Borrowing by AI companies represents a ‘mounting potential threat to the financial system,’ top economist says

Moody’s Analytics chief economist Mark Zandi warns that large tech and AI companies are issuing more debt now, even after inflation adjustments, than during the late-1990s dot‑com era—with the 10 largest AI-related firms (including Meta, Amazon, Nvidia and Alphabet) set to issue more than $120 billion this year—and much of it is incremental borrowing rather than mere refinancing. Markets have been willing to absorb long-dated, low‑spread issuance as hyperscalers are treated increasingly like quasi‑utilities, and demand signals such as Nvidia’s 66% yoy AI data‑center revenue growth support continued investment, but analysts caution that rapid tech churn in chips risks creating stranded hardware, AI pure‑plays like OpenAI lack profit cushions, and U.S. energy capacity could become a bottleneck. The combination of higher leverage, potential revenue disappointments and concentration risk creates a mounting financial‑stability and counterparty exposure issue investors should monitor closely.

Analysis

Moody’s Analytics chief economist Mark Zandi highlights that large tech and AI-related firms are issuing more bonds today, even after inflation adjustments, than during the late-1990s dot‑com era; the 10 largest AI companies — cited examples include Meta, Amazon, Nvidia and Alphabet — are forecast to issue more than $120 billion this year and much of that is incremental borrowing rather than simple refinancing. Zandi warns that elevated, creative borrowing could become a systemic threat if companies miss investor expectations and equity valuations decline, putting stress on newly issued debt.

Market participants are tolerating long-dated, low-spread issuance because hyperscalers are increasingly viewed as quasi-utilities and bond issuance is described by strategists as the cheapest route to finance a multi‑decade, trillion‑dollar AI infrastructure buildout. Nvidia’s reported 66% year‑over‑year AI data‑center revenue growth in Q3 is cited as proof of demand that has encouraged issuance, per the article and sentiment signals showing positive bias toward NVDA.

Key operational and credit risks flagged include rapid obsolescence of capital-intensive AI hardware due to fast chip innovation cycles, concentration and counterparty risk from unprofitable pure‑plays such as OpenAI, and a potential bottleneck from U.S. energy capacity that could slow realization of deployed capacity. The article and associated signals imply differentiated issuer risk: large, profitable hyperscalers can self‑fund but smaller or specialized AI players and dependent vendors (Oracle is noted as exposed) face higher default and stranded‑asset risk.

Investment implications are primarily credit‑centric: rising leverage across major tech issuers increases systemic exposure and warrants closer monitoring of maturities, covenant quality, spread behavior and execution against AI demand metrics; mixed sentiment and a cautious market impact score (0.45) suggest potential for spread widening if revenue trajectories diverge from current bullish expectations.

Key Decisions for Investors

• Monitor incremental bond volumes, tenor and covenant terms for the largest AI issuers (Meta, Amazon, Nvidia, Alphabet) and be prepared to reduce exposure to newly issued unsecured long‑dated paper if spreads widen
• Favor debt and equity exposure to cash‑rich hyperscalers with diversified revenues that can absorb buildout costs, and avoid or hedge exposure to lower‑rated pure‑play AI companies and capital‑intensive hardware suppliers vulnerable to rapid obsolescence
• Use Nvidia's AI revenue trends and U.S. energy capacity indicators as primary signal events to re‑rate demand assumptions, and tighten credit lines or add hedges if earnings or infrastructure constraints disappoint