Investment Strategy Insights: Financing the Next Phase of AI

Our team believes that differences of opinion not only make markets, but also often foreshadow substantial market moves. To share their diverse viewpoints and outlooks, investment leaders from our global multi-asset, equities and fixed income teams meet monthly. This report reflects the most recent monthly discussions and debates and intends to provide insights on the topic of the month along with snapshots of asset class views and convictions across the firm.

Financing the Next Phase of AI

The AI narrative has largely been defined by demand. Whether discussing enterprise adoption, infrastructure buildout or returns on investment, investors have spent the past two years debating how large the opportunity could become. Increasingly, however, attention is shifting toward a different question: How will the next stage of growth be financed?

Recent earnings reports suggest the underlying economics remain supportive. While investor attention has focused on large capital raises and share price weakness among some hyperscalers, companies continue to reaffirm plans for significant increases in AI-related spending next year. Cloud and AI businesses continue to generate strong growth and attractive returns, while many of the operational challenges facing these companies are concentrated in slower-growing legacy businesses rather than in AI itself. Incremental AI related investments also appear to be generating returns above existing corporate averages, supporting management teams' willingness to continue investing despite rising funding requirements.

The more pressing issue today is financing rather than fundamentals. While the economic case for AI investment remains largely intact, the scale of capital required to sustain the buildout is becoming increasingly difficult to ignore. Across credit markets, the common theme is not deteriorating credit quality, but the growing challenge of absorbing an exceptional volume of issuance over a relatively short period of time.

The investment-grade market provides perhaps the clearest example. Hyperscalers now represent a rapidly growing share of the investment-grade universe, and investors are demanding increasingly large concessions to absorb new issuance. Recent debt offerings have generally been well received, but order books have become less oversubscribed and borrowing costs have gradually increased as investors seek additional compensation for growing concentration and supply risk. This appears to be a technical rather than a fundamental issue, reflecting the practical challenge of financing one of the largest capital investment cycles in modern history.

Similar dynamics are beginning to emerge across other financing channels. In asset-backed securities (ABS), investor demand for data center exposure has become increasingly saturated despite generally healthy asset performance. New transactions have frequently required wider spreads, shorter maturities or structural changes to attract capital. Likewise, leveraged finance markets have absorbed substantial volumes of AI-related issuance over the past year, yet investors have become increasingly selective as future funding needs continue to expand. Private credit markets have also adopted a more cautious stance, particularly toward technology-related lending, reflecting a broaderreassessment of risk across capital markets.

Importantly, there is little indication that these financing challenges are being driven by weaker end-demand. Hyperscalers continue to pursue ambitious expansion plans, NVIDIA chip pricing remains firm and rising, and activity levels across the broader AI ecosystem remain robust. Evidence from private markets suggests that suppliers supporting utility infrastructure, grid expansion and data center construction continue to experience significant pricing power and elevated profitability, highlighting the strength of demand flowing through the system.

The challenge, therefore, increasingly lies on the supply side of capital rather than the demand side of compute. Higher financing costs, investor concentration limits and a more selective funding environment are likely to increase the cost of future projects. As a result, the pace of AI infrastructure expansion may moderate from the extraordinary levels observed over the past two years. In our view, however, this should be viewed as a new phase of slower and more sustainable growth, rather than a reversal of the investment cycle. The adjustment process of this downshift in rate of investment can be painful for the market, but it sets up a longer cycle and lowers the risk of overbuild.

Ultimately, the discussion reinforced a theme that has emerged repeatedly throughout this year's AI debates. The primary risks facing the sector are becoming less about demand and more about execution, financing and infrastructure. While the rate of growth is likely to decelerate as capital becomes more expensive and funding markets become more selective, there is little evidence to suggest that the broader AI investment cycle is nearing an end. Rather, the market appears to be transitioning from a period in which demand was the primary focus to one in which financing capacity is becoming an increasingly important consideration. The AI investment cycle remains intact, but its future trajectory is likely to be influenced as much by capital availability as by underlying demand. Such an environment favors the largest and strongest balance sheets, and may become increasingly challenging for smaller, leveraged players. Active credit selection takes center stage.