UniCredit: AI Debt and Bonds - FJElite

The five major hyperscalers plan to invest USD 4-6tn in data centres in 2025-30. As a large part of the funding will come from public and private debt markets, this historically unprecedented capex cycle creates a number of credit-relevant ramifications, including for European markets.

Hyperscalers are increasingly tapping foreign debt markets, having issued EUR 52bn of euro-denominated reverse Yankees in 2025 and 2026 to date. This corresponds to around 25% of total euro-denominated reverse Yankees placed during that time. We see scope for up to an additional EUR 5-10bn this year. So far, this issuance has had an adverse impact on the relative performance of the technology sector, although its impact on government bonds has been contained, given the limited issuance relative to outstanding government debt and hyperscalers’ avoidance of maturities heavily used by governments.

Roughly 40% of total investment will be funded via the bond market. While increased bond issuance will expand the technology sector’s weight and duration in credit markets, the greater concern is the risk of excessive investment, deterioration of credit quality and defaults. Moreover, rapid technological progress and improvements in AI model efficiency could reduce future infrastructure needs, making long-term demand difficult to forecast. Separately, the outcome of the US midterm elections increases the likelihood of greater regulation of hyperscalers, potentially creating additional challenges for competitive dynamics, financing structures and asset valuations.