AI’s infrastructure boom is getting more leveraged — and harder to track
Get the Finance newsletter
Daily finance — markets, central banks, M&A, the prints that move money. Free.
- Nvidia is partnering with Apollo, Blackstone, BlackRock, Brookfield, KKR and Goldman Sachs to develop AI infrastructure platforms that could exceed $500 billion, with CEO Jensen Huang telling CNBC Nvidia's chips are now an "investable infrastructure asset."
- Goldman Sachs analysts estimated hyperscalers carry $1.5 trillion in combined lease commitments for data centers, R&D, offices and equipment — up from roughly $200 billion five years ago — including about $1 trillion in "uncommenced" leases not yet on balance sheets (Aug. 6 note).
- PIMCO's Lotfi Karoui said the AI capex cycle is on track to be the largest investment cycle since 19th-century railway construction, with consensus forecasts pointing to hyperscaler capex surpassing $1 trillion per year from 2027 onward with "no clear signs of moderation" (Aug. 11 commentary).
- Situational Awareness, an AI-focused hedge fund, collapsed after leveraged equity bets failed, with assets falling from $45 billion to about $10 billion; Citadel then bought its publicly listed positions at a discount, with concentrated holdings including SK Hynix and CoreWeave.
- JPMorgan CEO Jamie Dimon told CNBC margin debt is "pretty high" and increases amplified volatility risk, while Ninety One's Sahil Mahtani argued elevated earnings expectations — not leverage itself — are the main risk to the AI trade, given "historically high" equity concentration.
- AIMA (Alternative Investment Management Association) distinguished the Situational Awareness collapse from Archegos in 2021 and the 2022 UK LDI gilt crisis, arguing hedge fund leverage does not currently pose systemic risk based on available evidence.
- Karoui noted hyperscalers are diversifying debt issuance into euro, sterling, yen, Swiss franc and Canadian dollar markets, with Amazon and Alphabet's euro-denominated bonds outperforming their U.S. counterparts in a potential sign of "demand fatigue" in dollar markets.
Why it matters: Goldman Sachs estimates $1 trillion of hyperscaler AI debt sits in 'uncommenced' leases not yet on balance sheets — obligations that will eventually inflate stated leverage — while a single leveraged AI hedge fund saw assets plunge from $45B to $10B in one tech sell-off, proving how abruptly crowded AI trades can unwind.
Ask SkimNews
