AI's $650 Billion Funding Wave: Credit Crisis or Just a Repricing?

AI capex is hitting credit markets when spreads leave little room for error

This is a funding surge meeting a thin cushion. Hyperscalers are targeting roughly ~$650bn in AI-related CapEx this year, and if a large share is funded through public bonds, the supply pressure can show up quickly. The timing matters because investors are being asked into that supply surge while US high-yield spreads at 289 bps still sit in the bottom decile. In other words, compensation is low and room for error is tighter.

Why tight spreads make the timing sensitive

Tight spreads are exactly what encourage issuance. When pricing is compressed, primary demand can look comfortable until it starts testing secondary-market resilience. The bullish read is that much of today's AI spend is being led by strong operators that are well-capitalized with strong cash flow. The more cautious read is that even healthy fundamentals may not fully protect spreads if risk appetite turns.

Why this still looks like a concentration, not a broad credit break

This supply wave is concentrated rather than evenly spread across the credit universe. Tech/AI-linked issuers in U.S. IG bond indices represent ~15%, so the immediate pressure is less about systemwide credit failure and more about whether a concentrated block of high-profile issuers can be absorbed without a larger repricing.

The borrower matters more than the AI label

The key question is not whether AI-related debt is real; it is. The more important question is who is borrowing, because that determines where spread stress is most likely to start.

Record IG supply adds to the backdrop

US investment-grade supply is now expected to rise about 25% to a record $2.25 trillion. At the same time, a global USD 5 trillion–USD 7 trillion global capital expenditure (capex) wave over the next five years is being discussed as AI competition pushes hyperscalers and related infrastructure players to spend aggressively. That combination makes this a meaningful supply environment, even if the strongest issuers can still fund comfortably.

Top-tier issuers look more resilient

The largest cloud vendors and other top-tier AI spenders have direct access to public IG markets and stronger balance sheets, so the most likely first-order effect is marginal spread pressure rather than an immediate widening episode. For now, the clearest credit distinction remains between well-capitalized leaders and borrowers that depend more heavily on layered or project-level funding.

The weaker part of the stack could feel stress first

The more delicate risk sits lower in the funding stack, where data-center securitizations, project finance, and private credit can introduce extra complexity. That matters because funding needs do not disappear when one channel tightens. If private or structured markets become harder to access, marginal borrowers may look to other lanes, which can push more refinancing activity back toward public bonds.

Reuters also notes that pressure in private credit has been a drag on inflows, which supports the view that investor caution may first show up outside public spreads before it fully feeds back into them.

Positioning should focus on borrower quality and funding structure

This looks more like a concentration trade than a blanket call on credit. With Tech/AI-linked issuers in U.S. IG bond indices represent ~15%, the safer preference is toward borrowers that are well-capitalized with strong cash flow.

What to watch:

• Issuance that moves from strong balance-sheet operators to more complex or collateral-dependent structures
• Signs that private credit and securitized channels are becoming less accessible or more expensive
• Spread behavior in adjacent credit segments if lower-quality exposure migrates toward more liquid markets

The practical takeaway is straightforward: stronger issuers are more likely to absorb the supply shock, while the real pricing test probably comes from borrowers where repayment depends more heavily on intricate structures or assets that could be more sensitive to delays in AI revenue cycles.