Foreign capital backs North American AI expansion

Foreign investors are increasingly playing a role in financing North America's artificial intelligence (AI) infrastructure development, as technology firms seek to fund their spending nearing $1 trillion, according to JPMorgan Asset Management. Charles Wu, head of Asia-Pacific alternatives institutional client strategy at JPMorgan Asset Management, stated this during a conference in Singapore.

He noted that banks and investors from various regions, such as Asia and the Gulf, are becoming more involved in providing funding for computing assets primarily located in North America. Wu explained that while US hyperscalers mainly rely on operating cash flow to finance their AI capital expenditures, they also raise approximately $250 billion from bond markets and an equivalent amount through bank loans.

This trend is raising questions about the level of debt associated with a few major technology companies. JPMorgan Asset Management is monitoring factors like lease obligations, concentration risk, and financing through structures that may not be reflected in corporate balance sheets. The organization estimates that the complete data-center build-out could cost around $5 trillion by 2030, with about $2 trillion potentially financed through investment-grade credit markets.

JPMorgan strategists reported that hyperscaler bond issuance totaled $194 billion in the first half of 2026 and is expected to reach $279 billion by year-end. This financing has sparked discussions on whether traditional credit indices adequately capture the full exposure of AI infrastructure. The Bank of England highlighted the growing AI-related debt burden, estimating its issuance at around $450 billion by early September, which is more than double the amount in 2025.

The central bank cautioned that a potential market correction could result from unrealized expectations regarding future AI earnings. Notably, some investors are adopting a differentiated approach to AI-linked debt, recognizing that borrowers vary in their financial strength. JPMorgan Asset Management asserts that major hyperscalers still maintain robust balance sheets with debt-to-earnings ratios well below typical investment-grade issuers.

In contrast, Jean-Christophe Aubert, senior director of infrastructure investments at PSP Investments, highlighted that firms are scrutinizing concentration risks and the ability to monetize large infrastructure assets. The geographical distribution of AI infrastructure investments may shift as the focus transitions from training models to inference or everyday use, as proximity to users becomes a critical factor for computing needs.