Investors Are Growing Wary of AI-Related Debt

AI hyperscalers have ramped AI data-center capex and debt issuance, with AI-related global debt totaling $236B as of May 31 and projected to reach $570B by end-2026. Bond investors are cooling: hyperscaler bond cover ratios fell from 5x in February to 2x in July, implying higher funding costs ahead. In parallel, SpaceX bonds are trading at a 1.62 percentage-point average spread over Treasuries (BB/high-yield) despite $29B in long-term debt, signaling rising credit skepticism that could presage weaker equity performance.

Analysis

The key market mechanism is not "AI is over"; it is that the marginal dollar of AI capex is now being priced by credit investors rather than equity investors. When bond demand weakens this quickly, the cost of funding new data-center buildouts rises before reported earnings do, which usually forces either slower capex growth or lower future FCF conversion assumptions. In the next 1-3 months, that is most likely to hit the highest-multiple hyperscalers first as investors start haircutting long-duration AI spending plans rather than current revenue.

The second-order winners are capital providers that can step in where public bond demand is fading: structured credit, private credit, and debt capital-markets intermediaries. That makes APO a more interesting relative long than the hyperscalers if the market continues to prefer private, higher-yielding financing channels; MS can still collect fees, but a flatter primary market means less easy issuance volume and less embedded carry. The losers are not just AMZN/GOOG/META/MSFT equity holders; any adjacent power, colocation, and data-center supply chain names that depend on an uninterrupted capex acceleration story are vulnerable to order deferrals if ROIC skepticism spreads.

Contrarian view: the consensus may be over-reading a funding signal as a demand signal. These firms still have immense internal cash generation, so the near-term risk is not balance-sheet stress but a multiple compression as investors demand proof that AI capex is accretive. The thesis is falsified if hyperscalers keep raising capex guides while bond spreads stabilize or tighten into the next earnings cycle, or if AI monetization accelerates enough to re-rate the entire funding stack.

Key Decisions for Investors

• Short a basket of hyperscaler equities (AMZN/GOOG/GOOGL/META/MSFT) on the first rebound toward pre-event levels; use a 1-3 month horizon and size for multiple compression rather than bankruptcy risk.
• Pair trade: long APO / short MSFT for 1-3 months as a relative expression of tighter public credit vs. private-credit displacement; stop if hyperscaler spreads tighten back materially or issuance resumes at prior clearing levels.
• Buy 2-4 month put spreads on XLK or QQQ if you want broader AI sentiment exposure without single-name earnings risk; best entry is on any oversold bounce rather than into weakness.
• Set an alert on hyperscaler bond spreads and cover ratios: if new issue demand stays below ~3x or spreads widen another 25-50 bps, expect equity analysts to cut AI capex ROI assumptions and downgrade forward EPS multiples.
• If you need a cleaner long, favor capital-structure beneficiaries over equity beta; avoid chasing the hyperscalers until the market shows that incremental AI spending is self-funding rather than debt-funded.