The AI 'hyperscalers' are on a $1T borrowing binge after years of printing cash
12 hours ago
- #AI-infrastructure
- #capital-expenditure
- #corporate-debt
- Major capital spending booms historically end in bankruptcies, consolidations, and victors benefiting from infrastructure.
- The AI buildout is led by five hyperscalers (Alphabet, Amazon, Meta, Microsoft, Oracle) with commitments nearing $1 trillion, primarily for data centers.
- Hyperscalers are issuing bonds to fund AI infrastructure, with $121 billion in new debt issued in 2025, up from $40 billion in 2020.
- Bond investors focus on risk compensation, wary of overinvestment leading to supply gluts, as seen in past cycles.
- AI buildout shifts internet companies from asset-light to asset-heavy models, affecting financial profiles and investor valuations.
- Long-term bonds (30-100 years) are being issued, attracting pension funds and insurers, signaling confidence in hyperscalers' stability.
- Oracle stands out with lower credit ratings, higher borrowing, and aggressive financing strategies compared to other hyperscalers.
- FOMO drives competitors into escalating investments, risking overcapacity, as Moody’s warns of on- and off-balance-sheet debt risks.
- Despite risks, hyperscalers' strong balance sheets suggest resilience, but outcomes will only be clear post-buildout.