is the machine of lending by investment funds rather than banks, in miniature. Insurance companies take money from the public, and the question regulators always ask is what they buy with it. Here the answer was loans to a billionaire's holding companies, routed through middlemen, from a lender in which the borrower holds a stake. Nobody puts a market price on that loan. Nobody has to. The policyholder sees a solvent company either way.

The terms of the loans and what was pledged against them are not reported. What the story shows is that when one captive source of insurance money shut, another opened within weeks, and that the pipework letting insurance money flow to connected borrowers is wider than the single case under investigation. For the crash odds this is the third answer: nothing got more dangerous this week. We just found out how it already works.

Signs of the times

Stuff you wouldn’t have believed was possible until 2026.

23 OctBloomberg

Hedge funds are betting more heavily against the euro than at any point since the Covid crash

Traders betting with borrowed money held about €11.8bn ($13.2bn) of bets that the euro will fall in the week to October 20, the largest such position since March 2020, according to the CFTC, which polices American futures markets and publishes these positions. They added 1,305 contracts to reach 94,138 as France's budget fight and street clashes rattled European markets. Speculators have been betting against the euro since February, but this month's French turmoil has deepened the position by raising doubts about whether the European Central Bank can keep raising rates while the Federal Reserve, America's central bank, is expected to head the other way. The euro ended the week 0.4% lower. The last time hedge funds were this gloomy on the euro, the world was shutting down for a pandemic; this time the trigger is one country's parliament rejecting a budget line by line.

eurofrancepositioning

23 OctYahoo Finance

Apollo is pitching the New York Yankees to its lending clients as 'AI-resilient leisure'

Apollo has described its $2.6bn financing of the New York Yankees, agreed in August, as 'AI-resilient leisure' in a note to investors in one of its funds that borrow money, lend it to mid-sized private companies and pass the interest to shareholders, according to a Yahoo Finance discussion of the headline. The hosts could not establish whether any of the Yankees deal actually sits in the fund being marketed, or whether the deal, which they understood to be mostly an ownership stake with perhaps some preferred shares, is simply the poster child for what the firm would rather talk about than its software loans. Either way, the direction is the signal: a lender whose business for a decade was financing companies now reassures clients by pointing to the one asset that by definition cannot be automated, a baseball team.

private creditmarketingai

23 OctBloomberg

A rehab chain co-founded by a convicted fraudster wants to sell $500m of ungraded municipal bonds, some not due until 2067

Flyland Recovery Network is selling eight addiction-treatment centers to Stepstone Health Foundation, a nonprofit formed in December 2025, in a $500m deal arranged by RBC Capital Markets. Stepstone plans to raise the money by selling $292.5m of bonds due 2062 that get paid first and $210.7m of bonds due 2067 that get paid after them and pay no interest until then, and none of it has been graded by the firms that rate how likely a borrower is to pay you back, which is the check a buyer normally wants before lending for 42 years. Flyland's co-founder, Kyle Robertson, pleaded guilty in 2014 to conspiracy to commit fraud over a telemarketing scheme that took $14.5m from more than 7,000 victims, and served 13 months; the sale document discloses this alongside his own 2009 addiction. He and the current executives will keep running the centers for at least 15 years on a fee of 4% of gross revenue. Columbia Threadneedle has already declined to buy, citing the leadership.

munisunratedhealthcare

23 OctBloomberg

US states and cities have already borrowed $500bn this year, the fastest pace on record

Borrowing by states, cities and other local bodies passed $500bn this week, almost 6% ahead of the same point last year and on course to top $600bn by year-end, according to data compiled by Bloomberg. In 2025 the $500bn mark was not crossed until early November. The borrowing is happening into the highest interest rates in years because, as MacKay's Eric Kazatsky put it, there are 'capital projects that need to get funded', airports and infrastructure, and waiting has stopped being an option. The buyers are there: the Bloomberg index of these bonds is up 2.8% in October, its best October in at least 30 years and its first positive month since June, as the same rush into bonds that took the ten-year US government rate back under 5% lifted everything that pays interest.

munisissuancerates

The rumour mill

What the crash-callers are saying, checked against real reporting.

ConfirmedClaimed by GoldSilver

“CoreWeave's bonds due 2031, carrying a 9.75% coupon, now yield close to 16%.”

CNBC did report CoreWeave's notes due October 2031, which were sold paying 9.75%, trading on October 20 at a price that gives a buyer holding them to the end a 15.97% annual return, with other trades between 15.58% and 15.78% in the week before. A buyer demanding nearly 16% a year from a company carrying roughly $50bn of debts is a buyer who doubts it will pay; that is distressed territory, and it matches the market for insurance against CoreWeave failing to pay, which puts the odds of default within five years at better than even. What would move it is a guarantee with Nvidia's name on it, or CoreWeave managing to borrow again at anything near the rate it originally paid.

ConfirmedClaimed by Anthony Pompliano

“Lone Pine Capital, up over 40% in the first half of the year, gave back the entire gain by end of September, finishing flat.”

Institutional Investor reported that Lone Pine's main fund, which owns some shares and bets against others, fell 31% in the third quarter after a strong first half and finished roughly flat for the year; Bloomberg's figures were a 44% gain through June and 0.6% through September. Funds have bad quarters. Losing nearly a third in three months at a flagship fund is not a bad quarter; it is this month's industry-wide hedge fund loss in a single name. The question is how far the fund shrank its total bets to stop the bleeding.

Partly trueClaimed by Heresy Financial

“The head of Goldman Sachs' trading desk says the long end of the US Treasury market is 'totally bidless' — there are no buyers at all.”

A Goldman desk head was quoted on October 1 describing government IOUs that do not come due for twenty or thirty years as 'still totally bidless', but the quote circulates only through unverified secondary sites, and in trading talk it means buyers are scarce at current prices, not that there are none. September's auctions of ten-year and thirty-year government debt drew $2.71 and $2.61 of bids for every dollar on offer, and this week the ten-year rate fell below 5% as investors ran to safety. An auction next month where buyers demand a higher rate than the market expected would settle it; this week's money flows argue the other way.

Partly trueClaimed by Peter Schiff

“US producer prices rose 1.3% in September alone — an annualized 17% inflation rate.”

The Bureau of Labor Statistics did report a 1.3% rise in September in the prices producers charge, and compounding one month twelve times does produce about 16.8%. But the official figure for the past twelve months is 6.3%, and one month of fuel-driven price rises is not a trend. The honest version is that producer prices rising 6.3% a year is already high enough to keep alive the expectation that interest rates will go up; the 17% framing is arithmetic dressed as a forecast.

UnsupportedClaimed by Michael Bordenaro

“US school districts carry $5.1 trillion in outstanding bond debt, fraudulently issued and impossible to repay.”

The best available estimate, from Bellwether Education Partners, puts school districts' long-term debt at about $570bn at the end of the 2023–24 school year, roughly a ninth of the figure claimed. The $5.1tn number traces to a single investor's allegation repeated on a partisan blog, and S&P, one of the three firms that grade how likely a borrower is to pay you back, has recently given high grades (in the AA range) to school bonds in Indiana and Missouri that are repaid out of property taxes. There is a real story in local-government borrowing this year; this is not it.

Partly trueClaimed by Coin Bureau

“Under MSCI's new index rules, Strategy could be removed from MSCI indices, forcing index funds to sell roughly $1.8bn of MSTR stock.”

MSCI, which decides what goes into the stock indexes that many funds copy, did adopt rules on October 15 that screen out companies with no operating business, and Strategy is a plausible casualty. But the list of affected companies is not due until November 11, with the change taking effect November 30, and the $1.8bn is an estimate of what index-tracking funds hold, about 13 million shares, not a confirmed forced sale. If Strategy is on the list the selling is mechanical and dated; until then it is a possibility priced into a stock already 13% off its high.

UnsupportedClaimed by Wealthion

“Revaluing US gold reserves from their $42/oz book value to market prices would credit the Treasury's account with spendable cash without new debt.”

No revaluation has happened, no bill to change the official price has been introduced, and Treasury Secretary Bessent has said this is not what he has in mind. The mechanism exists on paper: a certificate for the gold at a higher price could be deposited at the Federal Reserve, America's central bank. But it needs Congress, and with gold now under $4,000 the sum involved is a fifth smaller than when the idea was fashionable. It remains a thought experiment.

Earlier developments

Dispatches from previous readings. The same argument, no longer the news.

The AI capex bubble22 Octhigh

Meta's default insurance at a record; insurers hold the debt

The AI buildout's debt is now priced as a risk at the top of the capital stack and held, partly unlabelled, by insurers at the bottom of it.

The cost of insuring Meta's debt against default reached an all-time high on Thursday. Five-year credit default swaps widened 7.6bp to 114.79bp, the biggest one-day jump Markit has recorded for the name since at least 2025, after Meta raised the floor of this year's capex guidance to $135bn — the range is now $135–145bn, from $130–145bn (Reuters). The shares closed down 7.1% at $648.76, 17% off their high. Oracle's CDS sat close to its own record, Amazon's near its widest since March 2020, and the Philadelphia semiconductor index shed more than $744bn of value in the session.

Put the number in context. 115bp is not distress; CoreWeave trades at roughly ten times that. What it says is that the market has stopped treating Meta as a company that funds AI from cash flow and started treating it as a borrower, and is pricing the issuance it expects to follow the $825bn of commitments we covered yesterday.

Which raises the question of who buys. Moody's answered part of it on Thursday. US life insurers held $85.2bn of AI-related investments at the end of June, up 24% in six months, including $32.2bn of the five hyperscalers' debt — about 6.6% of everything they have outstanding (Bloomberg). The faster-growing bucket is $39.5bn of debt backing data centres, GPUs and TPUs, up 34%, and the share of that rated junk has risen from 4.7% to 6.1%. Moody's also says the direct figure understates real exposure, because bonds backed by hyperscaler guarantees and leases are not counted as hyperscaler risk.

That last sentence is the thesis of this site in one line. The guarantees and leases that have filled our dispatches for a month land on insurance balance sheets where they are classified as something else — project finance, infrastructure, lease receivables. At 1.4% of invested assets the sector can absorb a great deal. The question is whether 1.4% is the right number, and Moody's own report says it is not.

Bond market dysfunction23 Octhigh

Spreads at a 17-month wide; Treasuries have their best day in years

The AI trade's unwind has become a cross-asset event — credit wider, bank capital repricing, Treasuries bid — which is the shape a real risk-off takes.

Bloomberg's global corporate bond index widened to 129bp on Friday, its widest since May 2025 (Bloomberg). Asian high-grade dollar spreads reached their widest since last July; the regional CDS index opened 1.5bp wider at 78, China tech 3–5bp wider, Korea and India investment grade 3bp. None of these is a large number. Together they are the first time this month the AI selloff has shown up in credit outside the AI names themselves.

The other side of the trade is where the money went. Treasuries had their best day in more than two years, with buying across the curve and a rush to cover shorts in futures. The 10-year is at 4.97% this morning, below 5% for the first time in a month; the 30-year is 5.29%; the long-bond ETF rose 1.9% on Thursday. This matters for our reading. The rate shock that drove October's editions was the thing most likely to break something. A rally driven by fear of AI debt eases it, for now.

The funding end of the bank system is moving too. The average spread on Bloomberg's index of contingent convertibles — AT1s, the riskiest bank capital — has risen to 253bp from below 200bp in August and September, the first time it had ever been that tight (Bloomberg). BNY Mellon's latest preferred priced 32.4bp wider than an essentially identical deal three months earlier, a Belfius AT1 was indicated below par immediately after issue, and orders for Banco BPM's deal shrank while the book was open. Banks are paying a concession to sell capital that was being thrown at them in the summer.

Then the caveat. The high-yield spread we hold (21 October) is 327bp, tighter on the day and nowhere near the 800bp trigger, and the 129bp global figure is an investment-grade index. What has widened is the premium on the safest corporate debt and the riskiest bank debt at the same time, while the middle has not yet printed. Each camp has a champion: Ken Griffin says AI is almost certainly a bubble, Blackstone's Jon Gray says demand is "almost parabolic". This week the spread index moved toward Griffin.

Bond market dysfunction22 Octhigh

Where the junk selloff is actually landing

The aggregate spread is calm because the stress is in the tail, which is where every credit cycle starts and where the funds that grew fastest are now marking losses.

Three facts from Thursday, each about a borrower with no cushion.

Zorlu Enerji, a power producer owned by one of Turkey's largest conglomerates, will not pay the roughly $55m coupon due today on its $1.1bn 2030 bond and will ask holders for up to six months' delay (Bloomberg). The bond was indicated in the mid-50s on Thursday; Fitch cut it to B- in September; the company hired Houlihan Lokey a month ago; its sister company Vestel is already restructuring. This is what the end of cheap dollar borrowing looks like for a Turkish industrial: debt in dollars, costs in a lira that has strengthened in real terms, and minimum wages rising.

JetBlue's $2bn of 2031 bonds fell to about 40 cents on the dollar from around 80 at the start of the month, and the stock closed at $2.65, its lowest since the 2002 IPO (Bloomberg). Those bonds are secured on the loyalty programme, and the market now values that security at less than half of par. United's Scott Kirby said this week that other airlines are "next up on the firing line" after Spirit liquidated.

And the funds that own this kind of paper are marking it. Arini's master fund lost 6.9% in the first half of October on top of a 15.7% loss through September; Sona's lost 3.1% and is heading for its worst month in a decade (Bloomberg). Both run more than $20bn and neither had ever lost money in a calendar year. Arini is a major holder of Altice International and Aston Martin; Sona was hit on Virgin Media O2.

The high-yield index at 327bp says none of this is happening. Bloomberg's account is that the pain is "concentrated among companies with some of the weakest credit ratings", which — our inference — is why a broad index does not show it. Markets rarely widen uniformly; they drop the bottom first. Pearlstone's Ivelina Green put it plainly: "People are not aware of the pain we are experiencing in credit because it happened over such a short period of time."

The AI capex bubble22 Octmedium

OpenAI takes $150bn off its price; Solidigm pulls $10bn IPO

When the leading AI company cannot fill a round at its asking price, the buildout's funding shifts further toward debt that someone has to insure.

OpenAI has lowered the valuation target on its pre-IPO round to $1.25tn from the $1.4tn it pitched earlier this month, after Morgan Stanley's wealth clients — offered the deal alongside Dragoneer — declined to bite (Financial Times). "There has been zero clamouring," one person told the FT. "All the prior times OpenAI have been on there's been a frenzy. This time is a bit of a thud." The round is a $30bn "bridge" to a listing now expected in the first half of 2027; investors have until 30 October to commit, Coatue and UAE funds are in talks to lead, and the company may raise less than $30bn.

Keep the scale straight. $1.25tn is still a step up of almost 50% from the $852bn mark set when OpenAI raised $122bn in March. The company is not being repriced down; it is being repriced up by less than it wanted. What has changed is the marginal buyer. Wealth-management clients are the last layer of demand in a private round, and they have "indigestion" after $40bn in March 2025, $122bn this March and nearly $100bn raised by Anthropic this year. Several told the FT they would wait to see Anthropic's IPO price first — a public mark arriving to test a private one.

SK Hynix, meanwhile, scrapped the $10bn US listing of its Solidigm flash-memory unit and halted pre-IPO talks, citing "recent capital market conditions" in a filing (Bloomberg). Bloomberg's account is that the withdrawal was driven mainly by Korean regulators and minority shareholders objecting to the group's control structure — SK Hynix shares rose 4.5% on the news — so this is not a purely market-driven failure. But it joins Firmus Grid and AirTrunk on the list of data-centre-adjacent listings pulled this month.

The pattern this week: OpenAI took a lower price, Nebius took a one-year bank loan, Solidigm took nothing. The equity window for the AI buildout is not shut, but the price of using it is rising, and the alternative is debt — which is where Meta's record CDS comes from.

Private credit and BDCs23 Octmedium

Direct lending's worst quarter since 2023 sends lenders downmarket

Private credit's problem this quarter is not that money is leaving but that it cannot find enough borrowers, so terms loosen at the weakest end of the market.

Direct lending volumes fell to $38.9bn in the third quarter, the lowest since the second quarter of 2023, according to JPMorgan (Bloomberg). The managers' response is to chase smaller deals. Barings and CVC Credit Partners are among lenders offering to finance Sagard's sale of EOS Corrugated — a maker of tools for cardboard manufacturers with about €25m of annual earnings — with a package of 5.5 times those earnings, priced as low as 475bp over benchmark. Bloomberg's sources call the leverage and terms "atypical for smaller companies with shorter track records and thinner cushions".

This is the dynamic that matters most in private credit, and it is the opposite of a run. Too much committed capital, too few buyouts to lend against, and fund economics that punish undeployed money mean the competition is for borrowers, not for funding. The borrower wins: more leverage, a tighter spread, on a company that in a downturn has nowhere else to go. The loans will sit in funds at par because nobody has to mark a €25m-EBITDA credit until it misses a payment.

The listed managers were bid hard on Thursday — Blackstone up 8.3%, Blue Owl 5.9%, Ares 5.7%, KKR 3.6% — on the same day Blackstone's Jon Gray called AI demand "almost parabolic" and the junk market was being sold. The BDCs rose with them (Ares Capital 2.2%, FS KKR 4.3%), though Prospect Capital is 31% lower on the month. We cannot reconcile a quarter of "small ball" with a managers' rally except as a bet that the fee stream outlives the credit cycle — which is precisely the thing that has not yet been tested.

One regulatory footnote: the FCA on Friday restricted six small UK private credit lenders from taking new customers or writing new loans over anti-money-laundering failings, having promised more scrutiny of unregulated lenders after the collapse of property lender MFS earlier this year (Reuters). Small firms. But a regulator that has decided to look.

Household credit22 Octhigh

Sallie Mae cuts its outlook by a third as charge-offs rise

A federal borrowing cap has pushed graduate student debt onto a private lender's balance sheet just as that lender's losses started rising.

Sallie Mae now expects to earn $1.95 to $2.05 a share this year, down from $3.10 to $3.20 — a cut of more than a third — after third-quarter net charge-offs rose 17% to $90.4m (Wall Street Journal). Provisions for credit losses rose 19% to $214m, net interest income fell to $339m from $373m, and the company swung to a $32.3m loss from a $135.9m profit a year earlier. The stock fell 4.5% and sits 17% off its high.

The company's explanation is specific and partly reassuring: a third-party debt-management firm mishandled a small group of borrowers who could afford to pay and let them run straight from delinquency to default, and those borrowers are taking longer than expected to get back on track. Delinquencies are up only 0.2 points. If that is the whole story, this is an operational error, not a credit cycle.

The more interesting number is on the other side of the ledger. Originations rose 15% to $3.4bn, and graduate loan originations more than doubled, because federal programmes now cap how much graduate students can borrow and Sallie Mae has stepped in to fill the gap. That is the thesis of this site in miniature: credit the federal government stopped extending has migrated to a private balance sheet, where it is funded by private capital rather than the Treasury, and where the first sign of trouble is a guidance cut rather than a line in a budget. The provision increase is partly the cost of booking that growth — the company says the loss was "primarily due to provision and expenses related to higher originations".

So the thing to watch is not this quarter's charge-off but the next two years of a doubled graduate book, underwritten in 2026, by a lender whose stock is already 17% off its high before any of that lending has seasoned.

Bond market dysfunction22 Octhigh

France pays 5% to borrow while US government bonds become the refuge

The rate shock has turned around: investors are buying US government bonds for safety again, and the strain has moved to a major European government that cannot print the currency it owes.

France's government paid more than 5% to borrow for ten years on Thursday morning, the most since 2002 (the ten-year rate reached 5.0601%), and the extra France pays compared with Germany reached its widest since 2011 (1.64 percentage points, per LSEG data cited by the Wall Street Journal). Germany's borrowing cost went the other way, down 0.017 points to 3.407%. So did America's: the interest rate the US government pays to borrow for ten years fell 0.059 points to 5.088% during European trading, and the rate for thirty years hit a four-week low of 5.376%. By the time we built this edition the US ten-year rate was 5.04%.

Read those three moves together. France, Germany and the United States borrow in the same world, and this morning investors sold one of them to buy the other two. "Today's tentative risk-off, with U.S. Treasurys supported and struggling equities, argues for continued pressure on eurozone government bond spreads," Commerzbank's Hauke Siemssen wrote. In other words, when investors sell anything risky and buy American government bonds instead, the weaker European governments are the ones left paying more. The Journal attributes the rush into US bonds to falling stock prices and "concerns about the sustainability of debt-funded investments in artificial intelligence infrastructure", and notes that worries about growth now outweigh worries about inflation.

That is the mechanism. For most of October the story was a rate shock: US government bonds sold off, and everything priced from them repriced. Today those bonds are the refuge, and the pressure has moved to the one large borrower that cannot print the currency it owes. The gap over Germany is what investors demand for lending to France rather than Germany in the same euro; at 1.64 percentage points they are demanding more than at any time since the euro crisis, and the Journal names budget worries and social unrest as the reasons.

What it does not yet show is anything breaking. The Journal reports no failed auction, and no sign that French banks or insurers, the natural owners of French government bonds, are being forced to sell. But a major government paying 5% raises the cost of every French mortgage and company loan priced from it, and the rush into US bonds happened because investors fear the AI boom, not because they stopped fearing American debt. Yesterday we recorded the French gap above 1.5 percentage points for the first time since 2012. Today it is 1.64. That is a step, not a repeat, and it is ignition: nothing new was stacked on the pile, but a spark in a core European bond market got closer.

The AI capex bubble22 Octhigh

Meta's $825bn: the borrowing sits on other people's books

The largest single set of AI promises is growing by $125bn a quarter, and the debt that pays for it lives with landlords and their lenders, not with Meta.

Meta's quarterly filing on Thursday puts a number on what it has promised to pay: about $825bn of future spending commitments, much of it for computing power and AI infrastructure, according to Bloomberg. In July the figure was almost $700bn. That is roughly $125bn of new obligation signed in a single quarter, and adding that much every three months is not a routine thing even for a company of Meta's size.

What it is made of matters more than the total. $355.7bn is leases not yet on Meta's balance sheet, the list of what it owns and owes: data centers and energy infrastructure whose leases start between now and 2036. Another $454.5bn is contracts Meta cannot cancel, many for computing rented from other companies, energy for data centers and Reality Labs hardware. Meta is also obliged to buy up to $14.7bn of cloud computing over five years, an amount that may fall if the provider sells to other customers.

A lease that is not yet on the balance sheet is a building someone else is paying to put up on the strength of Meta's signature. The developer borrows, the investors who lend to the developer look through to the tenant, and the debt sits with them rather than with Meta. Last week the bonds funding data centers rented to Meta, Nvidia and Oracle were trading well below the price they were sold at; this filing says the pipeline of such promises is $355.7bn deep and runs for a decade. That is the migration this site keeps describing: the obligation is Meta's, the borrowed money is somebody else's, and nobody has to admit a loss until a lease payment is late.

On Wednesday Meta raised its spending plan for the year and announced a new class of shares that carry no vote, which could be used to pay for acquisitions and other expenses. Shares without votes let a company raise money by selling stock without the people in charge giving up control, and a company whose shares are 10% off their high does not invent a share class for fun. The filing does not say how much of the $825bn falls due in 2027, nor how much of the leased capacity has been financed by lenders who would lose if Meta's plans change. Those are the numbers that would tell us where the risk actually sits. For the meter this is fragility: $125bn more borrowed on Meta's word in a quarter, stacked where nobody has to price it.

The AI capex bubble22 Octmedium

Citi lends Nebius $3.5bn for one year as the market for new listings stalls

When the stock market will not fund the chip-renting companies, a bank steps in with money due in a year, and the gap between a short loan and a long-lived asset moves onto a regulated bank's books.

Nebius, which builds data centers and rents out Nvidia chips, said on Thursday it has secured a $3.5bn loan from Citi to pay for construction and chip purchases, Reuters reports. The loan has to be repaid in 364 days and starts at 2.75 percentage points over the benchmark rate banks use for short-term lending (roughly 6.6%, with that benchmark, term SOFR, at 3.88% today). The company already owes about $10bn on bonds that can be swapped for shares, after selling $5.75bn of them in August, and its shares are down 24% in a month and 31% from their high.

Look at the repayment date. A loan due in a year is a bridge, not a source of funding: it buys chips that lose value over several years, and it will be repaid, on the company's own account, from "further fundraising to support growth in 2027 and beyond", customer prepayments and the cash the business throws off. Chief operating officer Ophir Nave said Nebius could also borrow against more than $40bn of long-term contracts with customers graded safe to lend to, including Microsoft and Meta. That is the same thing Meta's filing describes from the other side today: the contract that is a cost for the tenant is what the landlord pledges to its own lender.

The reason a one-year bank loan is doing this job is that the stock market has stopped funding these companies. Bloomberg reports that Nscale, the Nvidia-backed British company in the same business, is grappling with delays, equipment failures and power constraints at some sites, and with investors balking at the $30bn value it wants to list at; Jensen Huang said on Thursday he told the company to "work around the clock". Firmus Grid abandoned a $30bn stock listing in Australia and Elice had a difficult first day of trading in Korea. Huang said demand was high and these companies had excellent customers.

He may be right about demand. The question the Nebius loan raises is who carries the risk between now and the next round of borrowing. Until this week the answer was investors in those swap-for-shares bonds and buyers of new listings. For the next year it is Citi, a regulated bank, the one place this site's thesis says risk had moved out of. A bank lending where the stock market will not is either a vote of confidence or a bridge to nowhere, and the repayment date is when we find out which. For the meter this changes little in total: the borrowing is not new, it has changed address, and the new address is one regulators can see.

Hidden leverage and shadow banking22 Octhigh

SoftBank loses half its value; the bets against it are crowded

The AI boom's most indebted listed company has lost half its value while the bets against it hit a five-year extreme, so the next move will be a forced one.

SoftBank shares fell 6.9% on Thursday, taking the fall from their June 2 record to 50%, Bloomberg reports. Four months ago the group briefly overtook Toyota as Japan's most valuable company; at ¥24.7tn ($157bn) it is now smaller than Hitachi and Recruit. On Monday the cost of insuring its debt against the company failing to pay over five years reached a four-year high ($4.32 a year for every $100 insured, or 4.32 percentage points).

SoftBank matters here because it is the biggest listed bet on the AI boom made with borrowed money: a holding company whose loans are secured on what it owns, and what it owns is stakes in AI companies. Kazuyuki Muramatsu of Nagomi Capital put the loop plainly: if the mood on AI sours, the Vision Fund has to admit its holdings are worth less, which raises what lenders charge SoftBank the next time it borrows. Those higher borrowing costs are what the shares are now pricing in. This is what makes holding companies built on borrowed money dangerous in a downturn: the thing pledged to the lender and the borrower are the same thing, so a fall in the asset tightens the credit at exactly the moment the asset needs patient money.

The other side of the bet is now crowded too. Goldman's desk that lends to hedge funds says those funds have bet on falls in about five times as much stock as they have bought this month. The borrowed money behind those bets against stocks is larger than in 99 of every 100 months over the past five years (the 99th percentile), while the net exposure of funds that bet both ways is near a three-year low. Unwinding has been expensive: funds focused on technology, media and telecoms are down 7.4% in October. Goldman's note floats "short squeeze risk alert" with a question mark, meaning a scramble by those bettors to buy back stock if it bounces, and Bloomberg observes that the bets against were added during a sell-off with identifiable real-world causes.

So two things are true at once. The biggest borrowed-money bet on the trade has lost half its value, and the bets against the trade are as heavy as they have been in five years. Positioning like that does not tell you direction. It tells you the next move, whichever way, will be forced. That is ignition rather than fragility: no new debt was added, but the chance of a sudden, involuntary move got closer.

Hidden leverage and shadow banking22 Octmedium

Highland Spring's owner defaults, and Apollo is one of the lenders

A prime-property empire built on borrowed money is defaulting with a non-bank lender among its creditors, which is where this cycle's untested valuations live.

The property empire of Mahdi Al Tajir, the billionaire owner of Highland Spring, has been hit by loan defaults and by creditors moving to seize assets, the Financial Times reports. The family is looking to sell chunks of a portfolio that runs from luxury developments in London to estates in Perthshire; last month the 94-year-old put a 620-acre golf resort near Gleneagles up for sale, a course finished in 2009 that never opened to the public, with a hotel and housing that were never completed. National Bank of Kuwait lent £12.75mn against it, according to the bank's 2024 accounts.

The names of the lenders are the point. The FT says the family's property companies have borrowed heavily in recent years from Gulf banks and from US institutional investors including Apollo Global Management. Apollo is not a bank. It is a fund manager lending alongside an insurance company's own money, and the price its loans are carried at is set by the lender, not by a market. A default on a prime London development or a Scottish estate is therefore the kind of event this site's thesis cares about: a loss in a place that does not report it the way a bank would, surfacing only because a creditor has moved to collect.

Trophy property is the kind of asset most dependent on cheap money. It produces little income, it is hard to sell quickly without cutting the price, and its value rests on the next buyer accepting an even smaller return. With the US government paying more than 5% to borrow for ten years, and France now with it, that buyer is scarce, which is why a resort that was never opened is the thing being sold. The FT's account does not give the size of the total debt or how much sits with Apollo rather than the Gulf banks. What it does show is an owner of prime assets, bought with borrowed money, selling under pressure from creditors with a private lender among them: the first visible test of such a valuation in that corner of the market this month. For the meter this does not make a crash more likely. It is the other kind of story, where we find out how bad something already was.

Bond market dysfunction21 Octhigh

Argentina at 12%: where global funds cut first

The AI sell-off is being paid for in the countries global funds cut first, and a government with $25bn due next year has just lost its market price for borrowing again.

Argentina's dollar bonds have been the worst performers among developing countries' debt over the past two months, Bloomberg reports, with the interest rate the country would pay pushed to nearly 12% earlier this week and the extra it pays over the US government at its highest in more than a year. The country owes roughly $25bn in payments on debt in foreign currency next year, and the presidential election is a year away.

The move surprised the people who own the bonds. Vontobel, Carmignac, Schroders and ProMeritum had all said the debt was too cheap to ignore. "Most large investment funds were overweight and surprised by the magnitude of the selloff," Ramiro Blazquez of StoneX told Bloomberg. "Argentina is one of the first places investors cut exposure when they reduce risk because it's a higher-risk credit, and that magnifies the move." Overweight means they held more Argentina than their benchmarks called for; they were betting on it.

Two mechanisms are at work. The first is arithmetic: a government paying 12% cannot sensibly borrow again to pay off $25bn in the bond market at that price, so the money has to come from somewhere else, reserves, official lenders or a deal to pay creditors less, and the market is pricing the uncertainty about which. The second is who owns it. Bloomberg reports that foreign money managers hold substantial positions while Argentine investors have been cutting theirs, leaving fewer buyers when the appetite for risk drops. A large position concentrated in foreign hands is the shape that falls fastest, because the holders all want to reduce at once and the local buyers who would normally absorb them have gone.

This is how an AI sell-off in New York and Seoul becomes a funding problem in Buenos Aires. Nobody has reassessed Milei's handling of the budget this week; the bonds fell because they are where global funds cut first. For the meter this is ignition, and a small step: a borrower with $25bn due next year has lost its market price for borrowing again, and that is worth remembering when we look at the next borrower that depends on a crowd of foreign holders staying put.

The AI capex bubble20 Octhigh

Insuring Oracle's debt has never cost more, and the borrowing has not even started

Bond investors are already charging for the next $450bn of AI borrowing before it is sold, while share investors are still pricing the demand it is meant to fund.

Insuring $100 of Oracle's debt against the company failing to pay cost as much as $3.43 a year on Tuesday, more than it ever has, and the same measure for Meta rose as much as 0.063 percentage points to an all-time high of 1.148 percentage points, according to ICE data cited by Bloomberg. Companies worldwide have already borrowed more than $450bn this year to fund AI. Meta and SpaceX report earnings this week, and once the big tech names come out of the period in which they cannot sell new debt, bondholders expect them to come straight back for more. Lenders are already charging for borrowing that has not arrived.

The insurance moves first for a reason. A bondholder who owns so much of a bond that selling it would push the price down buys protection instead, so the insurance is where fear shows up before the bonds themselves. Pimco's Lotfi Karoui put it plainly: "We have yet to find some kind of a landing zone for the level of debt issuance that the market needs to absorb." In other words, nobody yet knows how much new borrowing the market can take before buyers stop turning up. Bonds that already exist are being repriced for bonds that do not. Oracle is the barometer because, of the giant cloud companies, it funds its building most heavily with debt, and it borrows against contracts it has signed rather than cash it has earned.

Notice that the stock market disagrees. Oracle shares rose 1% on Monday and the S&P is within a percent of its record, on the same day insuring Oracle's debt cost more than ever. Shareholders are pricing the demand for AI; lenders are pricing who pays for it.

One more layer. The Wall Street Journal reports, from lien filings in California, that Larry Ellison called off a $7.5bn sale of Oracle stock this summer and instead borrowed more from Citigroup, HSBC, JPMorgan, UBS and BNP Paribas, pledging Oracle, Tesla, SpaceX and Paramount shares as security, to fund his son's bid for Warner Bros. Oracle disclosed in September that 413 million of his shares, worth around $58bn, are pledged against his personal debts, up from 346 million a year earlier. That is not a new risk; it was standing already, and what changed is that we can see it. It is worth knowing that the founder's security and the company's credit are the same stock, now 13% below its high.

On the meter, the sum of this is a spark creeping closer, not more tinder. The debt was all there last week. What is new is that the market has started charging for the next round before it is sold, and if that price keeps climbing through this week's deals, it is the first place a wave would break.

The AI capex bubble20 Octmedium

Anthropic's listing papers: $750bn of computing to pay for, and a fifth of the revenue from two customers

The largest AI stock sale yet rests on a customer list that is a handful of names and a computing bill larger than everything all companies have borrowed for AI this year.

Anthropic's draft papers for selling shares to the public, seen by the Wall Street Journal, say the company could spend up to $750bn in the coming years on computing capacity, including deals it is still negotiating. That is more than every company in the world has borrowed to fund AI this year put together ($450bn). Revenue in the first half of 2026 was about $16.3bn, up from $1.1bn a year earlier, and spending on computing and infrastructure was nearly $9bn, up from $2.3bn. The loss for the half was more than $80bn, against $13bn a year earlier, a figure the filing says largely reflects an accounting adjustment tied to financing agreements.

Take the three numbers in order. The revenue growth is extraordinary, and we mean that literally: almost no company of any size has ever multiplied its sales like that in a year. Around a fifth of it came from just two customers, and the filing says a "significant portion" will come from a limited number of customers for the foreseeable future. That matters well beyond Anthropic, because this is the demand behind every data-center bond we have written about this month. The lease payments that back those structures are, several steps removed, these two customers' invoices.

The $750bn is the other side, the supply. It is not a debt yet: it is a ceiling that includes deals not yet signed, and the filing lists uncertain access to computing among its risks. What has to stay true for a number that size to look prudent rather than reckless is that revenue keeps compounding at this pace.

The $80bn loss is the one we cannot explain. The filing attributes it to financing agreements, which suggests a paper charge on instruments issued to investors rather than cash leaving the building, but that is our inference and we may be wrong. Until the final filing is public, the honest statement is that Anthropic's accounts can swing by tens of billions on how its own financing is valued, which matters for a company aiming at a $2tn valuation.

On the meter this is tinder rather than a spark, and it is mostly tinder we are measuring rather than tinder being added. The spending is prospective. What the filing gives us is a fact we did not have: the demand at the top of the whole AI financing chain is, for now, two customers wide. The pile did not grow today, but we now know how narrow its base is.

The AI capex bubble21 Octhigh

NextDC stops building; Nvidia pays for what the stock market would not

When a builder freezes, a stock sale is cut in half and the chip seller funds the customers public investors turned away, the money behind the buildout has moved from markets that price losses every day to places where nobody has to.

NextDC, the data-center builder listed on the Australian stock exchange, lost as much as 45% of its value in Sydney on Wednesday after cutting its earnings forecast by more than a third, halting all new development and land purchases, and saying it is considering whether its contracts allow it to claim force majeure, the clause that excuses a company from delivering because of events outside its control, according to Bloomberg. Companies cut forecasts all the time. A builder stopping every new project and asking whether it can be excused from its own contracts is not that. The board is also weighing selling development sites. "We are not going to continue any work on any sites that don't have an existing customer commitment," said chief executive Craig Scroggie.

The cause is not a shortage of customers. It is power: community opposition delayed electricity to its M3 site in Melbourne, and the company could not deliver capacity it had already sold. Commonwealth Bank estimates Australia's buildout could reach A$150bn by 2030, on a grid retiring coal plants faster than it adds storage. A builder claiming force majeure against its own customers is telling the market that the physical limit has beaten the financial one.

The same day brought two more signs that the stock market's door is narrowing for these companies. DayOne, the Singapore data-center operator backed by the investment firms Coatue and Hillhouse, is weighing delaying its US stock market listing or cutting it to $2.5-3bn, roughly half the $5bn it had planned at a $20bn valuation (Bloomberg). And Nvidia is in talks to invest hundreds of millions of dollars in Firmus, the Australian company that rents out AI computing and abandoned what would have been one of the country's biggest stock listings this month, The Information reported via Bloomberg.

Put them together. Public investors declined to fund Firmus at the price offered; the company whose chips Firmus buys is now proposing to fund it instead. That is the supplier replacing the market as the source of its customer's money, and it has a name, vendor financing. On the meter it is tinder moved rather than removed. A listed share is priced every day and a loss on it shows up that afternoon; a stake sitting on Nvidia's balance sheet, the list of what it owns and owes, is priced when Nvidia says it is. The risk did not shrink. It went somewhere harder to see, and that makes the pile bigger, not smaller.

Bond market dysfunction20 Octhigh

France now pays more over Germany to borrow than at the height of the euro crisis

The euro area's second-largest economy is being charged a premium last seen when the currency was fighting for its life, and there is no rescue fund big enough for it.

France closed Tuesday paying more extra interest than Germany to borrow for ten years than on any day since January 2012 (the gap closed above 1.5 percentage points), Bloomberg reports. We have noted the gap passing that line during the trading day in the last two editions; a close above it is the threshold crossed, because the closing price is where the market settles rather than where it flinches. January 2012 was the depth of the euro crisis, and France is a far bigger borrower than the country at the center of that one.

Sonal Desai, who runs bond investing at Franklin Templeton, said: "I don't think anyone is really ready for a eurozone debt crisis version 2.0, and this one could be significantly worse if it were to come to pass." Her arithmetic: France is a €3tn economy carrying about €3.6tn of debt. "France has more than 10 times the debt that Greece had. A simple bailout is not on the cards." Franklin's funds lean toward Italian and French bonds that come due soon rather than in decades, which is what you hold when you want to be out before anything breaks.

The mechanism that matters is the one that nearly broke the euro in 2011. Banks hold their own government's bonds, so when those bonds fall in price the banks' cushion of capital shrinks, and weaker banks need rescuing by the same government that is getting weaker. Desai herself lists the differences: France is still graded a safe borrower by the firms that rate them, its banks hold more capital than in 2010, and the European Central Bank has tools it did not have then. Those are real. But the premium investors demand to hold France over Germany is where it was when Greece was restructuring, and it got there without a single rating firm downgrading France in the window.

This is a spark getting closer, not new tinder; France's debt was all there last week. What we are watching is not the size of the gap but whether French banks' own borrowing costs start to follow it. That is the loop closing, and it is the thing that would move ignition rather than fragility.

Household credit20 Octmedium

Car-Mart heads for court; the car loans it sold to investors stay whole

Bundling loans and selling them on is doing its job, the bondholders keep the cars, but the company that made the loans fails, a private lending fund inherits it, and the jump in long-term borrowing costs has reached mortgages and home sales.

America's Car-Mart, a used-car dealer that lends to buyers with poor credit, is working with its lenders on a sale that would run either through bankruptcy court (Chapter 11, the process that lets a company keep trading while its debts are sorted out) or a shareholder vote, Bloomberg reports. The company failed in June to pay what it owed on a $300m loan from Silver Point Capital, a loan it had taken to pay off a credit line secured on its car loans, and has been living on week-long reprieves from its lenders since; the latest runs to October 22. Living a week at a time on a lender's patience is not how a company survives. It is how one is sold.

The interesting part is what survives. The company's bundles of car loans, sold in slices to bond investors, are expected to be left untouched, and are quoted at around 98 cents on the dollar. That is close to whole, and the price implies there is more car loan inside the bundles than there is bond, a cushion built in at the start. So the structure does what it was designed to do. The loans were carved out of the company years ago, the bondholders keep them, and the corporate shell that made them fails around them. The loss lands on shareholders and on Silver Point, which in a court sale typically ends up as the owner. A fund that stepped in as lender of last resort in June is the likely proprietor in November.

Whether this is one weak lender or the start of something is not yet visible in prices: the shares of Credit Acceptance and Ally, two other car lenders, are down 4% and 3% over 20 days, nothing like Car-Mart's path.

The backdrop is the other end of household borrowing. Pending US home sales fell 7.3% in September to the lowest level since records began in 2001, lower than in 2008, as the interest rate on a thirty-year mortgage climbed above 7% to its highest since November 2023, the Financial Times reports. Housing starts fell to 1.22mn. A government paying five percent to borrow is no longer a story about bond traders; it is the number on a family's mortgage quote.

On the meter, Car-Mart is a small piece of tinder being resolved in an orderly way, which takes a little off the pile. The housing numbers are the opposite kind of fact and the larger one: evidence that the cost of long-term borrowing has reached households, which is where a slow problem turns into a fast one.

Hidden leverage and shadow banking21 Octmedium

Three companies hand the keys to their lenders

Lenders taking over companies is the orderly way the pile gets smaller, but the same few distressed-debt funds now own the outcomes, and even the strongest banks are paying a full percentage point more for capital than a year ago.

Three companies built on borrowed money moved toward being owned by their lenders in the same 24 hours. None of them is a crash. Each is how debt actually leaves the system when it goes well: in a court or across a negotiating table, one company at a time.

Optimum, Patrick Drahi's US cable business, is closing in on a restructuring that would hand control to its lenders and end the litigation between them, with the gap between what Drahi demanded for his stake and what creditors would give him having narrowed; a deal could close before the end of November, Bloomberg reports. This is the fight in which Optimum accused its lenders of breaking competition law for banding together, and in June warned them they risked triggering more than $4bn of tax liabilities. The lenders held together anyway, and that is the point: a borrower's best weapon is splitting its creditors, and it did not work.

Getty Images filed for bankruptcy protection in Delaware with the terms already agreed with its creditors, listing $1-10bn of assets and the same range of liabilities, after skipping an interest payment due September 1 on its unsecured notes; up to $533m of new money from the Getty family and Koch Equity Development takes it private, with suppliers paid in full (Bloomberg). The business was undermined by customers using AI-generated images instead of stock photos. Getty is an AI loser rather than an AI borrower.

Telesat has about US$1.71bn coming due on December 6 and has said its cash flow cannot cover what it owes. Lenders including Silver Point, GoldenTree and Sound Point tabled a second proposal on October 14 that would repay them in full through a $1.7bn sale of new shares to existing holders at $31.50 and a $1.5bn loan; Telesat has acknowledged the proposals but not negotiated, a source told the Globe and Mail. The shares fell 9.8%.

Silver Point appears in two of today's workouts and in Car-Mart's, which tells you how few hands these outcomes now sit in. The price of capital for the strong is rising too: Bank of America sold $2bn of notes that never come due at 7.2%, against 6.25% for the same instrument in July 2025, and did so days after raising $8bn (Bloomberg). On the meter, the three handovers take tinder off the pile, a little at a time, and that is worth saying plainly. The bank number is the part that does not. When one of the best-rated borrowers in the country pays a point more than it did a year ago, the cost of everything beneath it is rising with it.

What would change our mind

The specific, observable things that would move the number - in either direction.

The next readings on what shaky companies pay to borrow: if the extra interest over the government keeps widening through 4 percentage points, Thursday's stress is spreading beyond AI and airlines; if it drifts back toward 3.2, Friday's calm was real.

Would move the number

JetBlue's call on Tuesday, specifically any disclosure that the card processors are demanding security against ticket cash, which would turn a story about a bond price into a story about cash running out.

Would move the number

Whether the SpaceX vehicle finds a price in the next fortnight, and how much extra interest it has to pay; a failed or sharply repriced deal from a top-rated borrower would shut the template for everyone below it.

Would move the number

The French budget debate next week and Moody's warning of a possible downgrade: a defeat on the main bill with French government bonds failing to catch any of the rush to safety would move the pressure on France's debt from paused to live.

Would move the number