AI Infrastructure Domino Model — Interactive Dashboard
AI Infrastructure Domino Model
Built July 22, 2026 · By Pascal ⚙️ · ~$397B probability-weighted expected losses across three interlocking systems
Soft Landing
30%
Soft Landing
Capital Destroyed$30–50B
Pension Losses<$10B
Banking ImpactNone
GDP ImpactPositive
Timeline2027: Oracle upgrades; 2029 refi smooth
EV of Losses$12B
Modal Outcome
45%
Managed Correction
Capital Destroyed$200–400B
Pension Losses$30–60B
Banking ImpactMinimal
GDP ImpactNegligible
TimelineQ1 2027: junk; 2028 correction; 2029 refi painful
EV of Losses$135B
Tail Risk
25%
Cascade
Capital Destroyed$800B–1.2T
Pension Losses$100–200B
Banking ImpactPolitically Toxic
GDP Impact−0.3 to −0.5%
TimelineQ3 2026: miss; 2027: Ch.11; 2028–29: crisis
EV of Losses$250B
Probability-Weighted Expected Capital Destruction · Total: ~$397B
3%
34%
63%
Soft Landing: $12B EV
Managed Correction: $135B EV
Cascade: $250B EV

Live Baseline — July 22, 2026

🟡
Oracle 5yr CDS
203 bps
18-yr high · Jul 21, 2026
🟡
Oracle Stock
$127
$366B market cap
🟡
OpenAI ARR
$25B
Flat since Feb 2026
🔴
Anthropic ARR
$47B
~2× OpenAI · 35–40% enterprise
🟡
Oracle Credit
BBB−
S&P stable · Moody's Baa2 neg
🟡
CoreWeave Debt
$30B
Junk-rated B+/Ba3/BB−
🔴
Oracle 5yr Default Prob
15.6%
Market-implied · CDS-derived
🔴
Capex / Revenue
3.7×
Sustainable threshold: <2.0×
Oracle FCF FY2026
−$23.7B
Est. −$42B in FY2027
Oracle RPO
$638B
47% = OpenAI contracts
Total AI Infra Commitments
$1.2–1.5T
Off-balance-sheet SPV complex
Construction Loan (Refi 2029)
$38B
JPM + MUFG · Refi cliff 2029–30
OpenAI Cash (Est.)
~$40B
Burning ~$2B/month
Oracle FY2026 Revenue
$67.4B
FY2027 target: $80–85B

Monthly Timeline · Jul 2026 — 2030

👆 Click any month to see scenario signals for that period

OpenAI Revenue Gap Calculator

Assumptions

OpenAI ARR Growth Rate (YoY)40%
Oracle Annual Obligation (2028)$45B
Total Infrastructure Costs (2028)$170B
Available Cash (2027 start)$40B
Cash Exhaustion Estimate
Adjust sliders to calculate

Year-by-Year Coverage Analysis

YearOpenAI RevOracle ObligationTotal CostsAnnual DeficitRevenue CoverageBarStatus

The Domino Chain — Cascade Path

Click any node to expand details. Status reflects current July 2026 conditions. Colors: ■ healthy · ■ stressed · ■ triggered · ■ future risk

Investor Exposure Map

Hover over any institution to see their specific exposure and risk profile

17 Falsifiable Predictions Tracker

Interactive Decision Tree

Historical Parallel — Telecom vs. AI Infrastructure

📡 Telecom Overbuild 2000–2003
🤖 AI Infrastructure 2026–2030
Key Difference: Telecom peak-to-major-bankruptcy: ~18–24 months (2000 peak → Jan 2002 Global Crossing). AI has a longer fuse — hyperscalers are cash-generative. Correction will be slower but potentially larger in absolute dollar terms.
Key Similarity: 24 of 30 largest telecom companies went bankrupt by 2003. If Oracle (the "Global Crossing" of AI) fails, it takes Stargate, Vantage Data Centers, and the construction loan syndicate with it. But data centers (unlike fiber) will be absorbed by surviving hyperscalers at distressed prices.

Capex Supercycle — The Overcapacity Gap

Hyperscaler Capex vs. AI Infrastructure Revenue

Current 3.7× ratio Sustainable: <2.0×

~45% of current capex is building ahead of demand — creating ~$300B/year in "speculative" infrastructure

Total Capex (Top 5 Hyperscalers)
AI Infrastructure Revenue
2× Revenue Threshold (Sustainable)
YearCapex (Top 5)AI RevenueRatio2× ThresholdOvercapacity GapStatus