Financing the AI boom: credit markets at a crossroads

AI: moving into the mainstream

The exponential growth in AI is driving a surge in capital expenditure (capex) with investments in hyperscale data centres, advanced semiconductor plants and power grids fuelling the current economic cycle even as traditional growth engines slow. Until recently, this accelerating AI expansion was predominantly an equity market story, but the way this boom is financed is now changing rapidly, with major implications for credit markets.

AI capex is driving a rapid shift across capital markets

Initially, tech giants relied on strong free cash flow and high equity valuations to support investments in data centres, chip development and cloud infrastructure. This funding model enabled sizeable capex while maintaining relatively conservative capital structures.

However, as demand for AI-driven technology and the resources to power it far exceeds supply, players are racing to secure the huge amounts of capital needed to fund their ambitions. For instance, some market estimates suggest that global data centre capacity alone may need to increase sixfold by 2030, which would require around US$3 trillion in investment. While hyperscalers may fund roughly half, the remainder must come from external sources as the amounts involved are simply too large, even for these deep-pocketed tech companies.

As a result, we are witnessing a transition from equity markets to credit markets as the key source of capital for AI infrastructure. Public and private bond markets will likely shoulder a significant proportion of funding needs in this capital cycle, with asset-backed securities, infrastructure funds and novel financing structures all playing a role. We believe the public investment-grade corporate bond market will become the preferred avenue for financing, given its size and ability to absorb large quantities of issuance on both the short and longer end of the curve.

This transition is happening at remarkable speed and comes with plenty of innovation as large, high-quality issuers seek to capitalise on the depth of demand in the public bond market. As more issuers across the AI ecosystem adopt bespoke financing solutions, we expect these novel deal structures to evolve further, with hybrid structures that blur the lines between corporate and project finance becoming increasingly normalised.

Critical questions for credit investors

The AI infrastructure cycle offers both promise and caution. While the long-term potential of AI is compelling, the pace and scale of financing raise important questions for bondholders, who have limited upside if AI projects succeed, but meaningful downside if financing structures falter or returns disappoint.

How durable are today’s AI infrastructure investments?
Over half of AI-related investment targets chips and hardware, but rapid advances mean these assets may have shorter lifespans than before. Data centres built for training large models could see their value shift as AI workloads evolve and supply catches up with demand. Issuers able to adapt infrastructure and balance sheets to changing technology will be better positioned to maintain credit quality and avoid stranded assets.

Will AI investments deliver the anticipated returns?
AI’s commercial potential is significant, but the timeline for adoption, productivity gains and profitability remains uncertain. Many businesses are still determining how to integrate AI, and revenue from new AI infrastructure may take time to materialise or face temporary setbacks. If the expected boost is delayed, some projects could struggle to meet return targets, increasing the risk of funding pullbacks or repricing in credit and equity markets.

Could looser lending standards make the market more vulnerable?
The surge in AI-related financing has led to more complex and less transparent structures, with some deals relying on looser underwriting standards. Recent US defaults — though not directly tied to AI — have revealed challenges and irregularities, reminding investors that excess and opacity can quickly translate into broader stress. As more risk is warehoused outside traditional banking channels, investors should remain alert to potential ripple effects across the capital structure.

The scale of AI-related investment has made the sector a significant driver of US economic activity, but this reliance introduces vulnerability. Within credit markets, this risk is illustrated by the growing proportion of AI-linked issuers in the liquid USD investment-grade universe. They now represent approximately US$1.2 trillion in debt, or 14% of the universe, surpassing even US banks. If treated as a standalone sector, this cohort would be the largest in the universe, underscoring how closely the credit cycle is now tied to the trajectory of AI.

Should confidence in the viability or sustainability of AI investments weaken, the impact could be swift and far-reaching, potentially triggering repricing and increased volatility across risk assets. A better understanding of AI’s trajectory is fast becoming a key input when seeking to establish the likely direction of the economic and credit cycle. It will also increasingly affect the fortunes of virtually all issuers, meaning careful analysis and ongoing vigilance by credit investors is imperative.

Unlocking the AI infrastructure opportunity

This wave of capital undoubtedly comes with new risks but also promising potential relating to areas of opportunity such as:

Differing performance of AI-related projects
Not all AI-linked issuers will benefit equally. We expect wide dispersion in the viability, scalability and monetisation of AI-related investments. Some projects may deliver durable cash flows and strong fundamentals, while others may struggle. This uneven landscape creates a rich environment for bottom-up credit selection, where understanding issuer-level fundamentals and capital structure positioning will be critical.

Deepening markets and evolving deal structures
The scale of AI financing is expanding the credit universe with new issuers and structures. Alongside an increase in traditional corporate debt supply, we are seeing an increase in joint ventures, securitisations and private placements. Recent deals are blurring the lines between public and private credit markets, combining investment-grade execution with private credit features such as bespoke structuring and asset-specific protections. This creates additional complexity for credit analysis, which is compounded by the rating agencies’ uneven and unorthodox treatment of these financing structures, particularly when debt is kept off balance sheet.

Potential policy tailwinds
AI infrastructure is increasingly supported by government policy — particularly in areas tied to energy, national security and digital sovereignty. These tailwinds may create durable demand in adjacent sectors such as utilities, infrastructure and clean energy. Issuers in these areas could benefit from more predictable cash flows and supportive credit profiles, offering long-term exposure to real economy assets indirectly linked to AI’s growth trajectory.

Together, we think these dynamics create an attractive backdrop for active credit investors. Success will depend on the ability to distinguish between the winners and losers and identify whether risk is being priced appropriately. Deep research, cross-sector insight and disciplined security selection will be essential to unlocking differentiated returns.

Turning insight into action

How to approach the AI investment theme ultimately depends on risk appetite, investment objectives and interpretation of the balance of risks and opportunities. But more than ever, it is important to keep the inherent asymmetry of investment-grade credit in mind — upside in the form of price appreciation, consistent income and principal repayment, but meaningful potential downside in the form of permanent loss of capital. Strict adherence to the following core principles can, in our view, help to manage that asymmetry:

• Maintain discipline: We still expect positive total returns as yields remain elevated. We anticipate an accommodative macroeconomic backdrop, robust corporate fundamentals and strong market technicals. However, as spread valuations are relatively tight, we believe investors should be highly selective about where to deploy risk.
• Seek relative value: With limited pricing dispersion, we think it is important to focus on sectors with supportive fundamentals and identifiable positive catalysts that may be indirect beneficiaries of AI. Conversely, we would advocate mitigating exposure to AI-sensitive issuers that offer no material spread advantage but are vulnerable to technical headwinds and uncertainty around future profitability, such as data centre REITS, as well as issuers that are exposed to deteriorating underwriting standards in private credit, notably insurance and business development companies (BDCs).
• Be dynamic: Flexibility and liquidity are central, in our view. In this fast-moving environment, we think investors need to be able to adjust exposures quickly to take advantage of market dislocations. A dynamic approach can also help to capture second- and third-order effects as the AI investment cycle unfolds, by acting as a provider of capital during periods of volatility.

The bottom line

We are excited by the opportunities this cycle presents and believe the far-reaching implications of AI capex across public and private markets are potential game changers. This backdrop suits a dynamic, research-driven approach focused on generating consistent returns through security selection. In our view, active management — backed by deep sector expertise and a cross-asset-class perspective — will be essential to navigate the rapid, complex developments unfolding across credit markets.