Private credit is holding firm despite AI debt wave: Moody's Ratings

Moody’s Ratings’ Marc Pinto said record credit default swaps are not a red flag for AI data-center debt, with stress instead concentrated in older, speculative-grade private-credit loans. He sees no near-term systemic risk, citing private credit’s small size relative to the $80 trillion in debt Moody’s rates.

Analysis

The useful distinction is between a market-price signal and underlying borrower stress: record CDS activity can reflect hedging or positioning and is not, by itself, evidence that AI data-center borrowers are deteriorating. The more plausible fault line is legacy speculative-grade private credit, where stale marks and infrequent transactions can delay recognition of weaker recovery values. That creates a potential lagged NAV and fundraising problem for exposed lenders even if it does not become a systemic event.

Near term, avoid treating broad AI-infrastructure credit as the same risk bucket as troubled older loans. Over 1–3 months, watch whether private-credit stress shows up in BDC non-accruals, payment-in-kind income, NAV marks, fundraising, or wider spreads; those would be more actionable than headline CDS volumes. Over 6–18 months, tighter or more selective lending could raise financing costs for marginal data-center projects while advantaging well-capitalized developers and projects supported by credible long-term customer commitments. The counterpoint is that concentrated customer exposure, power constraints, or delayed buildout could still weaken otherwise attractive infrastructure financing.

The contrarian risk is that “small versus rated debt” addresses system size, not opacity, leverage within funds, or correlated lender exposures. The thesis weakens if disclosed credit metrics deteriorate across multiple lenders, or if stress spreads from older loans into financing for new infrastructure. There is not enough borrower-level or portfolio-exposure data here to justify a directional credit trade.

Key Decisions for Investors

• Do not buy broad protection on AI/data-center credit solely because CDS activity is at a record; require evidence of borrower fundamentals or financing terms worsening first.
• For existing private-credit exposure, prioritize diligence on older speculative-grade vintages: track non-accruals, PIK income, realized recoveries, NAV marks, and fundraising rather than relying on reported headline yields.
• Consider a relative-risk posture rather than a sector-wide short: favor transparent, senior infrastructure exposure with verifiable customer commitments over opaque legacy speculative-grade lending, subject to underwriting and valuation checks.
• Set an alert for a broader credit signal: deterioration across multiple lenders’ reported credit metrics, widening private-credit spreads, or weaker financing availability for new data-center projects would challenge the contained-stress thesis.