Oil at $95 and a strategic reserve at 1982 lows

Archived reading, published Wed, 02 Sep 2026 02:24:07 UTC (19 days ago). This is not the current state of the meter.

See the live reading →

CRASH-O-METER

Cracking

how close are we

+1 since the last reading

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 29 sources and rewrites this page.

Why it moved: Ignition up one to 42: Brent settled at $94.65 and traded as high as $97.04 into 2 September, a five-week high and a further escalation of the shock now driving the global bond rout, with the yen through 160 for a third straight session ahead of a near-certain BOJ hike — moves the equity tape, still frozen at the 1 September close, cannot show. Fragility holds at 91: Dell's guidance raise and the BDC write-down data are sharper measurements of risk already standing, and nothing was unwound.

Reporting from 9 Jan to 1 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.

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.

Signs of the times

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

The AI bottleneck is disk drives

Dell's COO Jeff Clarke told the 1 September earnings call the company has "spotty CPU shortages" and "shortages with disk drives" — in the same quarter it raised its AI server forecast to $74bn from $60bn. The constraint on machine intelligence is spinning rust.

aisupply chainhardware

1 SepReuters (verified earnings call quote)

A backlog five times a quarter's sales

Dell booked a record $60.9bn of AI server orders in three months and reported a $95bn backlog — roughly five times the $16.4bn of AI servers it actually shipped in the quarter. Demand is no longer the question; delivery is.

aibacklogcapex

1 SepDell Q2 FY27 release (fact-checked)

The reserve is 41% full

The US Strategic Petroleum Reserve held 289.7 million barrels as of 21 August, its lowest since 1982 and about 41% of its authorised 714 million barrel capacity, with another 39 million barrel release planned — just as Brent went through $95.

oilpolicyinflation

1 SepReuters / Department of Energy (fact-checked)

Abu Dhabi's spy chief owns half a stablecoin bank

Sheikh Tahnoon bin Zayed — deputy ruler of Abu Dhabi and UAE national security adviser — and co-investors hold 49% of the holding company of a US bank that has just received preliminary approval to issue USD1, the stablecoin launched by the Trump-family-linked World Liberty Financial. His conglomerate separately completed a $30m dirham-stablecoin transfer, among the largest in the Middle East.

stablecoinsgeopoliticsbanking

31 AugSemafor

Mortgages, secured on bitcoin

Coinbase and Better have launched bitcoin-backed mortgages after a waitlist projected more than $260m of demand — housing debt collateralised by an asset that fell 21% and rose 21% in separate twenty-day windows this year.

cryptomortgagescollateral

31 AugBitcoin.com weekly recap

The rumour mill

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

Confirmed

“The US Strategic Petroleum Reserve is currently being depleted.”

Department of Energy data show 289.7 million barrels as of 21 August, down 3.7 million in a week and the lowest since 1982 — roughly 41% of authorised capacity, with a further 39 million barrel release planned.

Claimed by Gregory Mannarino

Partly true

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

The roughly 900,000 downward benchmark revision is real, but it covers the twelve months to March 2025 and is the BLS's standard annual reconciliation to tax records, not an invention. The newest preliminary benchmark, published 28 August, was −79,000.

Claimed by Gregory Mannarino

Partly true

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

Multiple outlets put CAPE around 41.0–41.4 through July and August 2026, the highest since September 2000 — but there is no single clean "first crossing" date, and it has been printing around 40 since June.

Claimed by Wealthion

Partly true

“Consensus is that at least 15 gigawatts of power for AI chips will be missing in 2027.”

Musk's actual post said ~15GW of AI compute produced in 2027 cannot be switched on in 2027, and blamed transformers, wiring, cooling and networking as much as generation. The bottleneck is the electrical supply chain, not only electricity — which matches Texas freezing new grid connections.

Claimed by Meet Kevin

Earlier developments

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

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.

Crypto and TradFi contagion1 Sepmedium

Twenty-one banks, one stablecoin, no evident demand

The banks are conceding that instant-settlement tokens will take deposit share, and building a joint venture rather than a defence.

Goldman Sachs, Bank of America, Citi and Deutsche Bank are among 21 financial institutions that said on Tuesday they will form a company this year to issue a dollar-pegged stablecoin in the first half of 2027, then expand into the euro and other G7 currencies. The group had ten members when it was announced in October 2025. A rival consortium of 37 institutions, Qivalis, plans a euro stablecoin later this year. Reuters adds the line that makes the story: "there are few signs of demand for stablecoins issued by banks." (Reuters) What is a bank stablecoin, mechanically? A token backed by reserve assets that settles instantly and around the clock, competing with the same bank's own deposits. If a corporate treasurer moves $500m from a demand deposit into a consortium stablecoin, the bank loses a funding liability it can lend against and gains a claim on a segregated reserve pool it cannot. That is the tension nobody in the announcement addresses: banks are building the thing that disintermediates their cheapest funding, because the alternative is watching Tether do it. The market they are entering is dominated by a private issuer, and the demand for a regulated alternative is, on Reuters' reading, hypothetical. Meanwhile Circle minted roughly 5bn USDC in a single week, and Coinbase and Better launched bitcoin-backed mortgages off a waitlist projecting over $260m of demand. Stablecoins are becoming a funding market. Their reserve assets are Treasury bills. That link — crypto deposit flows to front-end Treasury demand — is now the most consequential crypto-to-TradFi channel, and it is far larger than the digital asset treasury companies everyone watches.

Bond market dysfunction1 Sephigh

Tokyo joins the rout at three per cent

The last thirty years of cheap global duration rested on Japanese savers having nowhere better to go; at 3% they do.

Japan's 10-year government bond yield reached 3% on Tuesday, the first time since 1996. Five-year JGBs hit a record 2.26%. In the same session UK 10-year gilts touched their highest since June 2008 and the 30-year gilt rose nine basis points to 5.88%, the highest since March 1998. US 10-year Treasuries closed at 4.79%, the 30-year at 5.27%. Brent went through $92 with just five ships transiting the Strait of Hormuz. The Japanese number is the one to hold on to. For thirty years, Japanese institutions — life insurers, banks, pension funds, the largest pool of foreign savings in the world — could not earn a return at home and so bought foreign bonds, much of it US and European paper, often hedged. A domestic long bond yielding 3% changes that arithmetic without anyone announcing anything. It does not require a policy decision, a headline or a crisis; it requires an actuary to run the numbers again at the next asset-allocation meeting. What we cannot see yet is whether that is actually happening. Repatriation flows show up with a long lag in the balance-of-payments data. What is visible is the coincidence: yields rising together across the US, Germany, the UK and Japan, which Reuters and Bloomberg both attribute to energy-driven inflation and fiscal supply rather than any single country's politics. Big Tech's bond issuance is being named in the same sentence — the AI buildout is now large enough to be a supply factor in the global government bond market, which is a sentence that would have read as nonsense two years ago. Minneapolis Fed president Kashkari says the Treasury market is "functioning as it should". On the evidence — no failed auctions, no dealer seizure — he is right. This is a price problem, not yet a plumbing problem.

The AI capex bubble31 Augmedium

A vehicle that buys chips and leases them out

The chips are collateral only if there is a second-hand market for three-year-old accelerators, and nobody has tested that.

Broadcom is reported to be in talks with lenders to raise more than $60bn, possibly $100bn, through a special-purpose vehicle that would buy Broadcom's own custom AI chips and lease them to Anthropic. Bank of America is quoted estimating that total exposure across the structure could reach $370bn by 2029. A previous version of the arrangement, set up in June with Apollo and Blackstone, is said to have raised $35bn. This is tier-2 reporting and the numbers move between accounts, so treat the range as a range. The structure, though, is worth understanding because it is becoming the standard form. A vendor wants to sell chips. The customer cannot fund them out of cash flow. So a separate company is created, funded with borrowed money from private credit and insurance balance sheets, which buys the chips and rents them to the customer. The vendor books a sale. The customer books a lease. The debt sits in the vehicle, which appears on nobody's consolidated balance sheet in a form the reader of an annual report would recognise. Reporting cited by CNBC puts roughly $3tn of off-balance-sheet AI commitments across nine large technology companies; the BIS calls it "shadow borrowing". The question that decides whether this is clever or dangerous is: who owns the residual? If Anthropic stops paying in year four, the lender is left holding used accelerators two generations behind the frontier, in a building that may not have power. Nothing in the reporting tells us how the residual value is set, who guarantees it, or what the chips are assumed to be worth at the end. Until that is disclosed, the loss-bearer is unidentified — which historically has meant it is everybody.

Fed, Treasury and policy1 Sephigh

$150bn of data centres that voters said no to

For once a piece of the AI pipeline is getting smaller rather than more leveraged, and that genuinely lowers fragility.

Dozens of US data-centre projects worth roughly $150bn have been stalled or cancelled because of local opposition, the FT reports. President Trump responded on Monday by telling Americans opposing them that they wanted to end up "backwards and poor", and urging the country to "let Data Reign". Vice-president Vance put "probably 99 per cent" of the backlash down to higher energy bills in surrounding areas. We have been arguing for a fortnight that the binding constraint on the AI buildout was shifting from capital to power. This is the next step: the constraint is becoming permission. Texas has frozen new grid connections while it works out which of 474 gigawatts of requests are real. Pennsylvania says only 20 of more than 100 proposed projects have actually applied for permits. And now a specific dollar figure has been attached to projects that are simply not going to be built. For the fragility argument this cuts in an unfamiliar direction. Capex that does not happen is debt that does not get raised and leases that do not get signed. $150bn of cancelled projects is $150bn of risk that will not be stacked. Set against that, every cancelled site makes the revenue assumptions underneath the sites that do get financed harder to verify — if a third of the pipeline is political vapour, the demand forecasts justifying the rest were built on the same spreadsheet. The money has noticed. An advocacy group tied to Leading the Future, a super PAC that has raised more than $75mn from Marc Andreessen, Ben Horowitz and OpenAI's Greg Brockman, is running an advertising campaign for data centres in Kansas, Ohio and Wisconsin. When an industry has to buy public consent by the state, its costs have changed.

The dollar, gold and reserve status1 Sepmedium

Gold fell 7% during a war and an oil shock

If gold is falling because people need cash rather than because real yields rose, someone is being margined and we cannot yet see who.

Gold is down 7.07% over five sessions and 2.63% on Tuesday alone, at $4,391. Silver is down 6.59% on the week. This happened while Brent went above $92, while the US and Iran exchanged direct attacks, and while long-dated government bonds sold off across four continents. Gold is still up 3.4% over twenty days, so this is a sharp move inside an uptrend rather than a break — but it is the wrong direction for the week's news. Two explanations fit. The first is mechanical: gold pays no coupon. With the 30-year Treasury at 5.27% and the dollar index up 0.79% on the week, the cost of holding a metal that yields nothing has risen, and gold sells off when real yields rise regardless of what is happening geopolitically. That is the boring answer and it is probably most of it. The second is that gold is what you sell when you need cash. In a liquidation, the asset that goes first is the one with a deep market and an unrealised gain — and after a 20-day run, gold qualifies on both counts. Tuesday's tape is consistent with that: equities down, credit slightly weaker, gold down, dollar up. That is the signature of positions being reduced rather than risk being reallocated. We cannot distinguish the two from price alone, and we are not going to pretend otherwise. What we would watch is the central-bank bid underneath, which runs on a much slower clock: India's reserves hit a record $729.3bn in the week to 21 August, with gold holdings up $2.8bn, and the Reserve Bank of Australia is reported to have cut its target dollar share from 55% to 45%. Those buyers do not care about a bad week.

Crypto and TradFi contagion31 Aughigh

Bitcoin fell 1.8%. Its holders fell 6%.

Digital asset treasury companies are a leveraged claim on a volatile asset funded by continuous equity issuance, and the leverage lives in the share price, not the balance sheet.

Bitcoin closed Tuesday down 1.82% at $77,121. The companies whose business is holding it fell three to four times as far: Strategy -5.78%, BitMine -8.00%, SharpLink -5.15%, Coinbase -6.21%, Galaxy Digital -5.17%. Over twenty days the same names are up 28-30% against bitcoin's 21.6%. The wrapper has a beta to the coin, in both directions, and it is running near three. The mechanism is in Strategy's own filing. It bought $369.7m of bitcoin in the week to 30 August, resuming purchases after a ten-week pause. To fund it, the company raised $602.8m by selling common stock and used $151.8m to buy back preferred shares. Read that again: it issued substantially more equity than it spent on the asset it exists to hold, and spent part of the difference retiring a fixed claim ahead of the common. This is the machine working as designed. If the shares trade above the value of the coins per share, issuing stock to buy more coins raises coins-per-share and is accretive; the premium funds the purchases, and the purchases justify the premium. The reflexivity is the product. It runs in reverse just as smoothly — if the premium goes, new issuance dilutes rather than accretes, and the buying stops, which was presumably what the ten-week pause was about. The filing tells us what was bought and how it was funded. It does not tell us where the multiple stands today, and we cannot compute it from the data in front of us. What Tuesday's tape shows is that the equity is repricing faster than the asset — which is what you would expect if investors are marking the premium, not the coin.

Household credit1 Sephigh

Cards are getting better. Cars are not.

A car loan is the debt people default on last, so 5.49% of balances 90 days late is a statement about the bottom third of the income distribution, not about credit conditions.

New York Fed data for the second quarter show 5.49% of auto loan balances 90 or more days delinquent, with 3.00% of balances transitioning into that state over the previous four quarters. Equifax, summarising Moody's, says subprime auto borrowers are at or near a 15-year high. Over the same period credit cards improved: bank-issued card 30-plus-day delinquency fell to 2.85%, and 60-plus-day delinquency across all cards fell to 2.69% from 2.87% a year earlier. FICO reports 30-day auto delinquency actually easing to 2.6%. These are not contradictory, and the divergence is the interesting part. The early-stage numbers are stabilising while the late-stage numbers are not. That is what it looks like when a cohort that fell behind two years ago is not recovering — the flow into distress slows, the stock of distress keeps compounding. FICO's own breakdown supports this: 90-day-plus delinquency rose only in the lowest score bands and was flat everywhere else. Add 3.2m student-loan borrowers reported at least 30 days late in the previous six months, carrying an average 38-point score decline, and the same households appear in both series. The market is pricing it selectively rather than systemically. Affirm fell 6.08% on Tuesday and 10.23% on the week; Capital One is off 4.26% over twenty days and Ally 6.81%. But Credit Acceptance, the deep subprime auto lender, is up 1.25% over the same twenty days and sits 1.6% from its high. Deep subprime funds itself through the securitisation market, and while high yield spreads sit at 263bp that funding stays open. The equity is telling you that the lenders' cost of money matters more, for now, than their borrowers' ability to pay.

Bond market dysfunction1 Sephigh

An oil shock arrives in the middle of a bond rout

The long end is now absorbing an inflation shock while the only official tool pointed at it is a liquidity backstop that cannot change duration supply.

Brent rose 1.7% to $92.1 a barrel on Tuesday and only five ships transited the Strait of Hormuz. Wheat hit a three-and-a-half-year high on attacks in the Black Sea. Long bonds sold off worldwide: the 30-year gilt jumped 9bp to 5.88%, its highest since March 1998, and the 10-year gilt reached its highest since June 2008 (Guardian). The US 10-year broke above 4.75% and the 5-year above 4.50% on Monday, both for the first time since January 2025. The 30-year Treasury sits at 5.26%, having closed above 5% on 55 days this year — the most since 2006 (Bloomberg). The plumbing matters here. Treasury's response to the earlier long-end move was to double liquidity-support buybacks from $2bn to at least $4bn per operation, starting 9 September and running to 4 November, funded out of a Treasury General Account near $1tn. That is a market-functioning tool: it buys illiquid off-the-run bonds to keep dealers able to make prices. It does not change the supply of duration, and it does nothing about an inflation impulse. The first announcement bought a day of relief before yields round-tripped. What makes this configuration different from the last fortnight is the direction of the Fed. Kevin Warsh told Jackson Hole that 2% is a "firm and fixed target" and that "we still have work to do." An energy shock into a central bank that has already priced out easing, with a hike live for 16 September, is the one combination in which the long end and the front end can rise together. Both did. Against that: no failed auctions, no dealer balance-sheet seizure, no forced unwinds. This is repricing, not breakage.

The AI capex bubble1 Sephigh

Texas asked for proof and 474 gigawatts appeared

The demand figures underwriting both utility capex and data-centre project debt are partly duplicative filings that cost nothing to make.

Requests for grid connections from very large power users — mostly data centres — now exceed 700 gigawatts across parts of the Midwest, Mid-Atlantic and South, more than ten times industry estimates of what US data centres actually consume today and roughly as much electricity as every home in the country uses. In Texas alone, requests went from about 48GW in 2023 to more than 474GW now. Texas has become the first major hub to freeze new grid connections while it investigates what is real. Pennsylvania says only 20 of more than 100 proposed data centres actually applied for permits (Reuters). The mechanism is worth being precise about. An interconnection request is free or nearly free to file, and developers file with multiple utilities for the same project to preserve optionality. Those queues then become evidence: utilities cite them in capex plans and rate cases, power producers cite them when talking about contracted demand, and lenders to neoclouds and project SPVs treat a signed power agreement as the thing that makes a GPU-stuffed shed financeable. Reuters reports that utilities have cut their data-centre demand figures once upfront payments were required — which is the queue telling on itself. Two consequences pull in opposite directions. Fake demand means the buildout, and the debt attached to it, is smaller than the headline pipeline implies. That lowers systemic risk. But it also means the scarcity story underwriting power valuations is thinner than the queue suggested, and the binding constraint on real projects is now permission rather than capital — which is not something a lender can solve. The prices have noticed. Talen is down 13.6% over twenty days and 22.5% off its high; Vistra is 19% off. Applied Digital is down 22% over twenty days, Core Scientific 31%.

Hidden leverage and shadow banking31 Augmedium

The trade that cares about volatility, not yields

The largest leveraged position in the Treasury market is indifferent to yield levels and highly sensitive to exactly the volatility now arriving.

Analyst estimates put the cash-futures Treasury basis trade at about $830bn, close to double its early-2020 peak, with a further $305bn in swap-spread arbitrage, financed by roughly $3tn of hedge fund repo borrowing — more than double early 2023. Hedge funds' gross Treasury exposure is estimated to have doubled to about $4tn between 2023 and September 2025, $2.4tn long against $1.6tn short, or around 8.5% of outstanding Treasuries. The top 50 funds account for about 90% of the repo borrowing (matterfact). These are commentary estimates, not official data; the BIS, IMF and Fed published nothing new on this in the window. The structure: buy the cash bond, short the future, capture the small spread, and finance the bond in repo at a haircut of a fraction of a percent. Leverage of fifty or a hundred to one is normal because the spread is tiny. Crucially, the position is close to neutral on the direction of yields. What it is not neutral on is volatility. A jump in realised volatility raises futures initial margin and repo haircuts at the same time, and the only way to meet both calls quickly is to sell the cash leg — into a market where dealers are also getting hit. That is why the relevant number this week is not the level of the 30-year but the MOVE index, up 25.5% over five sessions from a low base. Commerzbank called the basis trade "shaky" on 26 August. The RBA's Michele Bullock put it more plainly: "People don't know where the leverage is." The SEC, separately, has subpoenaed banks about one hedge fund's communications with its lenders over the leverage it was taking on. Nothing has broken. But a global long-end selloff is the precise input this position dislikes.

Private credit and BDCs1 Sepmedium

The managers fell. Their loan books rose.

The listed BDCs are trading at or near NAV while their own filings show the worst credit metrics since at least 2021 — one of those two numbers is wrong.

On Tuesday the private credit managers were sold and the funds they run were not. Blackstone -2.4%, Ares Management -2.1%, Apollo -1.9%, KKR -1.5%, Blue Owl -1.6%. Meanwhile Blackstone Secured Lending rose 0.5% and closed at its period high, Ares Capital rose 0.2% and sits 0.5% off its own high, Blue Owl's BDC was flat and FS KKR is up 2.9% on the week. That split is a specific claim about where the damage is. Manager equity is a levered bet on future fundraising, AUM growth and performance fees. BDC equity is a claim on a loan book carried at net asset value. Pricing the first down while marking the second up says: fundraising slows, the book is fine. The reported book is not obviously fine. Non-accruals across the ten largest BDCs reached 3.95% of cost in Q2, up from 3.75% in Q1; the median across the twenty largest rose 80bp in a quarter to 2.8%, the highest since around 2017. Blackstone Secured Lending's own figure went from 0.3% in Q1 2025 to 3.6%. Across 83 BDCs, investments still held were written down by $2.04bn, the largest quarterly write-down in a thirteen-quarter dataset, and positions marked below 90 cents or on non-accrual went from 0.9% to 3.2% (JD Supra, Accelerate). The PIK share of BDC loans has risen from roughly 6% in early 2022 to about 10%. The reconciliation available to us is unattractive: a diligenced, first-lien-heavy book marketed by BlackRock reportedly cleared at 88–92 cents on the manager's own marks. If that is the clearing price, then the discount the market is applying to the fee stream belongs on the loan book as well.

The dollar, gold and reserve status31 Aughigh

De-dollarisation is real and it is not today's trade

Reserve diversification raises the long-run cost of US duration; it does not explain a week in which gold and Treasuries fell together and the dollar rose.

The Reserve Bank of Australia cut the target share of US dollars in its reserve portfolio from 55% to 45% in 2024/25, raising the euro from 20% to 30% — a ten-point shift Deutsche Bank described as a quiet reduction. A Kansas City Fed working paper puts the dollar's share of global FX reserves at 57% in Q1 2026, against 72% in 2000. India's reserves hit a record $729.33bn in the week to 21 August, up $12.42bn, with gold holdings up $2.80bn to $114.22bn (Bloomberg). Every one of those facts is real. None of them explains this week's prices, and it is worth being clear about that rather than pretending the narrative and the tape agree. The dollar index is up 0.64% on the week at 99.55 and the broad index up 0.58%. Gold is down 4.8% over five sessions to $4,414 and silver down 5.9%. Over twenty days both are still up — 4% and 8.9% — so the trend has not broken. But over five days, gold, silver, equities and bonds have fallen together while the dollar rose. That pattern has a boring name: liquidation. When everything falls at once and the funding currency rises, the marginal seller is meeting a margin call, not expressing a view on reserve composition. Reserve diversification is a decade-scale flow of a few percentage points a year against a $100tn-plus asset base; it does not set the weekly price of gold. The useful reading is narrower. Reserve managers are gradually reducing their structural bid for Treasuries at the same time as the US is issuing record duration. That shows up in term premium over years, and in the 30-year's 55 days above 5% this year.

Crypto and TradFi contagion31 Augmedium

Bitcoin is now mortgage collateral

Crypto risk is being wired into mortgage credit and collateral chains before anyone has disclosed the loan-to-value or the top-up terms.

Coinbase and Better have launched bitcoin-backed mortgages, after a waitlist that the companies projected at more than $260m of demand. In the same week Circle minted roughly 5bn USDC, BlackRock's IBIT processed more than $5bn of direct bitcoin-to-ETF conversions after cutting the minimum ticket from $25m to $1m, Franklin Templeton received clearance to use its $2.6bn tokenised BENJI money-market fund inside ETFs and mutual funds and as collateral, and Schwab — a $13tn platform — added SOL, AVAX and LINK (Bitcoin.com weekly recap). The mortgage product is the one that changes a plumbing diagram. A crypto-collateralised mortgage takes an asset that can fall 30% in a fortnight and puts it behind a thirty-year liability on a house. The reporting we have does not say what the loan-to-value is, what the top-up mechanic looks like, who holds the loan after origination, or whether it is intended for securitisation. Those four facts determine whether a bitcoin drawdown produces a margin call on a homeowner, a writedown at a lender, or a downgrade in a securitisation trust. Until somebody publishes them, this is an unmeasured channel between a volatile asset and housing credit. The tokenised money-fund clearance is the quieter version of the same thing. A money fund share used as collateral is fine until the fund gates; then the collateral is illiquid precisely when it is needed. Bitcoin is at $78,300, down 2.4% on the week but up 23.5% over twenty days. Coinbase fell 3.3% on Tuesday, Galaxy 3.7%, the ether treasury companies 2–4% — the equity claims on crypto sold harder than the coin did, which is the ordinary order of things when leverage is somewhere in the stack.

What would change our mind

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

High yield OAS moving above 350bp from 263bp, or HYG breaking 5% below its recent high — the first genuine sign the credit market has stopped accepting private credit's marks.

Would move the number

A listed BDC cutting its dividend, or a large non-traded BDC gating redemptions in the US rather than Australia — that would turn a slow markdown into a funding event.

Would move the number

Dell's next quarter shipping materially below the run-rate its $74bn guide requires, or a neocloud debt deal (Volta's $5bn, Nebius's $5bn) being pulled or repriced sharply wider.

Would move the number

Brent back below $80 with the 10-year under 4.50% and the yen back inside 155 — which would take the ignition number down several points, because the current reading is largely an energy-and-fiscal story.

Would move the number