Why fixed income is critical for the AI supercycle

The AI infrastructure buildout is increasingly a fixed income story. Total data center capital expenditures could reach USD 5.5 trillion from 2026 through 2030, according to J.P. Morgan Chase,¹ creating financing needs that extend well beyond corporate balance sheets.

Demand for compute capacity is accelerating across hyperscalers,² power utilities, semiconductor manufacturers, construction firms, real estate developers, and capital markets. The emergence of more advanced agentic AI systems has intensified infrastructure demands even further. As a result, capital markets are playing an increasingly central role in financing one of the most consequential investment cycles in modern economic history.

The five fixed income markets financing AI

Given the scale of investment required, debt is playing a critical role in financing AI infrastructure, and we expect issuance to increase over the next few years. Total AI‑related debt issuance for 2026 had already exceeded full‑year 2025 issuance as of late June, according to J.P. Morgan.

Much of this issuance will probably be long‑duration bonds, increasing the impact of new hyperscaler supply on the broad fixed‑income market. We don’t anticipate fading demand for AI‑related bonds, but issuers have started to offer more attractive spreads to offset the long‑duration supply pressure. The higher credit quality of many hyperscalers should continue to draw buyers to their bonds, although the scale of the new deals has started motivating investors to seek wider spreads.

Companies can primarily use five distinct fixed income markets to fund AI‑related capex, with some overlap between them: public investment‑grade corporate bonds, securitized credit, private credit and hybrid debt, leveraged finance, and equity‑linked debt.

1. Public investment‑grade corporate bond markets

Technology‑related issuance of investment‑grade corporates has risen sharply since mid‑2025, driven by hyperscaler demand. Several large platform companies have executed multi‑billion‑dollar multi‑tranche offerings in recent quarters, in some cases representing their largest‑ever financings. We believe AI‑related bond issuers will use the investment‑grade corporate market to meet the bulk of their funding needs.

As these issuers grow as a percentage of investment‑grade corporate credit benchmarks, credit curves³ may steepen as investors demand more spread for longer‑maturity debt. Within the technology sector, the market may further differentiate credit quality, leading to a tiering of tech credit spreads and possibly creating more prospects for active managers.

The new AI‑related corporate supply could create benchmark concentration dynamics similar to that in the banking industry after the global financial crisis of 2008–2009, when large banks recapitalized by tapping the investment‑grade corporate market. This led to financials becoming the largest sector in the major investment‑grade corporate bond indexes. The upward pressure on spreads from the heavy issuance also opened up opportunities for fundamental credit analysis to find sound credits that had been swept up in the negative technicals.

2. Securitized credit markets

Asset‑backed securities (ABS) and commercial mortgage‑backed securities (CMBS)⁴ in the securitized credit market provide another avenue for AI‑related funding, albeit on a much smaller scale than the investment‑grade corporate market. ABS or CMBS can be useful for financing data centers that are generating cash flows or for power generation agreements that produce cash flows under contract, for example. As more AI campuses mature, we think that securitization tools will play a growing role in refinancing activity.

3. Private credit and hybrid structures

While investment‑grade public companies have not historically relied heavily on private debt funding, these markets are a deep pool of capital to help finance AI capex. Capital‑intensive AI projects increasingly rely on custom financing structures that combine elements of corporate credit, project finance, and securitization. Issuers can structure these deals as off–balance sheet debt under certain conditions or with amortization schedules that match the cash flows generated by the project. Lease or service payment streams can enhance collateralization, and project operators can share credit risk with debtholders.

4. Leveraged finance: high yield bonds and loans

Leveraged finance (high yield bonds and broadly syndicated loans) is also likely to play a supporting role in funding the AI infrastructure buildout, particularly for non‑investment‑grade participants across the extended ecosystem. While the largest hyperscalers will primarily access investment‑grade and private markets, high yield bonds and loans are well suited for data center developers, power and infrastructure contractors, equipment suppliers, and emerging AI infrastructure platforms whose balance sheets or project risk profiles fall below investment grade.

Through the first half of 2026, neocloud companies—which provide “GPU as a service” infrastructure for AI providers—have increasingly used the leveraged finance market to raise funding. We expect that trend to continue into next year. 

According to J.P. Morgan, the leveraged finance markets have sufficient capacity to absorb a meaningful share of AI‑related funding over time—potentially on the order of USD 150 billion over the next five years. As a result, high yield bonds and loans are likely to finance discrete projects, growth platforms, and secondary beneficiaries rather than the hyperscalers themselves, making them an important—but potentially more cyclical and risk‑sensitive—component of the overall AI funding mosaic.

5. Equity‑linked securities: convertible bonds and preferred equity

Non‑investment grade issuers can also tap the convertible bond⁵ market to help fund AI‑related capex. Some neocloud companies have recently issued zero‑coupon convertible debt, minimizing their funding costs in exchange for potential dilution of their equity.  

At the other end of the credit quality scale, some hyperscalers have issued preferred stock, which combines characteristics of both debt and equity. Preferred equity entitles holders to a stream of higher dividends (similar to bond coupon payments) than common stockholders.

Preferred stock is generally not a viable funding source for smaller, riskier AI infrastructure companies, which may struggle to support a regular dividend obligation. By contrast, the strong balance sheets and cash flow profiles of hyperscalers provide them with the flexibility to consider preferred equity as part of their funding mix.

Highly complex environment demands integrated research capabilities

The AI environment is highly complex. AI infrastructure exposure manifests differently across chips, utilities, data centers, power generation, and industrial components. At the same time, as we have highlighted, AI‑related financing spans multiple markets and capital structures, from public and private credit to securitized and leveraged finance. Many of these are considerably more complex than a traditional investment‑grade corporate bond with a fixed coupon and maturity date. Evaluating opportunities and risks therefore requires expertise across asset classes, industries, and capital structures, highlighting the importance of integrated research capabilities.

An integrated research framework can help to compare corporate credit curves against amortizing private placement debt structures and assess whether project‑level credit spreads appropriately compensate for construction, technology, or tenant concentration risk. Such an approach may also help identify indirect beneficiaries of AI capex, including companies in the capital goods, utilities, or industrials sectors, whose cyclical uplift may be underappreciated by the market. From a portfolio construction perspective, this can help avoid concentration in a single expression of an AI‑related theme while capturing more stable multi‑industry benefits.

For example, data center developer Hut 8 recently used a project‑level financing structure to fund a large data center in Texas. Rather than simply issuing debt at the parent‑company level, Hut 8 formed a separate entity, Beacon Point, which issued approximately USD 4.3 billion of secured bonds to finance construction of the facility. NVIDIA will be the site’s sole tenant under a long‑term lease.⁶ In simple terms, investors are lending against a specific data‑center project: the bond proceeds help build the site, NVIDIA’s lease payments are expected to help repay the debt, and the bonds are secured by the project’s assets rather than by NVIDIA’s entire balance sheet.

This structure shows why AI infrastructure credit analysis can differ from traditional corporate bond analysis. Investors must assess not only the tenant’s credit quality, but also the lease terms, construction timeline, power availability, collateral package, and whether the data center would remain valuable if the tenant’s needs change. That makes the Beacon Point deal a useful example of how AI financing can combine elements of corporate credit, real estate finance, project finance, and securitized credit in a single transaction.

Managing risks in an era of transformative change

AI‑related bonds have meaningful risks, which can be managed via credit analysis and selection alongside thoughtful portfolio construction. Examples of risks that are important to monitor include:

Supply: The broadest risks are the potential for overbuilding and oversupply of data centers, chips, and power assets, or for AI capex to moderate relative to the current very elevated expectations.

Deal risk: Individual AI‑driven debt issues could be overly dependent on a limited number of hyperscaler tenants or constrained by regulatory or permitting delays. Utilities subject to regulatory price constraints may be unable to fund an elevated capex burden over time.

Portfolio construction: Overlapping AI dependencies across sectors may create portfolio construction risk.

The AI infrastructure supercycle is reshaping not only technology markets but also the fixed income landscape. As financing needs expand across public and private markets, we believe a cross‑asset approach to research and underwriting will be essential for identifying opportunities and managing risks across this rapidly evolving ecosystem.