The artificial intelligence investment cycle is expanding beyond technology and equity markets. As companies borrow heavily to finance data centres and computing infrastructure, AI is becoming an increasingly important source of exposure for global credit investors. KKR is now warning that the scale and concentration of this borrowing could generate broader market volatility if AI growth slows.
The numbers illustrate the scale of the transformation. Technology companies are projected to invest nearly $8tn in AI infrastructure by 2030, while AI-linked debt already stands at approximately $600bn. According to KKR, this represents around 6.3% of the US investment-grade market.
AI Is Becoming a Credit Market Story
The rapid expansion of AI requires enormous amounts of physical infrastructure. Data centres, advanced chips, power capacity and communications networks all require capital, and technology companies are increasingly turning to debt markets to finance this expansion. Borrowing now extends across investment-grade bonds, high-yield debt and securitised products.
KKR estimates that as much as one-fifth of the investment-grade index could eventually become exposed to AI-related risks. This would represent an unusually concentrated investment cycle for a market traditionally associated with relatively conservative securities. KKR manages $796bn across private equity, private credit and other markets, giving its assessment relevance across several areas of institutional finance.
The current 6.3% AI-linked share of the US investment-grade market already stands well above historical sector concentration levels. KKR found that the highest sector exposure in the index averaged only 2.6% over the past 29 years. As AI borrowing expands, investors may therefore find that seemingly diversified portfolios are increasingly influenced by the same underlying technology investment cycle.
The $6tn Financing Gap
The expected scale of AI capital expenditure creates another challenge. Even if major hyperscalers including Oracle, Amazon, Meta, Google, Microsoft and SpaceX each reached a maximum index weighting of 3% — a typical single-issuer limit for bond portfolios — KKR estimates that they could raise only up to $1.7tn through the high-grade bond market.
Compared with projected AI infrastructure investment of almost $8tn by 2030, that would leave more than $6tn to be financed through other channels. This could accelerate the use of private credit, securitisation, leasing structures and other financing arrangements as companies search for additional capital.
Off-balance-sheet financing makes the picture more complex. Credit guarantees, leases and future commitments may not always appear as conventional corporate borrowing. As a result, KKR argues that the true AI exposure of some investment portfolios could be considerably higher than headline debt figures suggest.
Concentration Is Becoming a Central Risk
The key issue is not simply the amount of borrowing, but the extent to which different assets depend on the same economic assumptions. A bond issued by a technology company, a loan financing a data centre and a securitised infrastructure product may appear to be separate investments. However, all could ultimately depend on sustained demand for AI computing capacity.
This creates correlation risk. If AI infrastructure demand continues to grow rapidly, the financing system may remain supported by strong technology-sector cash flows. But if growth slows or expected returns from AI investment fail to materialise, several apparently different areas of the credit market could experience pressure simultaneously.
Early signs of greater investor caution are already visible. Investors have recently demanded higher yields on some AI-linked debt to compensate for perceived risks. Loans connected to Oracle data centres have also faced pressure because of construction delays and permitting challenges, increasing scrutiny of contractual protections and financial backstops.
Debt Investors Face Different Economics
The risk-reward structure is particularly important for fixed-income investors. Equity investors can potentially benefit from substantial upside if AI companies grow faster than expected. Credit investors generally receive a predetermined return, meaning their upside is limited even when the underlying business performs exceptionally well.
This difference makes concentration more consequential. Taking greater exposure to the same AI ecosystem does not necessarily provide bondholders with the potential gains available to shareholders. Investors therefore need to determine whether yields adequately compensate for sector concentration, structural complexity and potential correlations across issuers.
Contract terms are another critical factor. KKR highlighted the importance of examining leases, guarantees and force majeure provisions, particularly when debt is purchased in secondary markets where investors may have less visibility into the original agreements. The credit strength of a major technology counterparty may provide limited protection if the underlying financing structure is weak.
What the Market Should Watch Next
The AI infrastructure boom is increasingly becoming a capital markets story as well as a technology story. With nearly $8tn in projected investment by 2030 and only part of that requirement potentially financed through traditional investment-grade bonds, new financing structures are likely to play a growing role.
For institutional investors, monitoring exposure will require looking beyond individual issuers. Data centre financing, corporate bonds, private credit, leases and securitised products can create overlapping exposure to the same group of technology companies and the same assumptions about future AI demand.
AI-linked debt of $600bn may represent only an early stage of this financing cycle. As trillions of dollars flow into infrastructure, understanding where credit risk ultimately sits — and how different exposures are connected — could become as important as assessing the growth prospects of artificial intelligence itself.