Takeaway: growth is real, and so is the bill for it.
Two prints landed within hours on September 10. August PPI rose 0.4% m/m as expected but accelerated to 5.4% y/y from 4.8% in July, with goods up 1.1% — more than three-quarters from a 4.2% monthly jump in energy. Core PPI was 4.6% vs 4.3%. One detail matters for tech investors: electronic component costs jumped 3.4% in a month and are up 27.6% on the year, which Capital Economics ties directly to the AI buildout.
Then Oracle reported Q1 FY27 after the close: revenue $19.35B vs $19.14B expected (+30% y/y), adjusted EPS $1.92 vs $1.74, cloud infrastructure +121% to $7.4B, RPO up to $664B after $30B of new AI contracts "without requiring additional capital." The stock, down 22% year-to-date into the print, rose ~4% after hours. But read the balance sheet, not the headline: capex $28.5B vs $8.5B a year ago, negative free cash flow $5.4B, debt ~$125B, and a $20B at-the-market equity sale completed during the quarter. That is the whole macro in one company — triple-digit AI growth funded by bonds, equity dilution, and off-balance-sheet leases.
Market context at the close: S&P 500 ~7,592 (-0.6%), Nasdaq weak with NVDA $218.36 (-2.4%), TLT $80.78 (-1.2%), XLF $56.87 holding, XLE $64.93 (-0.6%), BTC $77,136 (-1.3%), ETH $2,459, CNN Fear & Greed at 33 (Fear) while crypto fear-greed sits at 69 (Greed) — equities scared of rates, crypto still positioned long.
Takeaway: scale, banks, and collateral are all different, and that difference is decisive.
Copenhagen professor Jesper Rangvid's March 2026 comparison is the cleanest framing: today's private credit is yesterday's shadow banking, today's oil spike echoes 2002–08, and today's equity valuations echo housing. But three breaks stop contagion:
Consequence: default rates of 4–5% in loans and 7–8% in private credit — Guggenheim's forecast — are a painful sector cycle, not a banking freeze. That keeps our systemic-crisis probability at ~10% on a 12-month horizon.
Takeaway: investment-grade spreads say calm; loan prices say stress is already here.
IG and HY spreads fully round-tripped to pre-Iran-war levels on 23% S&P earnings growth and coupon demand. Underneath: Fitch's US private-credit default rate hit 6.1% in July, a record; leveraged-loan defaults were 4.9% in March; 21% of tech loans trade below 80 cents vs 10% a year ago; non-traded BDCs faced ~$20B of Q1 redemption requests with gates/caps (BCRED gated Q3). Roughly one-third to one-half of loans that distressed typically default — on tech's ~20% index weight alone that adds ~2 points to the loan default rate.
The accelerant is concentration: software is mid-teens to ~30% of loans/private credit vs <5% of high yield, entering this shock with ~1.25 turns more leverage and half a turn less coverage. AI disruption (seat-to-usage repricing, collapsing net retention) plus floating rates plus Hormuz-driven input costs (plastics, food, packaging) is a triple hit fixed-rate IG issuers don't face. Recovery rates tell the rest: 44% in the last 12 months, with 44% of 2022 LMEs already re-defaulted. Opinion: passive loan/BDC exposure here is picking up pennies in front of a repricing — selection or avoidance, no middle ground.
Takeaway: Waller told you the reaction function; PPI just pushed it toward hiking.
Funds sit at 3.50–3.75% after a 9–3 July hold (three dissents for a hike) under Chair Kevin Warsh. Governor Waller's September 3 speech was explicit: disinflation (3-month core down from 4.76% in February to 3.05% through July, PCE 3.7%/core 3.3%) argues for holding — but if August data reverses, "it may be appropriate to raise" on September 15–16, since policy is "only slightly" restrictive.
August PPI was that hot surprise on headline (5.4%), even if monthly core cooled to 0.2%. CME odds of a September hike rose from ~64% to ~70%; JPM Wealth now expects a 25bp "credibility hike" (futures ~65% pre-PPI); UBS sees odds back above 60% after a strong August jobs report. Meanwhile the 30Y closed at 5.27% on July 31 and traded above 5.3% in August — first time since 2007 — and the 10Y sits near 4.80%. Opinion: one hike is not a cycle, but at these term premia one hike reprices everything — mortgages at 6.76%, AI debt, and software coverage ratios all tighten together.
Three asset-class regime levels: 30Y >5.50% forces AI multiple compression (~10–12%) and mortgage lock; Brent >$110 sustained (vs ~$105–108 now, WTI ~$100) flips consumption; weekly loan-distress >25% under 80¢ signals broad default wave, not tech-only.
Takeaway: your "safe" bond fund is now an AI bet too.
Vanguard's August 19 accounting: hyperscalers (Alphabet, Amazon, Meta, Microsoft, Oracle) issued ~$35B/year in 2020–24, $93B in 2025, and ~$132B year-to-date through July including a ~$53B mega-tranche — now ~11% of all US IG supply vs ~2% in 2023–24. Full-2026 AI-related issuance (chips, data centers, utilities) is estimated at $300–570B; Goldman counted $489B already. Consensus capex: ~$777B in 2026, over $1T a year 2027–30.
Two consequences: duration — issuance skews 30Y and even 100Y, lengthening index duration just as term premia explode; and correlation — the same AI cycle driving >50% of earnings growth now drives net bond supply, so an AI disappointment hits stocks and credit together. Add Nikkei-flagged ~$1.65T in off-balance-sheet lease/hidden obligations and Alphabet's first negative quarterly free cash flow since 2004, and CDS tells the story: Oracle 5Y at ~212bp (record), all hyperscalers drifting wider. Nothing here is insolvency — coverage is still strong — but compensation is thin with spreads near historic tights.
Takeaway: own the spike (energy), rent nothing long-duration, keep dry powder for loan dislocations.
| Call | Entry | Target | Invalidation | Horizon / Conviction |
|---|---|---|---|---|
| EQ-XLE-1 (flagship, reaffirmed): BUY XLE energy hedge | $64–66 (closed $64.93) | $72; trim 1/3 at $69.50 | Brent daily close <$88 | 1–3 months / HIGH |
| Duration stance: NEUTRAL–SHORT long IG; prefer T-bills / short-duration + TIPS to 30Y | — | Cover/add duration if 30Y >5.50% washes out | 30Y weekly close <4.60% | 1–2 weeks / WATCH |
| Credit stance: AVOID passive loans / perpetual BDCs; selective senior secured only | Deploy if loan distress >25% under 80¢ | Loan default rate topping 6% | Defaults fall <3.5% + gates lift | 1–3 months / HIGH (avoid) |
No contradiction with open calls (BTC hold, ETH long, QCOM speculative, AEP/CEG/VST longs, ORCL wait): energy is the macro offset by design, and Oracle is idiosyncratic AI-capex risk, not systemic bank risk. No new systemic short — a 2008 put is overpaying for insurance with banks capitalized and IG fundamentals (6th straight quarter of double-digit EPS) intact.
A. Data snapshot · B. Crisis-distance model · C. Sources
| Metric | Level | Note |
|---|---|---|
| SPY / QQQ / DIA / IWM | 757.83 / 708.69 / 520.75 / 287.70 | SPY -0.6%, QQQ -1.06% on PPI + oil |
| TLT / XLF / XLE | 80.78 / 56.87 / 64.93 | TLT -1.16% as 30Y surges |
| NVDA / ORCL / MSFT / JPM | 218.36 / 152.94 / 492.44 / 353.56 | ORCL -5.4% regular, +~4% AH on beat |
| BTC / ETH / SOL; total cap | 77,136 / 2,459 / 99.85; $2.64T | BTC dom 58.5%; crypto G greed 69 |
| PPI Aug / Core PPI | 5.4% y/y (4.8% Jul); core 4.6% (4.3%) | +0.4% m/m; energy +4.2% m/m |
| Fed funds / 10Y / 30Y / Mortgage | 3.50–3.75% / ~4.80% / 5.20–5.30% / 6.76% | 30Y first since 2007 |
| Oil | Brent >$105; WTI ~$100 | Hormuz + Bab el-Mandeb risk |
| Credit stress | PC defaults 6.1% Jul; loans 4.9%; 21% tech <80¢ | Fitch/S&P/Guggenheim |
| AI funding | Hyperscaler bonds $132B YTD; capex ~$777B '26 | 11% of IG supply; Vanguard |
| Oracle Q1 FY27 | Rev $19.35B; IaaS +121%; RPO $664B; FCF -$5.4B | $20B ATM equity done; FY27 rev ≥$90B |
Score = 0.35×oil persistence + 0.30×real-rate tightness (30Y − CPI) + 0.20×credit distress (loan <80¢ share + PC defaults) + 0.15×bank leverage. Today: medium-high oil, high real rates, medium credit, low bank leverage → selective squeeze, not systemic. Load-bearing assumption: banks can absorb private-credit spillover because CET1 >12% and exposure is via commitments, not SIVs. If banks hid larger total-return-swap / NAV-loan exposure (FSB opacity warning), the answer flips to 2008-lite — watch Q3 bank earnings warehouse provisions and BDC NAV cuts as the check.