Archived reading, published Sun, 23 Aug 2026 10:19:43 UTC (3 days ago). This is not the current state of the meter.
See the live reading →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 15 sources and rewrites this page.
Why it moved: Down one to 64. Ignition comes off two points because the growth data cuts against a near-term credit accident — S&P Global's flash composite at 56.0 is the fastest US business activity in over four years — and because last week's long-end selloff has now run four sessions without any echo in funding markets: rate volatility is at a 20-day low, VIX 15.1, high-yield 275bp. Fragility holds at 89: this window's new material (Nvidia's SPV collateral terms, the OCC trust charter, the NY Fed household file) is better measurement of risk that was already standing, not new tinder.
Reporting from 20 Aug to 23 Aug
Eight places it could go, scored 0–10. Tap one for the explainer.
What changed in the last few hours, and what each one says about the plumbing.
The long end has settled at a level that reprices every leveraged balance sheet in the system, and it got there calmly enough that nobody has been forced to do anything about it yet.
A stablecoin whose issuer holds its own reserves has the same maturity and liquidity mismatch a bank has, minus the backstops a bank gets.
The consumer credit book is deteriorating in level but not in rate of change, which is why lenders' shares are near highs and why the reckoning is being pushed into later quarters.
The long end is being cleared by leverage that funds itself one day at a time, which is fine until the day it isn't.
Capping long yields does not remove fiscal pressure, it relocates it — and three separate hard-asset markets are now pricing the relocation at once.
The AI buildout is being financed against collateral that is worth a lot precisely when nobody needs to seize it, and guaranteed by the vendor selling the collateral.
Stuff you wouldn’t have believed was possible until 2026.
Bessent said corporate tax revenue is being dented by the wave of factory and data-centre construction, which is immediately expensed against profits under the 2025 tax act. The buildout is now widening the deficit that is pushing up the yields the buildout has to borrow at.
Strategy sold about $333.7m of its own stock last week, bought and sold no bitcoin, and ended the week holding a $4.8bn dollar reserve. The original digital-asset treasury company spent the week raising equity to hold cash.
BofA's Bull & Bear Indicator rose to 9.5, inside "extreme bull" territory, in the same week the 30-year Treasury hit its highest yield since 2007. Weekly flows showed $40.1bn into equities and $21.4bn into bonds.
Mastercard has closed its $1.8bn acquisition of stablecoin payments firm BVNK, while Coinbase picked Abu Dhabi as its international tokenisation hub and a South Korean regional bank replaced SWIFT correspondent transfers with Ripple Payments. The plumbing is being bought, not built.
What the crash-callers are saying, checked against real reporting.
“PIMCO estimates capex could absorb about 94% of hyperscaler operating cash flow this year and next.”
PIMCO's May 2026 report puts hyperscaler capex at roughly 93–94% of operating cash flow across 2026–27, against 33–40% in 2022–23. On that arithmetic roughly $6 of every $100 earned is left over for dividends, buybacks and everything else — which is why the rest is being borrowed.
Claimed by Mark Moss
“Microsoft just changed the depreciation schedule on its offices and data centers from 15 years to 25 years.”
CFO Amy Hood said on the 29 July call that Microsoft will extend the estimated useful life of data centres and office buildings from 15 to 25 years — but effective from the start of FY27, so it is not yet in the reported numbers. Spreading the same cost over 25 years lowers annual depreciation and raises reported operating income; whether GPUs and buildings genuinely last that long is the whole question.
Claimed by Meet Kevin
“Data centre securitisation issuance was about $2.4bn in 2020 and grew to $15.5bn last year.”
Bloomberg data confirm annual data-centre ABS issuance rose from $2.4bn in 2020 to $15.5bn in 2025 — a sixfold increase in a market that packages lease payments from tenants whose own AI revenues are still speculative.
Claimed by Mark Moss
“Global debt stands at 350% of world GDP.”
The IIF puts total global debt above $350 trillion in early 2026, which is about 305% of global GDP — the $350tn figure has been converted into a percentage it does not support. Other estimates run to roughly 336%. The number is large enough without the rounding error.
Claimed by Thoughtful Money (Adam Taggart)
Dispatches from previous readings. The same argument, no longer the news.
The largest AI-fund blow-up of this cycle was resolved by one buyer's balance sheet rather than by the market, which means the price at which those positions would actually clear under stress is still unknown.
This is the first sign in a listed private-credit portfolio that the AI boom is producing credit losers as well as winners, and it arrived as a valuation markdown rather than a default.
Buybacks shift duration risk from investors to the government's own refinancing schedule, and the fact that they are being tried at all is the clearest measure of Washington's pain threshold on long yields.
AI capex and the federal deficit are now competing for the same duration buyers, which makes the long end of the Treasury curve the binding constraint on the buildout.
The main post-2020 regulatory repair to the Treasury market leaves affiliate and open-term repo outside central clearing, so the most leveraged corner of the world's most important bond market remains partly unmeasured.
Treasury companies rising faster than the coin they hold means premiums to net asset value are re-expanding, which is what funds their buying and what disappears first when sentiment turns.
Consumer credit stress is now at levels the lenders' share prices are explicitly not discounting, which means the loss is being absorbed somewhere less visible than a bank income statement.
The long end is where policy is currently straining, and the long end is financed by an unregulated $2tn position that unwinds by selling exactly what is already falling.
Yield suppression does not remove fiscal risk from the system; it relocates it to the currency, where the US has fewer tools and more foreign holders watching.
The AI buildout is increasingly financed against collateral whose value is set by the same company whose products the debt is buying.
A forced seller has appeared, and the assets being sold are sports teams rather than loans — which tells you which side of the balance sheet the marks are still holding.
When a treasury company issues stock to hold cash rather than coins, the flywheel that justified its premium is running in reverse even as the premium expands.
The cost of AI infrastructure debt is now set by junk buyers, whose participation is the least reliable part of the capital structure.
The market is now discriminating between hyperscalers and the leveraged periphery that builds for them, which is where any AI credit event starts.
Fiscal and monetary duration policy are pulling opposite ways in a market already clearing 30-year paper at the worst yields since 2001.
Insurance balance sheets are the largest untested holder of private credit, and the first public unwind of one is happening now.
Regulated fund cash management is being wired into blockchain settlement at the same time as the levered crypto equity complex is at its highs.
Non-accruals are the earliest hard number in private credit that the manager cannot mark to its own opinion, and they are now rising faster than the equity of the lenders implies.
The specific, observable things that would move the number - in either direction.
SOFR printing materially above the interest-on-reserves rate, or a repo spike around a settlement date — that would turn the basis-trade leverage from a standing risk into an unwind, and we would raise ignition several points.
Would move the number
The 30-year through 5.5% with the dollar and equities falling on the same day; a simultaneous move in all three is the correlation break that says something structural, not a duration repricing.
Would move the number
A non-traded BDC gating beyond its stated quarterly limit, or a listed BDC cutting its distribution — evidence that the redemption queues we have been tracking are actually forcing sales.
Would move the number
A filing from Nvidia or Broadcom quantifying the backstop and guarantee obligations and how they are recognised. If the exposure is smaller or better capitalised than the reporting implies, we would take fragility down.
Would move the number