AI data center financing is heating up, but credit investors are becoming more cautious

Overview

This report summarizes Morgan Stanley's discussions with macro accounts, core credit investors, corporate credit specialists, and securitized credit specialist investors regarding AI/data center financing. The report notes that AI-related debt financing remains a core topic in credit markets, with issuance volumes driving its above-consensus supply forecasts and influencing sector preferences and spread direction. Investors remain willing to allocate to AI/data center exposure, but sentiment has become more cautious due to macro and thematic risks.

Core views

The core view is that fixed income markets have shown flexibility in meeting AI financing demand, and companies are actively broadening financing channels, including non-USD debt, equity, DDTL, private 4(a)(2) formats, and securitized financing. In the IG market, the focus is on the financing mix of hyperscalers and ORCL, the impact of supply on spreads, credit quality differentiation, and index eligibility; the HY market is more focused on construction risk, structural evolution, and refinancing; and the securitized market focuses on supply-demand balance and whether it will be crowded out by corporate credit markets. The report believes AI-related credit may continue to see spread differentiation and maintains an underweight view on IG Tech.

Analysis framework

The report uses investor interview feedback, recent issuance and secondary market performance, capex forecasts, structured financing terms, and cross-market supply-demand comparisons to assess the relative value of AI/data center debt. The analysis is organized across IG, HY, and securitized credit markets, and summarizes the main sources of risk using a framework of construction risk, tenant risk, asset risk, corporate risk, and macro/thematic risk.

Methodology notes

• credit_risk_frameworkData Center Financing Risk Framework

  construction risk, tenant risk, asset risk, corporate risk, macro/thematic risk

  Construction risk focuses on whether data centers can be delivered on time and on budget; tenant risk focuses on tenant payment ability and credit quality; asset risk focuses on whether data centers will still have demand and technological suitability at lease renewal; corporate risk focuses on whether issuers can monetize the AI investment cycle; macro/thematic risk includes oversupply, AI demand falling short of expectations, scarcity of power and supply chains, local opposition, and regulatory restrictions.

• relative_valueCross-Market Credit Comparison

  structure and risk allocation among IG, HY, ABS, and CMBS

  The report compares how different credit markets handle construction risk, tenant risk, amortization mechanisms, index eligibility, and refinancing paths to determine which structures better offset AI thematic risks.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• IG Tech / hyperscalers credit

  The main issuers of AI capex and data center financing

  Strengths

  Higher issuer quality, broad investor base, and the ability to broaden funding sources through non-USD debt and equity.

  Weaknesses

  Supply continues to increase, while off-balance-sheet commitments, leases, guarantees, and SPV liabilities raise actual leverage.

  Comparison

  Compared with other IG sectors, the AI complex has already seen spread widening, while other industries have been relatively stable.

  Risks

  Credit quality differentiation, ORCL downgrade, oversupply, and delayed realization of investment returns.

• HY data center construction bonds

  Used to finance the data center construction phase

  Strengths

  Shorter maturities, with the potential for take-out or early redemption through lower-cost capital after construction completion.

  Weaknesses

  Construction delays, cost overruns, and limited issuer operating history are the core issues.

  Comparison

  Compared with IG, HY investors are currently more focused on project-level construction risk rather than unsecured corporate risk.

  Risks

  Construction risk, uncertain refinancing path, and rating upgrades dependent on project completion and structural performance.

• ABS / CMBS data center financing

  Securitized financing based on stabilized data center asset cash flows

  Strengths

  Usually tied to completed, cash-flow-generating assets, where the assets themselves matter more than the corporate entity.

  Weaknesses

  CMBS usually lacks amortization mechanisms, and both ABS and CMBS can also face ARD, LTV, DSCR, and refinancing constraints.

  Comparison

  Compared with corporate credit, securitized supply has not yet been clearly crowded out, but corporate credit markets may provide larger scale and competitive funding costs.

  Risks

  The securitized market's ability to absorb a refinancing wave, sponsor risk, asset re-leasing risk, and supply-demand mismatch.

Key data

• Global AI-related debt issuanceabout $336bn YTDAs of the report sample period, covering global credit markets.
• IG AI-related debt issuanceabout $222bn YTD; about $257bn including the AVGO chip financing dealAs of 2026-07-10.
• Morgan Stanley forecast for IG AI-related issuance$350-400bnIncluding $250-300bn from high-quality hyperscalers and ORCL, and about $100bn from other AI-related debt.
• Morgan Stanley total IG supply forecast$2.25tnThe report says this forecast is above consensus, with YTD at about $1.33tn.
• Hyperscalers and ORCL non-USD issuance$62bn YTDFinancing currencies include EUR, GBP, CHF, CAD, and JPY.
• Hyperscalers and ORCL USD issuance$132bn YTDInvestors believe this may be below the full-year $250-300bn range.
• Estimated 2027 capex growth+54%Based on revised estimates from Morgan Stanley equity analysts for high-quality hyperscalers and ORCL.
• Size of data center construction bonds not included in indicesabout $53bn outstandingMainly involves 144A-for-life construction bonds that are currently not index-eligible.

Impact & implications

For investors, the AI financing theme is no longer just a story of new supply, but is gradually shifting toward a comprehensive pricing of credit quality, financing structure, long-term asset suitability, and refinancing feasibility. Shorter duration, amortization, construction guarantees, IG wrappers, high-quality tenants, and stronger disclosure may be more favored; bonds lacking index eligibility, with longer duration, no amortization, or high asset re-leasing risk may face larger discounts. If rates fall while AI supply remains high, demand may slow and lead to broader repricing in credit markets.

Risks

• Data center overbuilding due to AI compute demand falling short of expectations or oversupply.
• Improvements in technological efficiency, chip iteration, liquid cooling, and changes in power systems may expose existing assets to obsolescence risk.
• Power, equipment, permitting, zoning, and local opposition may cause construction bottlenecks or delays.
• Actual leverage of hyperscalers and ORCL rises through off-balance-sheet commitments, leases, guarantees, and SPV liabilities.
• Long-duration, non-amortizing, or 144A-for-life structures may underperform due to limited liquidity and index eligibility.
• If rates decline while supply remains elevated, credit market demand may slow and trigger broader repricing.

What to watch

• Whether the pace of AI-related IG issuance in the second half of 2026 reaches or exceeds first-half levels.
• Whether 2027 capex forecasts for hyperscalers and ORCL continue to be revised upward.
• After ORCL rating changes, whether investors shift from supply discussions to fundamental stratification.
• Whether non-USD debt, equity, DDTL, and private 4(a)(2) formats continue to expand.
• Whether HY construction bonds are completed on schedule, and the take-out or refinancing path around the first call date.
• Whether 144A-for-life data center construction bonds convert to unrestricted CUSIPs and become eligible for index inclusion.
• Whether the ABS and CMBS markets can absorb a potential wave of data center refinancing.