AI debt financing enters an acceleration phase, with investment-grade bonds still the core funding source

Overview

The report covers dollar debt financing channels needed for AI-driven data center expansion, including public, private, unsecured, secured, securitized, and structured financing. Morgan Stanley believes AI infrastructure financing will be a core credit theme for the next several years. AI-related issuance has already started strongly in 2026, and the issuer base is expanding from hyperscalers and semiconductor companies to data center REITs, leveraged-finance issuers, and securitized credit instruments such as ABS and CMBS.

Core views

The key conclusions are: first, the upward revision in AI-related capex will continue to support debt issuance across multiple channels; second, investment-grade bonds remain the main financing channel for AI and account for about 80% of all AI-related financing; third, the fundamentals of higher-rated hyperscalers remain healthy, with low leverage and large cash balances, but their index weight is still limited; fourth, the supply surge has already widened credit spreads, with spreads for AA-and-above hyperscalers widening 19 bp since last September versus only 2 bp for the index; fifth, securitized credit performance metrics remain solid for now, but the heavy supply backdrop may continue to pressure valuations.

Analysis framework

The report uses a cross-channel financing tracking framework to compare issuance volume, maturity structure, spread performance, and underlying-asset performance across investment-grade bonds, leveraged finance, high-yield bonds, data center REITs, ABS, and CMBS. It benchmarks AI-related financing against the public investment-grade index, high-yield index, semiconductor debt, and the data center securitization market, while drawing on data from Dealogic, PitchBook LCD, CreditFlow, TREPP, Bloomberg, and Morgan Stanley Research.

Methodology notes

• Credit market trackingCross-channel AI financing monitoring

  Split AI-related debt supply by financing channel

  The report breaks AI infrastructure financing into investment-grade bonds, leveraged finance, securitized credit, and structured financing channels to identify issuance pace, maturity structure, and spread pressure across funding sources.

• Credit quality analysisHyperscaler balance sheet assessment

  Leverage, cash balances, and rating differences

  The report argues that AA-and-above hyperscalers still have very healthy fundamentals, with low leverage and high cash balances supporting credit quality; ORCL is the relatively weaker case because it is BBB-rated and has a negative outlook.

• Securitized credit analysisData center ABS and CMBS performance metrics

  DSCR, LTV, occupancy, and trigger points

  The report evaluates data center securitized credit using debt service coverage, loan-to-value, and occupancy metrics, noting that current performance remains above trigger levels, but higher supply and the rate backdrop will affect spreads and financing costs.

Asset mapping & comparison

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

• investment-grade AI-related bonds

  The largest funding source for AI financing

  Strengths

  They account for about 80% of AI-related financing, have deep market liquidity, and can absorb large hyperscaler and data center-related issuance.

  Weaknesses

  The surge in supply may put pressure on valuations, and related spreads have already widened meaningfully since last September.

  Comparison

  Compared with leveraged finance and securitized credit, investment-grade bonds are larger in scale and have higher-quality issuers, but they are more sensitive to public credit supply and demand.

  Risks

  Persistent heavy supply, wider credit beta, and refinancing pressure from higher long-term capex expectations.

• ORCL.US

  The hyperscaler credit case highlighted by the report

  Strengths

  ORCL has become the largest non-financial issuer in the investment-grade index, and AI and cloud infrastructure demand are providing both financing and growth themes.

  Weaknesses

  ORCL is the only hyperscaler with a BBB rating and a negative outlook; its cash bond and CDS spreads have widened significantly since October and are wider than those of peers.

  Comparison

  MSFT, because of its AAA rating and no issuance since 2018, is described in the report as the best-performing hyperscaler; META is the second-widest-spread name, while ORCL faces the most acute spread pressure.

  Risks

  Rising lease liabilities, a large amount of lease commitments not yet commenced, downgrade risk, higher financing costs, and capital expenditure execution risk.

• data center ABS and CMBS

  An important securitized financing channel for stable data center cash flows

  Strengths

  Outstanding balance is about USD 60 billion, and performance metrics such as DSCR, LTV, and occupancy remain healthy and above trigger points.

  Weaknesses

  Rapid supply growth in 1Q26 widened spreads by about 30-50 bp, and CMBS financing costs have historically been slightly higher than ABS.

  Comparison

  The report expects ABS to be the primary tool for U.S. data center securitization going forward, while CMBS will still play a role; in Europe, CMBS may be more important because cross-jurisdiction financing is easier.

  Risks

  Further supply pressure, rates staying elevated, valuation adjustments, deteriorating cash flow performance, or cash sweep triggers.

• semiconductor bonds

  AI-ecosystem related but underrepresented in credit indices

  Strengths

  AI semiconductor demand remains strong, and the report expects about USD 60 billion of issuance from this sector in 2026.

  Weaknesses

  Semiconductors account for only about 2% of the IG ICE BAML index, versus about 14% in the S&P, so their representation in credit markets is limited.

  Comparison

  Compared with hyperscalers, semiconductor debt faces more moderate maturity pressure, and the report says it remains relatively manageable before the maturity wall peaks in 2030.

  Risks

  Industry cyclicality, supply constraints, slower AI demand, and spread pressure from higher issuance volumes.

• leveraged-finance AI issuers

  An emerging channel for AI-related financing expansion

  Strengths

  Names such as CRWV, APLD, CIFR, and WULF show that the AI theme has begun to enter the high-yield and leveraged-finance markets, and some are first-time issuers.

  Weaknesses

  Issuer credit quality is lower, historical market data are limited, and AI-related names make up only about 2% of the high-yield bond index.

  Comparison

  Compared with investment-grade bonds, leveraged finance is more sensitive to risk appetite and financing windows, and maturities are concentrated around five years.

  Risks

  Limited credit history for first-time issuers, M&A or change-of-control events, closed refinancing windows, and AI assets failing to commercialize as expected.

• data center REITs

  A representative case of AI financing expanding from tech issuers to real estate and infrastructure issuers

  Strengths

  Issuers such as QTS, DLR, and EQIX show that data center REITs are participating in investment-grade issuance, and demand for QTS's first high-grade debt issue was strong.

  Weaknesses

  Project financing depends on large cloud customers and lease performance, while property and rate factors may affect valuations.

  Comparison

  Compared with traditional hyperscaler bonds, data center REITs are closer to infrastructure and real estate credit, with a higher degree of cash flow securitization potential.

  Risks

  Customer concentration, construction delivery risk, power and land constraints, higher rates, and cap rate changes.

Key data

• Reference date2026-03-31Unless otherwise noted, the report information is as of 2026-03-31; some data center REIT issuance data are as of 2026-04-06.
• Share of AI-related financing from investment-grade bondsabout 80%The report estimates that investment-grade bonds are the largest funding source for AI-related financing.
• Share of higher-rated hyperscalers in the public investment-grade indexabout 4%Issuance from hyperscalers rated AA- or above is growing, but their share of the public investment-grade index remains small.
• Semiconductor share in the IG ICE BAML indexabout 2%Compared with about 14% in the S&P, semiconductor bonds have a much smaller weight in the investment-grade credit index.
• Share of AI-related names in the high-yield bond indexabout 2%The high-yield AI issuers included in the report are CRWV, APLD, CIFR, WULF, FLASHC, TRACTC, BLKPRL, and PFORGE.
• Outstanding balance of data center securitized creditabout USD 60 billionSecuritized credit has become an important financing source for stable data center cash flows.
• 2026 securitized credit issuance forecastabout USD 30 billionAbout USD 9.5 billion has been issued year to date, and the report expects issuance to rise sharply through 2028.
• Change in data center note spreadswidened by about 30-50 bpRapid supply in 1Q26 and broader credit widening pushed spreads wider on data center-related notes.
• QTS first investment-grade debt issuanceUSD 4.6 billionQTS issued this on 2026-04-06 to support a Microsoft-related data center project; according to Bloomberg, demand was about USD 12.5 billion, and pricing tightened 25 bp from the initial price guidance.
• Index headroom after ORCL's February issuanceUSD 57 billionThe report says ORCL still had USD 57 billion of room before surpassing JPM as the largest issuer in the investment-grade index after its USD 25 billion February issuance.

Impact & implications

For investors, the AI infrastructure financing theme has extended from the equity capex narrative into credit supply, spreads, and structured-product performance. Investment-grade bonds remain the core channel, but continued issuance may pressure valuations for hyperscaler and data center-related bonds. Securitized credit provides a new tool for financing data center cash flows; current asset performance remains healthy, but future spread stability will depend on supply intensity, the rate environment, and changes in buyer demand. ORCL is a focal point in the credit market because of its scale, rating, and disclosure of lease commitments.

Risks

• AI-related debt supply is increasing across multiple channels at the same time, which may continue to widen credit spreads.
• Upward revisions to hyperscaler capex expectations will increase financing needs and may also make investors more focused on leverage and free cash flow.
• If ORCL, as a BBB-rated hyperscaler with a negative outlook, continues to see rising lease liabilities or financing needs, its credit spread could come under further pressure.
• Data center ABS and CMBS are currently performing well, but heavy supply, rising rates, and valuation adjustments could weaken the market's ability to absorb new issuance.
• Most leveraged-finance AI issuers are new or have short credit histories; if AI demand or capital market windows change, default and refinancing risk will be higher.
• The report discloses that Morgan Stanley may have business relationships with the covered companies, and investors should treat this research as only one input in investment decisions.

What to watch

• Whether 2026 AI-related issuance continues to run meaningfully ahead of the 2025 pace.
• Whether investment-grade bonds maintain their roughly 80% share of AI financing and how higher supply affects index spreads.
• Further spread changes for AA-and-above hyperscalers relative to the investment-grade index.
• Follow-up disclosures on ORCL's cash bond and CDS spreads, rating outlook, lease liabilities, and lease commitments not yet commenced.
• Whether the 2026 forecast of about USD 30 billion in data center securitized credit issuance is achieved.
• Changes in DSCR, LTV, occupancy, cash sweep triggers, and cap rates for data center ABS and CMBS.
• Follow-on dollar and non-dollar issuance demand from data center REITs such as QTS, DLR, and EQIX.
• The 2026 issuance outlook of about USD 60 billion for semiconductors and changes in the maturity wall before 2030.
• Trading performance and refinancing capacity of high-yield AI issuers such as CRWV, APLD, CIFR, and WULF.