Moody’s Signals Rethink Of CRE Risk Exposure As Off-Book AI Data Center Leases Top $662B

Why It Matters

The hidden lease exposure could strain credit ratings and limit financial flexibility for the tech giants as AI spending accelerates, prompting investors to reassess risk profiles.

Key Takeaways

• $662B off‑balance‑sheet leases exceed hyperscalers’ adjusted debt
• Leases start 2025‑2031, shifting liabilities onto balance sheets
• GAAP rules let firms exclude renewal guarantees from current reporting
• Residual value guarantees hide up to $28B potential payouts
• Moody’s may adjust debt metrics to reflect cash‑flow exposure

Pulse Analysis

The race to power generative‑AI models has ignited an unprecedented wave of data‑center construction, with hyperscalers such as Amazon, Meta, Alphabet, Microsoft and Oracle signing multi‑billion‑dollar lease agreements. While these contracts secure the physical footprint needed for AI workloads, they are structured to remain off‑balance‑sheet until the facilities are occupied. This accounting treatment masks a $662 billion liability—more than the combined adjusted debt of the five firms—creating a latent risk that will materialize as the leases commence.

Under U.S. GAAP, a lease liability is only recognized when the lessee is "reasonably certain" to renew, a threshold that many AI‑focused leases deliberately avoid. Companies instead rely on residual‑value guarantees (RVGs) that act as back‑stop payments for landlords if a lease is not renewed or the asset’s market value falls short. Because the probability of exercising these guarantees is deemed low, the associated obligations are excluded from current financial statements, allowing firms to preserve headline leverage ratios while still committing to massive future cash outflows.

For investors and credit analysts, the opacity of these off‑balance‑sheet commitments raises red flags. Moody’s warning suggests a potential shift toward cash‑flow‑based adjustments, which could tighten borrowing capacity and trigger covenant breaches if AI market dynamics soften. Regulators may also scrutinize whether existing lease accounting standards adequately reflect the scale of AI infrastructure spending, prompting possible revisions that would force earlier liability recognition. Stakeholders should monitor lease commencement schedules and any Moody’s methodological updates as early indicators of emerging financial strain.