The global credit market remains under two pressures that appear parallel but share a common origin: one driven by aggressive corporate borrowing, exemplified by AI infrastructure, and the other by France’s fiscal fragmentation and deteriorating sovereign creditworthiness…
The global credit market is experiencing two seemingly unrelated but fundamentally linked strains, both influenced by the same funding environment: on one side, the massive financing demands driven by AI infrastructure development; on the other, worsening fiscal and political uncertainty in France. Credit default swaps (CDS) for both asset classes have widened significantly recently, as investors demand higher compensation to continue financing corporate expansion and government debt.
This does not mean that AI companies and the French government face the same level of default risk. More importantly, the market is reassessing how much pressure the model of “high leverage, long-term returns, and continuous financing” can withstand in a high-interest-rate environment.
AI funding shifts from valuation disputes to credit risk
The most noticeable changes in the AI sector have appeared in the bond and CDS markets. According to FactSet data, CoreWeave’s five-year CDS spread rose to approximately 1,066 basis points as of Wednesday, indicating that the cost for investors to purchase default protection on its debt has reached an extremely high level. Oracle’s CDS spread over the same period increased to about 308 basis points, up from around 141 basis points at the start of the year; SoftBank’s spread rose to approximately 432 basis points, nearing a four-year high.
These companies occupy central positions in the AI cloud computing, data center, and capital investment chains, but share the common characteristic of requiring continuous massive capital investment.
NVIDIA CEO Jensen Huang stated this week that demand for "new cloud" enterprises remains extremely strong, while acknowledging that the industry is facing "capital scarcity": the need for funding clearly exceeds the available supply.
This is becoming a new paradox in AI trading. While the market previously focused on whether AI revenues could justify high valuations, the question has now expanded to whether companies can continuously raise sufficient funds at acceptable costs.
S&P Global warned this week that major cloud providers, including Amazon, Microsoft, Alphabet, Oracle, Meta, and SpaceX, may be exposed to higher AI-related debt risks than reported in their public financial statements.
Some data center projects are completed through special financing structures, long-term procurement commitments, or credit support provided to partner companies, so certain risks do not directly appear in traditional debt projects.
S&P expects that AI capital expenditures by major hyperscale cloud providers between 2025 and 2030 could total over $7 trillion. While rating agencies currently do not view the credit ratings of large tech companies as facing immediate downgrades, assessments are increasingly dependent on whether future AI revenues can offset today’s rapidly rising capital costs.
The market has already begun signaling price movements. The yield on CoreWeave’s bonds maturing in 2031 previously surged above 16%, and even after a recent decline, it remains elevated at approximately 15%.
The cost of credit protection in France has surpassed that of Italy.
Pressure from the other end came from France. France’s five-year sovereign CDS spread rose from around 30 basis points at the start of the year to nearly 90 basis points on Wednesday, surpassing Italy’s level of approximately 67 basis points. Meanwhile, France’s 10-year government bond yield breached 5%, and the yield spread between French and German 10-year bonds widened to over 160 basis points at one point.
Recent market volatility has intensified even further. According to Reuters data, the yield spread between French and German 10-year government bonds reached as high as approximately 178 basis points on Thursday, one of the highest levels since the European debt crisis in 2011; France’s 10-year yield peaked at around 5.14%.
Market concerns primarily focus on two issues: France’s public debt has approached 120% of GDP, and fiscal deficits have remained persistently high; at the same time, domestic political divisions are making it increasingly difficult to reduce the deficit.
The European Central Bank has previously downplayed expectations of direct support for the French bond market. President Lagarde stated that the “primary solution to France’s issues lies domestically” and must be addressed through fiscal policy. This means that if the current rise in yields primarily reflects France’s own fiscal fundamentals rather than disorderly market dysfunction, Paris will find it difficult to simply rely on the ECB to lower its financing costs.
France also faces a new rating review. Moody’s plans to reassess France’s Aa3 sovereign rating on Friday. Given that bond yields and CDS spreads have already risen significantly, how rating agencies evaluate France’s fiscal consolidation capacity will become a key determinant for the next round of market pricing.
The real risk is that financing costs create a negative feedback loop.
AI debt and French government bonds are not the same type of asset, but the current market is reevaluating them using similar logic: Can future cash flows or fiscal revenues afford increasingly higher financing costs?
For AI companies, if bond yields continue to rise, the capital cost of building data centers will increase further; higher interest expenses mean AI operations must generate profits more quickly to justify prior capital expenditures. If profitability is realized slower than expected, investors may demand even higher financing costs, creating a vicious cycle of “more expensive financing—worse cash flow—wider credit spreads.”
France faces a similar logic: the higher the yield, the greater the government’s future interest burden; increased interest payments worsen the fiscal deficit, prompting investors to demand a higher risk premium.
Credit stress is no longer confined to individual AI companies. The five-year spread of the North American CDX High Yield Index stood at approximately 355 basis points as of Tuesday, widening by about 39 basis points since the end of last year, indicating that investors are demanding higher compensation for overall low-rated credit risk.
Thursday’s market performance further reinforced this shift in risk appetite. Concerns over AI capital expenditures triggered a sharp decline in tech stocks, with the Nasdaq Composite falling 2.6% as U.S. investors rotated into Treasuries for safety; Meta’s stock dropped approximately 7% after announcing higher AI spending requirements.
Therefore, whether this round of "stress test" has ended cannot be determined solely by a single day’s bond rebound or tech stock stabilization. What truly needs to be observed is whether the funding market is willing to once again provide long-term capital to AI capital expenditures and highly indebted governments at lower costs.
As capital demand continues to rise and investors simultaneously demand higher future cash flows and greater fiscal consolidation capacity, credit spreads may remain a key source of market volatility in the next phase.