AI Bond Divergence Exposes Weaknesses in AI Lending

Artificial intelligence-related bonds are showing increasing divergence in risk assessment, with investors distinguishing between AI computing providers and the data centers that support them. CoreWeave, an AI cloud computing startup, saw its bonds plunge last week, highlighting growing concerns about highly leveraged AI firms. While data center bonds have surged to over $100 billion this year—about 25% of all AI-related dollar-denominated debt—these are generally seen as less risky than bonds issued by AI companies themselves. Data center bonds backed by tenants like Microsoft, which maintains a triple-A rating, trade with tighter spreads than those of highly leveraged firms like CoreWeave, whose junk-rated bonds reflect its heavy debt dependence. The bond market's differentiated response offers insight into where investors see the most significant risks in AI financing. According to Pimco strategist Lofty Karoui, AI-related credit deals should not be viewed as one large transaction, as differentiation will continue to increase across the sector.