Concerns over financing artificial intelligence (AI) infrastructure are increasing as credit markets assess whether they can support the large investments needed to build data centres, according to a Goldman Sachs research report. The report said high-yield bonds issued by data centre companies were facing pressure as investors examined the risks associated with the rapid expansion of AI infrastructure.

Goldman Sachs recommended evaluating each project separately rather than treating all AI infrastructure investments as having similar risk profiles. Amanda Lynam of the firm’s credit research team identified construction risks, borrowers’ ability to repay loans, refinancing requirements and claims against tenants as factors investors should examine when assessing data centre projects.

The report also highlighted power availability, regulatory approvals and insurance as issues that could affect project viability. The concerns come as rising demand for computing capacity drives investment in data centres and related infrastructure, while companies across sectors increase their exposure to the industry. This makes project-level assessment relevant to equity investors as well as bondholders.

Goldman Sachs cited Spanish renewable energy company Solaria as an example of the growing connection between energy businesses and data centre infrastructure. Solaria’s announcement of a data centre deal in Uruguay has attracted investor attention, while analyst Alberto Gandolfi assessed the risks linked to the transaction as skewed to the upside. The report said Solaria could make further announcements at its capital markets day in November.

The research concluded that AI infrastructure presents investment opportunities but requires scrutiny of construction costs, financing arrangements, power supply, permits and insurance. Investors must determine whether individual projects can manage these risks and obtain the funding required to support the sector’s continued expansion.