Private debt, asset-backed-facilities, bonds.
AI-related companies, and the broad definition that comes with it, have tapped almost all forms of financing across different currencies to feed the enormous amount of capex needed to build its infrastructure. US hyperscalers are forecast to spend about $800bn in 2026 alone, according to Goldman Sachs estimates.
But broadly syndicated term loans sit in a corner of finance that has been barely touched by the wave of AI-related liquidity, at least in Europe. In the US, CoreWeave’s 2031 delayed draw term loans and QTS’s $3.25bn term loan B have both shown that loan markets are ripe for AI deals.
The European high yield market has seen some activity, but issuance has been overwhelmingly driven by non-European companies, as in the case of CoreWeave's €2bn 8.5% SUNs due 2032 and SoftBank’s €500m 2030 7.125% and €500m 8.000% due 2032. These notes mirrored a traditional corporate finance structure, making them accessible for institutional investors, one senior banker explained.
When in market, European data centre developers have fallen short of expectations and relied on cheaper bank facilities instead. PolarDC’s record Nordic high yield bondhas been the only European high...
