Twelve banks, double the ask, GPUs as collateral

Archived reading, published Wed, 02 Sep 2026 10:25:41 UTC (18 days ago). This is not the current state of the meter.

See the live reading →

CRASH-O-METER

Cracking

how close are we

Fragility91

how much tinder is stacked up — moves slowly

Ignition42

how close a spark is — moves fast

Crash Lab watches the machinery under the current boom: the debt paying for AI data centres, private credit, the leverage that has moved out of banks into places nobody has to mark, the bond market and the dollar.

Every four hours it reads the day’s reporting from 22 sources and rewrites this page.

Status: Held at 67. Ignition stays at 42: the stock market is still sitting on its September 1 close, the VIX (the market's gauge of how much turbulence traders expect over the next month; under 20 is calm) reads 16.77, the extra interest that shaky borrowers pay compared with the government is 2.63 percentage points and tighter than a month ago, and the move in long-dated government bonds (the ten-year rate at 4.81%, the thirty-year at 5.27%) is the continuation of a selloff we already counted across three earlier runs, not a new one. Fragility holds at 91: the Bank of England's highest-risk collateral data (£17.8bn, up from under £1bn in mid-2024) is a better measurement of consumer-credit risk that was already standing, offset by a GPU-backed loan in Asia clearing at more than double the ask, an untested funding channel that opened rather than closed.

Reporting from 9 Jan to 2 Sep

Crash points

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

What’s moving the needle

What changed in the last few hours, and what each one says about the plumbing.

The AI capex bubble2 Sephigh

The banks want the GPU loan. The stock market doesn't.

When the chip vendor guarantees the cash flow that repays the loan that buys its chips, bank credit committees are pricing Nvidia risk without holding Nvidia paper.

GMI Cloud, an Nvidia partner, went to Taiwanese banks for a five-year loan backed by the revenue from its chips. It asked for NT$13.9bn. About a dozen banks offered roughly NT$30bn ($947m), and Bloomberg reports that if verbal commitments firm up the book grows to about NT$40bn. The money buys GPUs for a data center in Taiwan and is repaid out of the revenue they generate. Read the terms before deciding what the oversubscription means. Nvidia supplies the chips. Nvidia will also lease any unused computing capacity at an agreed price for as long as six years. And Nvidia takes half the revenue GMI Cloud earns from other AI companies. So the lenders are not really betting on GMI Cloud's ability to sell computing power. They are betting on a floor that the chip vendor itself has agreed to provide, on chips the same vendor sold. It is a guarantee on the residual value wearing the clothes of a commercial contract, and it does not appear as debt anywhere on Nvidia's books. This is the pattern we keep finding: the risk is real, someone has taken it, and it sits outside the place where risk is counted. The interesting part is the disagreement. Bank lenders are bidding twice the ask for exposure to GPU cash flows in Asia. Meanwhile the listed companies that build and run these assets have been sold hard: Core Scientific is down 29% in twenty days, Applied Digital 22%, Nebius 12%, CoreWeave 11%. Nvidia itself is up 2.6% over the same stretch. Stock investors are marking down the operators; bank lenders are lining up to lend against the operators' hardware, secured by the vendor. One of those two groups is wrong.

Hidden leverage and shadow banking2 Sepmedium

Store-card loans, pledged at the Bank of England

A twenty-fold rise in consumer-credit paper pledged at the Bank of England is a measure of what UK banks would rather not have to sell.

On August 18, British banks pledged £1.9bn of the Bank of England's highest-risk category of collateral at its weekly auction for six-month funds, the most since March 2020 and three times the previous week. The category, which the Bank of England calls Level C, includes loans linked to high-interest store cards and vehicle leases. Reuters' review of Bank of England filings puts the total stock of Level C collateral from the Indexed Long-Term Repo at about £17.8bn, up from £8.7bn a year ago and under £1bn in mid-2024. The mechanism is straightforward. Quantitative easing put cash in the system; the Bank of England has been reversing £895bn of it. Banks still need central bank reserves for wholesale settlement, so they borrow them, and they post whatever assets they have. Government bonds are the cheap option. When a bank is posting bundled store-card receivables instead, it is telling you something about the composition of what it has left, or about how much it would rather not sell. Two things stop this from being alarming on its own. The Bank of England says the facility is designed to accept a broad range of assets and that it protects itself through discounts and risk management, and £17.8bn is small against a system of that size. The Bank calls the growth an intended consequence of unwinding quantitative easing, and it is. What has changed is that the central bank has become a standing window where UK consumer credit can be swapped for cash, a class of asset the European Central Bank has tightened its own rules to exclude, on the reasoning that a central bank stamp of approval creates demand for paper that is hard to sell in a crisis. This is a twenty-fold increase in two years in a channel almost nobody was watching. We did not know its size this morning. Now we do.

Bond market dysfunction2 Sephigh

The buyback has been fully round-tripped

The market has tested the Treasury Department's response to rising long-term interest rates and found it does not change the price.

On August 19 the Treasury Department, which borrows the money the government spends, said it would at least double its buybacks of older government bonds, to $4bn per operation, starting September 9. The interest rate the government pays to borrow for thirty years fell. By Tuesday it was back at 5.27%, the level it traded at in the moments before Treasury Secretary Bessent announced the move. The ten-year rate is more than a tenth of a percentage point higher than it was then, at roughly 4.80%, and touched 4.81% in Asian hours on Wednesday, the highest in nearly three years (Bloomberg). The round trip is worth understanding, because it is not a failure of execution. Buybacks retire old bonds that trade thinly and fund the purchase with new borrowing. They improve trading conditions in those older lines. They do not reduce the amount of debt the market has to absorb, and they do nothing to the government's spending path, the inflation path, or the volume of AI-related corporate borrowing now competing for the same buyers. As Bank of America's Mark Cabana put it, "the rates market has not been able to hold any type of significant rate decline." One consequence is showing up in Asia. The gap between US and Chinese ten-year interest rates has widened to 3.12 percentage points, against a record of roughly 3.15 percentage points, as China's ten-year sits still at 1.69% (Bloomberg). The yuan has barely reacted, trading near its strongest since early 2023. Worth stating plainly: this is investors demanding more compensation for tying their money up for a long time, not visible plumbing failure. The evidence we have does not show poorly received auctions, dealer strain, or dislocation in the derivatives that track government borrowing costs, and the Federal Reserve's Kashkari has said he sees no sign the selloff is making the Fed's job harder. What it does show is that the one tool the Treasury Department reached for has now been priced out entirely, in ten business days.

Fed, Treasury and policy2 Sephigh

Japan budgeted for 3%. It just got 3%.

Japan's domestic interest rate is now competitive with the foreign bonds its institutions have spent thirty years buying.

Kazuo Ueda, the head of Japan's central bank, told reporters after the G20 meeting in Asheville that a rate rise "will be discussed thoroughly in every meeting, including the next one," referring to the September 17–18 meeting (Nikkei). Japanese money market pricing puts a 94% probability on a 0.25 percentage point move, bringing the rate to 1.25%. The number that matters more sits in a budget document. Japan assumed a 3% long-term interest rate when it calculated debt-servicing costs for the fiscal 2026 budget. On Tuesday the ten-year Japanese government bond hit exactly 3%, for the first time since September 1996, and it has more than tripled in two years. The five-year touched a record 2.265%; the two-year reached a 31-year high of 1.81% (Channel NewsAsia). Above 3%, the arithmetic in the budget stops working, on the largest government debt pile in the developed world. The transmission to everyone else runs through Japanese savings. For three decades, domestic institutions with no meaningful interest rates at home bought foreign bonds: US Treasuries, British gilts, French government debt, Australian paper. That flow was one of the quieter supports under demand for long-dated government bonds worldwide. At 3% domestic, with the currency near a four-decade low and a rate hike arriving, that trade is worth less. The yen is still weak despite a joint US-Japan intervention on July 31, which tells you how much force has been applied and how little it bought. We are not claiming a disorderly unwinding is underway; the evidence does not show one. But the buyer at the margin of long-dated bonds everywhere has just been offered 3% at home, free of currency risk, with a central bank that says it will keep raising.

Household credit1 Sephigh

Australia is running the whole experiment at once

It is the clearest live example of the loop from higher rates to developer insolvency to private lending funds shutting the exit door.

Sydney house prices are down about 7% from their February peak. Values are falling in 93% of suburbs across Australia's capital cities. Mortgage applications at the big banks have dropped as much as 20% since the May budget removed tax incentives favoring property investment, and sales volumes in Brisbane, Perth, and Sydney are off more than 20% year on year (Business Times). About 60% of Australian household wealth is tied up in property. What makes this worth a reader's attention outside Australia is that the country has, in miniature, every part of the structure we track. Rising global long-term rates raised borrowing costs. A tax change removed a buyer at the margin. Prices fell. On August 25, one of Sydney's largest developers, Bathla Group, went insolvent owing A$3.3bn. Its lenders were not banks. They were private credit funds, the kind that lend to companies directly rather than through the banking system. Ray White Capital, with roughly $242m lent to Bathla, told investors this week that the homes will be released into a market where they are worth up to 15% less than assumed, and reminded them that they were "never promised the opportunity to redeem their investments" because the funds are closed-ended (AFR). The open-ended funds have had to be more explicit. MA Financial capped redemptions on its A$2.3bn secured property loan fund at 1% of funds under management a month; CVS Lane suspended applications and redemptions outright (Fund Finance Friday). That is the full chain in four weeks: rates up, the property backing the loans down, borrower fails, values marked down 15%, exit door closed. Nothing here is systemic. This is a A$200bn market, fund-specific, and Trilogy has said its own redemption procedures are unchanged. It is a working demonstration of the sequence, at a scale small enough to watch.

Crypto and TradFi contagion1 Sepmedium

Bitcoin fell 1%. The wrappers fell six.

The stock-market wrappers around crypto carry a premium that can vanish without the underlying coin moving at all.

On Tuesday bitcoin was down 1.08%. Strategy, the original company that holds bitcoin as a corporate treasury asset, fell 6.02%. BitMine fell 7.77%, SharpLink 5.56%, Coinbase 5.97%, Galaxy Digital 4.72%. That ratio is the entire business model, seen from the wrong end. Over the last twenty days bitcoin is up 20.8% and Strategy is up 28%, roughly 1.3 times. On a single down day the multiple was closer to six. The gap is not just borrowed money amplifying the coin price; it is borrowed money plus a premium. A company like Strategy trades at the value of the coins it holds, plus whatever the market pays for the promise of buying more coins on favorable terms. The second part is not a hard asset. It can compress to nothing in an afternoon without a single coin moving, and when it compresses the company loses its ability to issue new shares above the value of its holdings, which is the only mechanism by which the promise was ever going to be kept. We cannot tell you where these premiums stand today. Our sweep of the week's reporting found no sourced item establishing that any of these companies is trading below the value of its holdings. Absence of a report is not evidence; it means the disclosure is not there to check. Meanwhile the connective tissue keeps thickening. Coinbase and Better have launched bitcoin-backed mortgages after a waitlist projected more than $260m of demand (weekly recap). A house bought with borrowed money against a volatile asset used as security is a specific structure with a specific failure mode, and it is now a product you can apply for.

Signs of the times

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

Seven months old, shopping $5bn

Volta Infra Holdings raised $300m at a $2.4bn valuation in early August, backed by Andreessen Horowitz, Altimeter, Nvidia, and Michael Dell. On August 27, JPMorgan began sounding out lenders for a $5bn debt package for its data-center buildout. That is twice the company's entire equity valuation, three weeks later, at a company founded about seven months ago.

ai capexleveraged finance

1 SepBloomberg

Japan budgeted 3%; it arrived

The Japanese government assumed a 3% long-term interest rate when calculating debt-servicing costs in its fiscal 2026 budget. On Tuesday the ten-year Japanese government bond hit exactly 3% for the first time since 1996, having more than tripled in two years.

sovereign debtjapan

9 JanChannel NewsAsia

Your house, borrowed against your bitcoin

Coinbase and Better have launched bitcoin-backed mortgages, after a waitlist projected more than $260m in demand. Charles Schwab, which oversees $13tn, added Solana, Avalanche, and Chainlink to its trading menu in the same week.

cryptohousehold credit

1 SepBitcoin.com weekly recap

Abu Dhabi's 49% of a stablecoin bank

Sheikh Tahnoon bin Zayed, the UAE's national security adviser, and co-investors own 49% of the holding company of a US bank that has received preliminary approval to issue USD1, the dollar-pegged stablecoin (a digital token meant to be worth exactly one dollar) launched by the Trump-family-linked World Liberty Financial. His conglomerate separately completed a $30m transfer in a dirham-backed stablecoin, described by its backers as one of the largest single stablecoin transactions in the Middle East.

stablecoinssovereign money

31 AugSemafor

The rumour mill

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

Partly true

“Margin debt has soared 50% in the past year to $1.5 trillion.”

The level and the rate of increase check out. NPR and FINRA (the body that polices American stockbrokers) data put money borrowed by investors to buy stocks above $1.5tn in June, roughly 50% higher than a year earlier. The added framing that this is "double the rate of the market's 25% gain" is the commentator's own arithmetic and is not confirmed by any outlet we can find.

Claimed by Wealthion

Partly true

“The government just erased almost 900,000 jobs it had previously reported as created, which were fabricated to make the numbers look better.”

The revision is real: Bureau of Labor Statistics benchmark data imply roughly 911,000 fewer jobs for the twelve months to March 2025, and the prior year's benchmark cut 898,000. But the benchmark revision is a routine annual reconciliation of the survey against unemployment-insurance tax records, and the most recent one, published August 28, was minus 79,000. No reputable source supports fabrication.

Claimed by Gregory Mannarino

Partly true

“The Shiller CAPE ratio just reached 41 for the first time since 2000.”

Multiple outlets put the ratio (which measures how many years of inflation-adjusted profits investors are paying for the stock market) at roughly 41.0–41.4 through July and August 2026, the highest since around September 2000. The "just reached, first time" phrasing implies a single clean crossing that the published data do not show.

Claimed by Wealthion

Earlier developments

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

Private credit and BDCs2 Sepmedium

Two ways to count a bad loan

The gap between reported and borrower-level non-accruals is the clearest available measure of how much credit stress private lenders can carry without it showing in the headline metric.

At the ten largest listed business development companies, loans on non-accrual reached 3.95% of debt at cost at the end of Q2 2026, up 20 basis points on the quarter, with the balance rising $89m to $3.3bn (web sweep of PitchBook LCD data). That is the reported number. There is a second one. If you count every loan a BDC holds to any borrower where at least one tranche has stopped accruing — the performing senior piece as well as the failed junior piece — exposure rises to $5.0bn, or 5.95% of total debt at cost, up 54bp on the quarter. The mechanism matters. A BDC lending to a company through several instruments can stop accruing interest on the piece that has clearly failed while continuing to book income on the rest, because the rest is contractually still current. That is defensible accounting. It is also a way of describing a borrower in trouble using a number that is a third smaller than the amount actually at stake with that borrower. The borrower-level figure is rising almost three times as fast as the headline one, which tells you the deterioration is concentrating in names the funds already have multiple lines into. One honest caveat: part of the reported increase is arithmetic rather than credit. Total debt at these ten funds shrank 2.3% to $83.6bn over the quarter, so a static stock of bad loans would have raised the ratio anyway. The numerator grew about 2.8%; the denominator fell. Both moved the wrong way, but not equally. Separately, Jefferies put Q2 non-accruals at ARCC at 2.4%, Blue Owl's BDC at 2.8%, Golub's at 2.9% and Blackstone Secured Lending at 3.6% — against 0.3% for BXSL in Q1 2025.

Hidden leverage and shadow banking1 Sephigh

The managers fell four times harder than the funds

Private credit's marks move quarterly and its equity moves hourly, and the widening gap between them is where any repricing will show up first.

On 1 September the private credit complex split in two. Blackstone fell 4.61%, Blue Owl Asset Management 4.62%, Apollo 3.64%, KKR 3.08%, Ares Management 2.90%. The funds those firms run barely moved: Ares Capital down 0.67%, Blackstone Secured Lending down 1.16%, Blue Owl Capital Corp down 1.22%, and BXSL is still up 3.9% over twenty days. This is not inconsistency. The two securities are claims on different things. A BDC share is a claim on a portfolio of loans carried at fair value, most of them not traded, marked quarterly by the manager with an auditor's sign-off. Those marks do not move on a Tuesday. A manager's share is a claim on future management and performance fees — which depend on raising the next fund, deploying it, and not having to explain last year's marks to the people you are asking for money. That claim reprices instantly, and it is levered to sentiment in a way the fund's NAV structurally is not. So when the two disagree, the manager equity is usually the faster instrument and the fund NAV the more authoritative one. Neither is the truth. What is worth watching is the direction of travel: five-day moves have the managers down 1 to 4.7% while the BDCs are flat to slightly up. The market is repricing the fee stream, not yet the loan book. Against that, the actual credit data is going the other way, slowly — non-accruals at multi-year highs, $2.04bn of write-downs across 83 BDCs last quarter, the largest in a thirteen-quarter window. If the loan books were being marked at the speed the manager shares are being marked, the two lines would have converged already.

Fed, Treasury and policy2 Sephigh

Treasury doubles its buybacks; the market gave it a day

A buyback that cannot hold the long end tells you the pressure on 30-year yields is fiscal and inflationary, not a liquidity problem policy can fix.

From 9 September through 4 November, the US Treasury will at least double the maximum size of its long-dated buyback operations, from $2bn to $4bn per operation, in the 10–20 year and 20–30 year sectors. The announcement came in August, after the 30-year yield touched roughly 5.3%, its highest since 2007. What a buyback actually does is narrower than it sounds. Treasury purchases older, off-the-run bonds — issues that have drifted out of the current benchmark and trade thinly — and funds the purchase by issuing new debt elsewhere on the curve. Total debt outstanding does not fall. Net duration barely changes. The stated purpose is liquidity: giving dealers a reliable bid for paper that is otherwise expensive to hold on a balance sheet. It is a plumbing operation, not monetary policy, and it is not QE. The market has read it as something else. Long yields fell on the announcement and gave the move back within a day; the 30-year closed 1 September at 5.268%, and Reuters reporting has it recovering roughly two-thirds of its post-intervention fall. The 10-year is at 4.796%, a 19-month high. Meanwhile Minneapolis Fed president Neel Kashkari says he sees "no sign" the selloff is making the Fed's job harder and that the Treasury market is "functioning as it should". Both can be true. There is no sourced evidence this window of tailing auctions, dealer balance-sheet strain or swap-spread dislocation — the things that actually signal a broken Treasury market. What there is: a fiscal authority that felt the need to double a liquidity backstop, and a curve that ignored it. That is a market repricing term premium, not one that has stopped working. The distinction matters, and it is worth not blurring.

Household credit1 Sepmedium

Charge-offs up, delinquencies down, both true

The consumer is splitting: revolving card credit is healing while auto and student debt concentrate the damage in lower-income households.

Two consumer credit numbers moved in opposite directions in Q2 2026. The credit card charge-off rate rose from 3.66% to 3.70%. The 30-plus-day delinquency rate on bank-issued cards fell to 2.85%, and the all-card 60-plus-day rate fell to 2.69%. This is not a contradiction; it is a lag. A card balance goes delinquent, ages through 60 and 90 days, and is charged off at around 180 days. Today's charge-off rate is the tail of last autumn's delinquency rate. A falling flow into delinquency alongside a rising write-off rate is what the end of a stress episode looks like — the losses arriving are the ones already baked in six months ago, and fewer are being created behind them. On cards, this is genuinely the better of the two possible pictures. Auto is not the same shape. New York Fed Q2 data has 5.49% of auto loan balances 90-plus days delinquent, with 3.00% of balances transitioning into 90-plus days over the prior four quarters. On subprime specifically the sources in this window disagree: Equifax's Moody's update calls subprime auto delinquency an all-time high, while another says it is near a 15-year high and stabilising. We cannot reconcile the two, and the difference matters — one is a peak, the other a plateau. Student loans are the clearest deterioration: FICO puts 3.2 million borrowers, 15% of those with a payment due, at 30 days or more delinquent at some point in the previous six months. The market is not treating this as benign. Affirm fell 6.1% on 1 September and 10.3% on the week; Synchrony is down 5.2% over five days. Deep subprime auto is the odd one out — Credit Acceptance sits 1.9% off its high.

Crypto and TradFi contagion1 Sepmedium

Twenty-one banks, one stablecoin, no demand

Bank-issued stablecoins convert deposits into T-bill-backed claims, shrinking the lending base of exactly the institutions regulators can see.

A group of 21 institutions including Goldman Sachs, Bank of America, Citi and Deutsche Bank said on Tuesday they will form a company this year to issue a dollar-pegged stablecoin in the first half of 2027, with euro and other G7 currencies to follow. The group was ten banks when it was announced in October 2025. It will compete with a separate consortium of 37 institutions, Qivalis, planning a euro coin this year. BBVA is in both. The demand evidence is thin, and Reuters says so plainly: Société Générale, the first major bank to issue a dollar-backed stablecoin, has $12.5m in circulation. Tether, meanwhile, has issued more than $180bn. What is worth understanding is the balance-sheet mechanics, because this is a funding-structure story dressed as a technology story. A bank deposit is funded by the bank and lent out; a stablecoin is funded by the holder and backed by reserves, mostly Treasury bills. If a bank's customers move balances from deposits into that bank's own stablecoin, the bank's loanable deposit base shrinks and its holdings of short-dated government paper rise. That is a smaller, safer, less profitable bank. It is also why the incumbents would rather issue the coin themselves than watch Tether do it — the fee income on $180bn of bill reserves is the whole business. Elsewhere in the same window: Circle minted roughly 5bn USDC in a week, Schwab added SOL, AVAX and LINK, and Coinbase and Better launched bitcoin-backed mortgages after a waitlist projected over $260m of demand. Crypto collateral is now entering the housing credit chain. Coinbase fell 5.97% on 1 September; bitcoin rose 0.53%.

The AI capex bubble1 Sepmedium

The builders are down. The vendor is up.

Equity in the debt-funded AI buildout is now trading as a levered bet on long rates, which is the transmission channel from the bond rout into the capex cycle.

Over five sessions to 1 September: Applied Digital −15.9%, IREN −12.8%, Core Scientific −10.8%, Nebius −10.2%, CoreWeave −7.1%. Over twenty days Core Scientific is down 29.5% and Applied Digital 22.4%. Over the same five days Nvidia rose 2.1%, Microsoft 2.0% and Meta 1.5%. That split is the clearest thing on the tape right now, and it maps precisely onto who is taking the financing risk. The last fortnight's deals: JPMorgan sounding lenders on a $5bn debt package for Volta Infra Holdings; Blue Owl arranging $2.4bn for IREN — $1.2bn term loan, $1.2bn senior secured notes — to buy Nvidia Blackwell Ultra GPUs for a Canadian campus; Nebius upsizing a debt offering to $5bn after closing $5.75bn of convertible notes on 25 August; Lambda closing a $926m senior secured term loan B on 27 August. In each case the chips get paid for at closing and the debt sits on the buyer. Equity in a levered datacentre builder is the residual claim after that debt, and its value is the difference between two large numbers — contracted lease revenue over ten years, and interest plus depreciation over the same period. Move the discount rate up 100 basis points, as the long end has done, and the residual moves a great deal more than 100 basis points. That is most of what these five-day numbers are. The hyperscalers and Nvidia are on the other side of the same trades and are not being marked down. PIMCO puts hyperscaler and chipmaker bond issuance at $182bn this year; nine tech companies carry roughly $3tn of off-balance-sheet AI commitments. The market is pricing the borrowers, not the guarantors.

Fed, Treasury and policy1 Sephigh

The spike absorber is 41% full

With the oil buffer nearly empty, an energy shock now has to be answered with interest rates rather than barrels — into the most rate-sensitive debt stock in history.

Brent settled at $94.65 a barrel, up 4.6% on the day, and traded as high as $97.04 into the Asian session on 2 September — a five-week high. We checked both figures. The move follows the first direct exchange of attacks between the US and Iran in a month; on Tuesday only five ships transited the Strait of Hormuz, and European gas hit its highest since March (Guardian). The piece of plumbing worth understanding is what is not available to absorb it. The US Strategic Petroleum Reserve stood at 289.7 million barrels as of 21 August — its lowest since 1982, and about 41% of its authorised 714 million barrel capacity, on Department of Energy data. A further 39 million barrel release is planned. The reserve exists precisely to blunt a Middle East supply shock, and it is close to the level at which drawdowns stop being a policy tool and start being a logistical problem. So the shock passes straight through to prices, and from prices to inflation expectations, and from expectations to the long end. That is the chain that has the 30-year Treasury at 5.27%, gilts at a 28-year high and the 10-year at 4.80%. And it lands on a Fed chaired by Kevin Warsh, who said at Jackson Hole on 28 August that "two percent is a firm and fixed target" and that short-term rates are the primary tool. Futures moved last week towards pricing a hike on 16 September. The two-year at 4.34% against three-month bills at 3.77% says the front end agrees. A tightening cycle restarting into a $40tn federal debt stock is a different regime from the one we have been describing for most of this year.

The AI capex bubble1 Sephigh

Dell grew 1.7% and raised the year by $14bn

The most-watched pass-through vendor in AI hardware is guiding to a second half nearly 30% larger than its first, on a backlog whose buyers are partly funded by private credit.

Dell's AI-optimised server revenue was $16.4bn in the quarter ended 31 July, against $16.1bn in the quarter ended 1 May. That is roughly 1.7% sequential growth — and roughly 100% year on year. In the same release Dell raised its full-year AI server forecast to $74bn from $60bn, and its adjusted EPS guide to $25.50 from $17.90 against a consensus near $18.92. We checked all four numbers; they hold. Do the arithmetic. Two quarters have delivered about $32.5bn. Getting to $74bn requires something over $41bn in the back half — a step change, not a continuation. Dell says the orders exist: a record $60.9bn of AI server orders in the quarter and a $95bn backlog, roughly five times what it shipped in three months. So the guide is a claim about conversion, not demand. Two things could break it. The first is supply: COO Jeff Clarke told the call the company has "spotty CPU shortages" and "shortages with disk drives". In 2026 the constraint on artificial intelligence is partly hard drives. The second is who the customers are. A meaningful share of the neocloud tier buys with borrowed money — Iren took $2.4bn arranged by Blue Owl explicitly to buy Nvidia Blackwell Ultra GPUs, Nebius upsized a debt offering to $5bn, Lambda closed a $926m term loan (reported deals). A backlog is only as good as the financing behind it. Dell's $95bn is, in part, a claim on private credit's continued appetite. One genuinely surprising number: Dell's infrastructure segment converted 15.0% of revenue to operating income, up from 8.8%. AI servers were supposed to be low-margin metal. That is either mix or pricing power, and the filing does not say which.

Private credit and BDCs1 Sepmedium

The largest write-down in thirteen quarters, and the funds barely moved

Equity holders of private credit's fee machine are being marked down while holders of the underlying loan books are not — and only one of those groups gets to choose its own marks.

Across 83 business development companies, investments still held were written down by $2.04bn in the latest quarter — the largest quarterly write-down in a thirteen-quarter dataset. Positions carried below 90% of cost or on non-accrual rose from 0.9% to 3.2% of the tracked universe (Fund Finance Friday). The granular numbers are worse than the headline rate. At the ten largest listed BDCs, non-accruals reached 3.95% of debt at cost, up from 3.75%. Count every tranche owed by a borrower with at least one tranche on non-accrual and the exposure is $5.0bn, or 5.95% of debt, up 54bp in a quarter. Loans already on non-accrual are carried at 55.8 cents on the dollar, down 144bp. PitchBook LCD identifies $772m of interest "at risk of imminent default" — 204bp of total cash interest. That last figure is the one to hold onto. It is not a solvency number. It is roughly two percent of the cash coming in the door, which is what pays distributions. Dividends are covered. They are covered by less than they were. Now the prices. Ares Capital is 1.4% off its recent high and up 1.4% over twenty days. Blackstone Secured Lending is 1.3% off its high and up 3.9%. FS KKR is up 6.9%. The vehicles that actually own these loans are trading as though the marks are right. Meanwhile the managers were hit hard on Monday: Blackstone −4.6%, Blue Owl −4.6%, Apollo −3.6%, KKR −3.1%. The market is discounting the fee stream and accepting the loan book. We cannot make those two views sit together. One of them is wrong.

Household credit1 Sephigh

Ninety-three per cent of Australian suburbs are falling

This is the live experiment for the whole private credit model: what happens to unlisted loan marks when the collateral behind them is visibly falling.

Sydney house prices are down about 7% from their February peak, and values are now falling in 93% of suburbs across Australia's capital cities, according to Cotality figures reported by Business Times. Mortgage applications have dropped as much as 20% since the May Budget removed tax incentives for property investment. Sales volumes in Brisbane, Perth and Sydney are down more than 20% year on year. Around 60% of Australian household wealth is tied up in property, on Commerzbank's estimate. This is the same story as the Australian private credit gates, seen from the other end. On 25 August the developer Bathla Group went into insolvency owing A$3.3bn. Ray White Capital, with roughly $242m lent against it, told investors this week its homes will hit the market worth up to 15% less than carried, and reminded them that they were "never promised the opportunity to redeem their investments" because the funds are closed-ended (AFR). The mechanism is worth naming. A closed-ended fund cannot be run on. That is genuinely stabilising — nobody is forced to dump collateral into a falling market to meet redemptions. But it does not make the loss smaller; it converts a fast loss into a slow one, disclosed in quarterly letters rather than in prices. Investors get no exit and no mark they can trade against. Australia matters here out of proportion to its size because it is the only place where a private credit book lending against development land is being tested against a falling collateral market in public. A$200bn of industry, a 7% decline, one insolvency, two gated funds. Watch the second markdown, not the first.

Bond market dysfunction1 Sephigh

Japan's yields are up and its currency is down anyway

If Japanese money stops funding foreign bond markets, every long-end yield in the developed world clears at a higher level, permanently.

The yen traded at 160.15 per dollar on Tuesday, through 160 for a third consecutive session, on the same day the 10-year JGB yield touched 3% for the first time since September 1996. Five-year JGBs hit a record 2.265%, the two-year a 31-year high of 1.81% (Business Times). That combination is the whole story. Ordinarily a currency rises when domestic yields rise: higher returns pull capital in. When yields rise and the currency falls at the same time, buyers are not being attracted — they are demanding compensation. The market is reading Japanese yields as a fiscal risk premium, not a policy tightening. Ministries have just filed what domestic media report as the largest initial budget request on record, and the government assumed a 3% long-term rate when calculating debt service for the fiscal 2026 budget. It is now at that assumption. Treasury Secretary Scott Bessent said he believes Tokyo will act to strengthen the yen — which, as one strategist noted, may effectively lock the Bank of Japan into hiking this month. A rare joint US–Japan intervention in late July pulled the yen back from a 40-year low of 163.99; most of that gain is gone. For everyone else, the plumbing point is this. Japanese institutions have spent three decades as the marginal foreign buyer of other people's government bonds, because domestic yields were nil. A 3% JGB, hedged, removes the reason. The yen ETF is down 1.6% over twenty days, and the global long end has sold off in lockstep. The world's largest pool of price-insensitive capital is becoming price-sensitive.

The dollar, gold and reserve status31 Augmedium

Gold fell 7% in a week with a war on

Gold selling off into a war and a bond rout is usually a tell about leverage somewhere else, not about gold.

Gold is down 5.7% over five days to $4,338, the GLD ETF down 7.3%, silver down 7.1%. That happened during a week containing a Middle East escalation, Brent through $95, a global bond rout and the yen at 160. The classic hedge did not hedge. We would treat this as positioning rather than a change in the fundamental case, and we are marking that as inference rather than fact. Gold is still up 2.3% over twenty days and near record levels. When a crowded, profitable, extremely liquid long falls hardest in the week everything else falls, the usual explanation is that it is being sold to fund losses elsewhere — you sell what you can, not what you want to. The structural bid is a different set of buyers entirely, and it is still there. The Reserve Bank of Australia has become the first major central bank to confirm cutting its US dollar holdings, reportedly taking the dollar target share of its reserve portfolio from 55% to 45% and lifting the euro from 20% to 30% (AFR). India's reserves hit a record $729.3bn in the week to 21 August, with gold holdings up $2.8bn to $114.2bn. A Kansas City Fed working paper puts the dollar's share of global reserves at 57%, down from 72% in 2000. Central banks buy on a decade horizon and do not use leverage. Hedge funds do both the other things. The dollar index rose 0.6% on the week while gold fell — a combination that, if it persists alongside falling equities, is the signature of a liquidity squeeze rather than a rotation.

Crypto and TradFi contagion31 Augmedium

The coin was flat. The companies that hold it were not.

The treasury-company model converts equity market sentiment into crypto demand, and it only works in one direction — while the shares trade above the coins.

On Monday bitcoin fell 0.28%. Strategy fell 6.02%. BitMine fell 7.77%, SharpLink 5.56%, Coinbase 5.97%, Galaxy Digital 4.72%. Digital asset treasury companies are supposed to be levered proxies for the coins they hold. Over twenty days they have behaved exactly that way: bitcoin up 21.5%, Strategy up 28.0%, the ether treasuries up around 29%. On a single risk-off day with the coin unchanged, they traded like small-cap Nasdaq equities instead. Both facts are true and they describe different buyers — a coin-linked bid on the way up, an equity-risk bid that leaves first. This matters because of how the model is funded. Strategy resumed buying after a ten-week pause, taking $369.7m of bitcoin in the week to 30 August, financed with $602.8m of common stock sales, alongside a $151.8m buyback of preferred shares (Bloomberg). Issuing equity to buy coin only adds value per share if the shares trade above the value of the coins already held. Sell $602.8m of stock at a premium, buy $369.7m of bitcoin, retire some expensive preferred — that is accretive. Do the same at a discount and every issue destroys value for existing holders, and the machine has to stop. Our sweep of this window did not turn up a sourced, current measure of where any of these companies trade against the value of their holdings, which is precisely the number that decides whether the model keeps running. That absence is worth noting. It is the single most useful disclosure in the sector and it is not readily available.

The AI capex bubble1 Sephigh

The IPO that tells you how the AI trade is financed

This is the clearest single document yet showing that AI infrastructure debt is being underwritten on the chipmaker's balance sheet rather than the borrower's.

SB Energy, the SoftBank-backed power and data-centre developer, filed for an IPO on Tuesday. The numbers in the filing: net losses of roughly $3.2bn in the first half of 2026, revenue of about $139m — nearly all of it from the legacy energy business — and not one operational data centre. (CNBC) The risk factors are more useful than the accounts. SB Energy says it is "substantially dependent" on OpenAI, which is simultaneously its anchor tenant and an equity investor, and whose CEO was an early personal backer. The filing states plainly that "near-term revenues, project-level financing arrangements, and development plans are significantly linked to OpenAI's continued performance under our lease." Then the third leg. Nvidia said in August it would provide $105bn of financing for the OpenAI data centre in Ohio that SB Energy is building. The company's own CEO explained why the chipmaker is in the structure at all: it "helps us to unlock things like investment-grade financing." Read that as plumbing. A developer with no operating assets and $3.2bn of six-month losses cannot borrow at investment grade on its own credit. It can borrow at investment grade against a lease from a private company that does not publish accounts, provided the chip vendor stands behind the paper. The credit being priced is Nvidia's. The cash flows are OpenAI's. The equity risk is about to be sold to public markets. The filing does not say how Nvidia's $105bn is structured — guarantee, backstop, direct lending or some combination — and that distinction determines who eats the first loss. It is the single most important unknown in the document.

The AI capex bubble1 Sepmedium

Texas was asked for ten times the power it has

The AI power shortage that justified a re-rating of the entire independent power sector was measured with an instrument that costs nothing to lie to.

Texas has frozen new grid connections for data centres while it works out which of them are real. The reason: requests from very large power users in the state went from about 48 gigawatts in 2023 to more than 474GW now. Across the Midwest, Mid-Atlantic and South, Reuters counts requests exceeding 700GW — more than ten times industry estimates of what US data centres actually consume today, and roughly what it takes to power every home in the country. (Reuters) Pennsylvania supplies the cleanest test. Of more than 100 proposed data centres, 20 have applied for permits. The mechanism matters because the interconnection queue has been doing work it was never designed for. It is a free option: a developer files a request, gets a place in line, and uses that position as evidence of a project when raising capital or negotiating with a utility. Nobody has to prove they can pay. Utilities have been building demand forecasts on it; lenders have been reading those forecasts as a floor under power prices and lease rates. When states impose upfront payments, the numbers collapse. Reuters reports major utilities already cutting their data-centre demand figures once financial guardrails go in. This cuts both ways for our argument. It removes speculative capex from the pipeline, which is genuinely less fragility, not more. But it also means the demand series that underwrote power-sector valuations was substantially fiction. Talen is down 13.6% over twenty days and Vistra 18.3% off its high, which suggests the equity market got there first.

Private credit and BDCs1 Sephigh

The first real number out of Australia is minus fifteen per cent

A closed-end fund cannot suffer a run, but it also cannot be forced to tell you what its loans are worth until it sells them.

Ray White Capital, one of the private credit firms most exposed to the collapsed Sydney developer Bathla Group with around A$242m lent, told investors on Tuesday that the homes securing those loans will be released into a market where they are worth up to 15% less than assumed. It added that investors "were never promised the opportunity to redeem their investments" because the funds are closed-ended. Its people described the feel of it as GFC-like. (AFR) That is the first hard loss estimate in an episode we have been tracking for a week, and it is the mark being tested rather than asserted. Note the structure: closed-ended funds cannot be gated because there was never a gate, which is honest but also means the loss will be discovered at wind-down rather than in a monthly NAV. MA Financial capped redemptions at 1% of fund assets a month; CVS Lane suspended them outright. The listed market is telling a split story. The managers were sold hard on Tuesday — Blackstone -4.6%, Blue Owl -4.6%, Apollo -3.6%, KKR -3.1%, Ares -2.9% — while the vehicles holding the actual loans barely moved: Ares Capital -0.7%, Blackstone Secured Lending -1.2%, and both within 1.4% of their recent highs. So the market is marking down fee streams and fundraising, not credit. Jefferies puts Q2 non-accruals at BXSL at 3.6%, from 0.3% in early 2025, and across the twenty largest BDCs the median rose to 2.8% of cost. One of those two prices is wrong.

Household credit1 Sephigh

The collateral behind the Australian gates

Australia is running the experiment we cannot run on US private credit: what a marked-to-model property loan book does when the property is repriced.

The loans that Australian private credit funds are gating against are secured on houses, and the houses are falling. Sydney values are down about 7% from their February peak, and values are now declining in 93% of suburbs across Australia's capital cities. Sales volumes in Brisbane, Perth and Sydney are down more than 20% year on year, and mortgage applications have dropped by as much as 20% since the May Budget removed tax incentives for property investors. (Business Times) About 60% of Australian household wealth sits in property, per Commerzbank. Consumer confidence hit "deeply pessimistic" in June. The sequence is worth laying out because it is the same one that runs in every credit cycle, just faster here. Rates rise and a tax subsidy is withdrawn. Buyers step back. Volumes fall before prices do, because sellers hold out. Developers who borrowed at construction-cost assumptions from 2024 find that the finished product is worth less than the loan. Bathla went insolvent on 25 August owing A$3.3bn. Ray White's lenders now expect the collateral to clear 15% light. Funds that lent to developers face redemption requests from investors who read the newspaper, which is why MA Financial capped withdrawals at 1% a month. None of this is systemic on its own — Australia's private credit market is around A$200bn and ASIC has called it the sector's "first real test". What makes it worth watching from here is that it is a full, compressed demonstration of the mechanism we argue is sitting untested in much larger markets: illiquid loans, monthly liquidity promises, collateral marked to an assumption.

Bond market dysfunction1 Sephigh

Doubling the buybacks while insisting nothing is wrong

The official line is that the long end is fine; the official actions are those of a debt manager who thinks it is not.

The US 10-year closed Tuesday at 4.796%, a 20-month high; the 30-year sits at 5.27%. UK 30-year gilts hit 5.89%, the highest since March 1998, and 10-year gilts the highest since June 2008. German 10-year bunds made a 52-week high. Brent is above $92 after US–Iran exchanges around Hormuz. (CNBC) Treasury Secretary Scott Bessent told CNBC the US bond market remains "the best performing market" in the world. Minneapolis Fed President Kashkari has said he sees "no sign" the selloff is making the Fed's job harder and that the Treasury market is "functioning as it should." Set that against what Treasury is actually doing. From 9 September to 4 November it will at least double the maximum size of its liquidity-support buybacks in the 10–20 and 20–30 year sectors, from $2bn to $4bn per operation. Buybacks of this kind exist to support trading in off-the-run bonds — the older, less liquid issues that dealers warehouse. You expand them when dealers are full. The important point about a buyback is that it is not stimulus. Treasury has to fund the repurchase, and it funds it mostly with new short-dated issuance. The operation swaps duration out of the market for bills; it does not reduce the debt. When it was announced, the relief lasted about a day. One genuinely new pressure: Standard Chartered's Steve Englander told CNBC that a Supreme Court tariff ruling removed roughly 40% of additional tariff revenue, which changes the deficit arithmetic underneath all of this.

What would change our mind

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

The extra interest shaky borrowers pay compared with the government passing through 3.50 percentage points from 2.63, or any listed fund that lends to mid-sized private companies cutting its dividend, or a large US private credit fund closing the exit door (rather than in Australia). That would be credit repricing rather than stock-market noise.

Would move the number

A GPU-backed or data-center loan pulled for lack of demand, or a deal for a cloud computing startup repriced at a higher interest rate. The GMI Cloud book clearing at more than double the ask is the strongest evidence that this funding channel is still wide open.

Would move the number

The ten-year rate holding below 4.50% and the thirty-year below 5% for a full week, which would be the first evidence that the repricing of long-term interest rates has an upper bound. We would cut both fragility and ignition on that.

Would move the number

Any sign of month-end or quarter-end funding stress: the overnight rate banks charge each other printing well above the rate the Federal Reserve pays on reserves, or a jump in the Fed's emergency lending window. That is the channel through which a move in interest rates becomes a crisis in credit.

Would move the number