The $570 Billion AI Debt Wave Is Straining U.S. Credit Markets — and Investors Are Starting to Push Back

The AI buildout is being financed by debt at a scale the U.S. corporate bond market has never seen. In 2026 alone, global AI-linked debt issuance has reached roughly $335 billion year-to-date — more than double the pace of 2025 . Morgan Stanley forecasts that total AI and hyperscaler-related issuance could climb to $400 billion in investment grade (IG) plus $65 billion in high yield and loans globally, potentially accounting for nearly 20% of all issuance . Bank of America revised its 2026 hyperscaler IG debt forecast upward by 25% to $175 billion after Amazon's record-breaking €14.5 billion euro bond debut in March . Bankers believe AI volume could push total U.S. IG issuance above $2 trillion for the first time ever .

These numbers are not just big. They are beginning to distort a market that had been calm. Below is what the evidence shows about where the strain is appearing and what it means for investors.

Spreads Are Widening and Hyperscalers Are Underperforming

The first visible signal is in credit spreads. Since September 2025, spreads on A-rated and above hyperscaler bonds have widened by 14–29 basis points, compared to just 4 bps for the broader IG market, according to ICE Data Indices . This divergence is directly attributed to the glut of issuance and growing concentration concerns. The IESE business school analysis notes that spreads have "so far absorbed the supply remarkably well, but not without strain" — even top-tier tech giants like Amazon, Meta, and Microsoft are seeing their bonds trade at wider levels .

Amazon's July 2026 Deal: Demand Is Cooling

Amazon's $25 billion "surprise" bond sale on July 7, 2026, drew orders of only 2.5x the bonds on offer, down from 3.2x in March. According to BofA, this was the weakest new-issue performance for any hyperscaler since Meta's $30 billion sale in October 2025 . Amazon had to dangle an extra 18 to 21 basis points of yield on its longest bonds to get the deal done . When Amazon set terms, AI-related bonds sold off sharply as investors raised cash to participate. The WSJ reported that spreads on hyperscaler bonds "increased significantly" on the day of Amazon's announcement .

SpaceX's Debut 30-Year Bond: A Case Study in Shifting Sentiment

SpaceX issued $25 billion in IG bonds on June 23, 2026, across five tranches. The 30-year tranche ($3.5 billion) carried a coupon of 6.65% — the highest in the deal . The 2036 tranche priced at a spread of 1.4 percentage points above Treasuries, which was already 0.4 pp wider than the average BBB-rated bond .

Within weeks, the situation worsened. By July 9, the bonds were yielding 1.7 pp above Treasuries . By July 11–12, the 30-year bond yield had risen to ~7.3% (vs. the 30-year Treasury at ~5.06%), meaning the credit spread had widened to more than 2 percentage points — a level normally associated with BB (high-yield) rated bonds . The Financial Times described this as "BBB SpaceX treated worse than BB bonds" . Demand skewed heavily to shorter maturities at issuance , and the secondary market repricing signals that investors view long-dated AI-funded debt as carrying material uncertainty and term premium risk.

Investor Sentiment: Fidelity Is Backing Away

In its midyear assessment on June 3, 2026, Fidelity told clients it is shifting its main bond investments away from the rapidly expanding segment of IG bonds issued by Amazon, Alphabet, Meta, Microsoft, and Oracle to fund AI infrastructure . The decision is not due to default risk but because investors are no longer sufficiently compensated for the risks involved. "Currently, investors are not adequately rewarded for holding corporate bonds," Fidelity's assessment stated .

The WSJ reported in mid-July that Wall Street is "conveying a clear warning" to tech firms on their unprecedented borrowing spree, and the IG market has faced challenges accommodating the supply . Goldman Sachs notes that credit spreads remain very low overall and investors are not yet broadly concerned about credit risk, but acknowledges that as AI spending continues to rise, supply could begin to overwhelm demand . Apollo's chief economist Torsten Slok has asserted that this level of AI borrowing is already diminishing demand for U.S. Treasuries and other fixed-income assets, though other market participants say they haven't observed this yet in corporate bonds .

Risks to Financing Costs, Free Cash Flow, and Equity Valuations

With spreads widening 14–29 bps and SpaceX's long-dated bonds trading at distressed-like spreads, the marginal cost of new debt is rising. For issuers planning to come to market repeatedly, every basis point of spread widening adds billions in interest expense over time. PIMCO has flagged that even as hyperscalers start from strong balance sheets, rising capital spending and falling free cash flow signal a shift toward greater leverage . Aggregate hyperscaler capex on AI infrastructure is projected at $770 billion in 2026, with cumulative debt issuance through 2028 expected to exceed $1 trillion .

If debt-funded AI capex does not generate the promised returns, the combination of higher leverage, higher interest costs, and potential write-downs could compress equity valuations. The bond market's cooling reception — weaker demand, wider spreads, allocator pullback — is an early warning signal that the cost of capital is rising before the revenue benefits have materialized.

A Two-Tiered Market

The evidence confirms that while the overall IG market has not yet broken, the hyperscaler-specific segment is showing clear strain. The broader IG index remains calm, but hyperscaler bonds are experiencing weaker demand, wider spreads, and active portfolio rotation away from AI-heavy concentrations. The risk path runs through rising financing costs, to thinner free cash flow, and eventually to lower equity valuations if the anticipated AI returns fail to materialize on schedule.