CITIC Securities: What signals does CDS spread reveal under the AI financing boom?

The Zhitong Finance App learned that CITIC Securities released a research report saying that the scale of AI capital expenditure continues to increase, and the financing costs of related entities are getting higher, compounded by changes in the macro environment in the third quarter, driving investors' concerns about the credit risk of AI technology companies and demand for hedging to rise at the same time, driving the AI technology enterprise credit default swap market to show a trend of rising trading volume and interest rate spreads. Although changes in market sentiment have generally widened CDS interest spreads for related companies, they have not evolved into an overall rise in industry credit risk, and CDS interest spreads of major AI-related companies have shown gradient differentiation. In a context where AI infrastructure construction and bond financing are still in the expansion stage, the cross-sectional differentiation of CDS interest spreads and their dynamic evolution may become an important indicator for observing the accumulation and transmission of credit risk in the current AI capital expenditure cycle.

CITIC Securities's main views are as follows:

The expansion of AI capital expenditure is compounded by rising financing costs, and the demand for credit risk hedging from technology players continues to heat up.

Investments in computing power infrastructure are driving up financing demand, pre-capital expenditure and delays in cash flow accentuate maturity mismatches, and market concerns about repayment capacity and refinancing pressure are increasing. Since the third quarter, changes in the macro environment have increased the market's sensitivity to credit risk, and CDS trading activity has increased at the same time as interest spreads. According to Bloomberg data, the average monthly turnover of AI-related technology entities under the single name CDS increased by 156.6% in the third quarter of 2026. CDS interest spreads for major technology companies generally widened and continued to rise since October, reflecting deepening market concerns about the implementation of investment returns and financing sustainability.

Bond financing in the AI-related industry is characterized by a misaligned distribution in the number of entities and the scale of issuance. The number of companies in the software service industry is larger, while telecom operators have a larger financing volume.

According to Bloomberg's caliber, there have been 1,490 effective debt issuers in the AI-related industry since 2023. Among them, there are 659 issuers in the software and service industry, corresponding to the amount of bonds issued since 2023 of US$494.23 billion. In contrast, there are 243 issuers in the wired and wireless telecommunications services industry, and the cumulative amount of bonds issued reached US$891.55 billion. In terms of rating distribution, there are 225 entities with valid ratings from the three major international rating agencies, accounting for 66.1% of the amount of bonds issued by related industries since 2023. Among them, there were 128 investment-grade entities, and the corresponding bond issuance scale accounted for 50.5% of the total sample. Among them, the BBB rating range accounted for the highest share of issuance, reaching 22.7%.

The CDS interest rate spreads of AI technology entities are divided in a gradient, and credit revaluation of highly leveraged companies has been particularly prominent in the past three months.

According to Bloomberg's caliber, as of October 20, 2026, CoreWeave's 5-year CDS spread reached 1,120.7 bps, and the increase in the past three months reached 270.9 bps. This is due to its highly leveraged computing power development model and the continued pressure on its own free cash flow. Meanwhile, Oracle's 5-year CDS spread, which had previously had its credit rating downgraded, also rose to 329.1 bps, widening 115.3 bps in the past three months. In contrast, with abundant cash flow and steady leverage, the average CDS interest spreads in the past three months were 51.5 bps and 42.9 bps, respectively. The overall pattern of differentiation was obvious.

Compared with the historical cycle, the current round of capital expenditure cycle related industry bond financing showed the characteristics of high scale and long term.

According to Bloomberg data, the AI-related industry's annual bond issuance scale rose from US$305.91 billion in 2023 to US$779.89 billion in 2025, reaching a further US$964.67 billion by October 20, 2026, which is significantly higher than the annual bond financing scale of the Internet telecom and real estate cycle. In terms of term structure, there is a clear trend of long-term bond issuance. The share of bonds issued over 10.33 rose from 14.4% in 2024 to 25.0% in 2026, with the share of ultra-long-term bonds over 20.33 rising from 8.4% to 13.8%, reflecting the further extension of corporate financing periods driven by long-term capital investment.

The key to the trend of credit evolution in the subsequent AI capital expenditure cycle is whether the gap between capital investment and cash flow can be reduced.

If relevant entities subsequently recover cash flow through order fulfillment to cover debt principal and fixed expenses, the leverage pressure is expected to gradually ease. Looking forward to the future, interest spreads on single bonds are more likely to be affected by differences in maturity, liquidity, and terms. Single CDS uses a specific issuer as a reference entity, which is more suitable for pricing reflecting the entity's overall credit risk, but it is still disrupted by hedging demand and liquidity. Follow-up suggestions combine order fulfillment and cash flow fulfillment to continuously track CDS interest spreads and differentiation among entities to identify the accumulation and transmission of credit risk.

Risk factors: AI demand and commercialization progress fall short of expectations; overseas monetary policies exceed expectations; impact of individual credit risk events.