The primary risk for investment-grade AI debt is not weak company fundamentals, but rather massive supply overwhelming investor demand. In contrast, the high-yield market's main concern is construction risk, including project delays and cost overruns on new data centers, representing a shift to asset-level analysis.

The primary threat to today's tight credit spreads is not weakening demand but a sustained surge in supply, particularly from AI 'hyperscalers'. The concern is how this new debt is employed, as it could fundamentally deteriorate the issuers' balance sheets over time.

Unlike corporate and high-yield AI financing that funds new builds, securitized products focus on stabilized, cash-flowing, and often multi-tenant data centers. This structure avoids construction risk, offering investors a more mature risk profile centered on occupancy, churn rates, and overall demand for compute.

Despite market hype, Madison avoids asset classes like office and data centers. They view them as binary investments—success or failure—with high capital costs and low liquidity. They prefer residential assets where underwriting mistakes are less catastrophic, protecting investors from the potential for major principal loss.

While construction risk is the dominant concern for high-yield data center debt, the structural shortage of power and compute capacity makes it unlikely tenants will exercise termination rights over delays. This suggests that any project-related valuation dips are likely temporary, presenting attractive buying opportunities for investors.

While data centers are a hot commercial real estate (CRE) sector, the property-level investments offer narrow spreads unsuitable for hedge funds. A more compelling relative value play is in the high-yield corporate credit of companies providing essential technology and services to these data centers.

Blue Owl's decision to back out of financing an Oracle data center reflects a growing concern among lenders about overexposure to Oracle's massive AI infrastructure commitments. This suggests a potential funding bottleneck for the entire ecosystem as lenders become more cautious.

Evaluating data center investments is like analyzing net lease real estate. With a tenant like a MAG-7 company, the investment is primarily a bet on the counterparty's creditworthiness, not the long-term value or potential obsolescence of the physical data center itself.

The financial market for AI infrastructure is maturing and becoming more risk-averse. Investors who previously funded speculative data center builds are now demanding long-term customer contracts upfront. This shift de-risks new projects but also indicates that the era of 'build it and they will come' is ending.