Here is why we see opportunities in rising AI debt

We’re seeing signs investors are worried about the rapid increase in debt issuance by AI hyperscalers and AI-related companies.¹ Some are interpreting this growth as a sign that credit in the sector is deteriorating; as a result, they’re reluctant to invest in the sector. We think this hesitation can lead to missed opportunities.

Increased debt loads are not a proxy for deteriorating credit. That’s because much of this debt is being issued by companies with low leverage and rising earnings, meaning they’re well-positioned to support that debt. Credit risk is not about absolute dollars raised. It is about what those dollars do to leverage, interest coverage and liquidity from a starting point that, for hyperscalers, is unusually strong.

With that in mind, we believe that widening spreads in some investment grade hyperscaler bonds can be an opportunity to add selectively to positions. Fundamentals still look durable, despite heavy issuance—even if AI capex takes time to translate into incremental earnings.

“AI debt” is a varied ecosystem

AI debt growth spans the breadth of the credit markets, but it is not evenly distributed.² AI debt sales in the U.S. dollar public markets have already reached $260 billion year to date.³ They are still dominated by the mega-cap hyperscalers, which are among the highest-quality credits in investment grade. Said differently, the vast majority of the debt is being raised by the companies with the greatest capacity to support it.

What about further down the quality spectrum? Leveraged credit markets are naturally home to companies with weaker credit fundamentals. We see reasons to lean more defensively in high yield: Issuer balance sheets are generally less robust, with many credits increasingly reliant on significant future EBITDA growth.

This high yield issuance has increased, but it’s still small compared to the overall market: We estimate that high yield AI-related debt is approximately $62 billion, a mere 4% of the overall high yield market. For leveraged loans, this is even smaller, at 1% of the overall market.⁴

Private credit adds another layer, including debt tied to the buildout of AI physical infrastructure. However, the market is difficult to measure with precision, and alternative investment funds are significant players. A helpful way to frame it is by “where you sit” in the ecosystem. While AI-related debt is growing across segments, in public credit markets, this is largely an investment grade story.

Issuance is elevated, but context matters

While AI-related debt issuance has risen significantly, it should be considered in the context of the broader investment grade market. Forecasts suggest $2 trillion in U.S. investment grade debt will be issued this year. After subtracting roughly $1 trillion in expected maturities and $500 billion in coupon income, this implies markets must absorb $495 billion in new debt.

While $2 trillion is an all-time record, the $495 billion net supply figure is less than net issuance in 2020. It also comes at a time of healthy corporate bond demand.⁵

Still, we are monitoring the changing composition of this issuance. If tech and hyperscaler debt comprise about 12% of the U.S. investment grade market today, sustained issuance may inflate this sector to the largest index weight, surpassing banks, which have dominated for nearly two decades. If that happens, investors who simply hold passive investments may wind up owning more of this exposure than they realize.

Why we’re comfortable with fundamentals

Today, these companies have strong credit profiles (excluding Oracle). This is reflected in their credit ratings, which generally fall between AA (very low default risk) and AAA (lowest expectation of default risk).⁶ Although free cash flow may be compressed as capex rises, we still believe these companies possess a great deal of flexibility.

Despite a year of heavy debt issuance, leverage levels haven’t moved much, as the new debt has been accompanied by rising EBITDA. However, these companies would be less levered than the average investment grade issuer, even if their earnings were not growing at all.

There is also off-balance sheet exposure to consider, but at present, our analysis suggests it doesn’t change the dynamic materially. While we acknowledge this is an item to monitor given that companies are holding more debt off their balance sheets, leverage levels in our view currently remain manageable.⁷

What about rising interest expense, from spreads or rates? Again, context is critical: Hyperscalers have a lot of room to absorb increased interest expense while maintaining investment grade profiles. On average, they have enough earnings to cover their interest expense 50 times over, which compares to roughly only 9 times over for the average investment grade company.⁸

So why have spreads widened? In our view, recent moves are tied to the cadence of issuance and uncertainty about what additional issuance lies ahead—not a repricing of credit risk. This makes it more technical than fundamental, and as such, we believe this should ultimately correct. Importantly, widening has not translated into broad spread widening for the overall investment grade index— suggesting the impact has been contained to just hyperscalers.

Conclusion

AI financing will likely remain choppy. Public equities, alternatives and credit markets all have meaningful exposure, but each of these has a distinct risk/reward profile. At the low end of the ratings spectrum, we take a more cautious approach. Across high yield corporate and data center debt, there is meaningful variation in underlying fundamentals and end-demand is considered more speculative, requiring an even more careful approach to credit selection. Private capital is expected to play a critical financing role. Real asset funds and select private credit strategies can give investors access to AI-buildout opportunities, offering the potential for robust returns in exchange for incremental risk and reduced liquidity. For investors seeking lower-risk and high-quality opportunities, select investment grade hyperscaler debt may offer attractive prospects. In that setup, spread widening driven by supply can create entry points as prices decline—as we think this widening is not justified by fundamentals.

We can help

For more information about potentially incorporating AI-related debt investments into your portfolios, or about credit investments in general, contact your J.P. Morgan team.