Nvidia’s five-year credit default swap spread climbed to about 0.82% intraday on July 27, a rise of roughly 14 basis points that marked its largest one-day increase since the contract began actively trading in November 2025, according to ICE Data Services data cited by Bloomberg. The move drew attention to a developing concern in credit markets: whether chip suppliers could increasingly be asked to support the financing of the AI infrastructure that drives demand for their products.
A CDS is a contract used to insure against a borrower defaulting on its debt. A higher spread means that protection has become more expensive, signaling that the market sees greater credit risk than before. Nvidia’s spread remained relatively low in absolute terms, and the move did not suggest an immediate expectation of financial distress. Its size mattered because Nvidia has generally been associated with substantial cash generation and a comparatively lighter balance-sheet burden than companies building and operating large data centers.
Oracle’s five-year CDS traded around 1.25% over the same period, Bloomberg reported, placing it above Nvidia on the same measure. The difference reflects how credit markets are separating AI beneficiaries according to their roles in the infrastructure buildout: Oracle faces direct construction, power and data-center spending commitments, while Nvidia is being examined for the possible financial support it may extend to customers purchasing its equipment.
Financing questions reach chip suppliers
The reassessment comes amid reports that OpenAI and projects linked to SoftBank discussed financing guarantees of as much as $250 billion. Final terms, including the guarantees’ caps, triggers and accounting treatment, had not been publicly set out.
Such arrangements can reshape the economics of a technology supply chain. A guarantee may leave a company with no immediate cash payment, but it can create a contingent obligation if the customer or tenant later fails to meet its commitments. In a large AI project, that risk could emerge years after equipment has been delivered, particularly if data-center revenue develops more slowly than the spending required to build and power the facilities.
Credit markets tend to focus on that downside scenario: who is contractually required to pay if a project falls short, whether commitments remain off-balance-sheet, and whether a supplier’s exposure is capped. These details can determine whether a support arrangement is viewed as a manageable commercial incentive or a potential future liability.
The concern is particularly relevant in AI infrastructure because many of the largest projects require coordinated spending on chips, networking, memory, land, power generation and physical data-center construction. Revenue may arrive only after capacity becomes operational and customers begin using it at scale. That timing gap can be difficult for companies carrying heavy capital-expenditure programs or debt-funded expansion plans.
Ohio site illustrates the scale of planned buildout
One project connected to the reported guarantee discussions involved SB Energy, a SoftBank unit, and the Portsmouth Site in Pike County, Ohio. A March notice from the U.S. Department of Energy described plans for 10 gigawatts of new power generation at the site, intended to support a 10GW data-center buildout.
A facility of that size would require an unusually large commitment across electricity supply, construction and computing equipment. The Department of Energy notice placed the project within a growing race to secure power for data centers, where access to generation capacity can be as consequential as access to advanced chips.
Nvidia has also been connected with an announced AI plan and cooperation framework with SK Group described as exceeding $500 billion. The framework included work on next-generation memory with SK Hynix and a 2GW data-center component involving SK Telecom.
The SK relationship points to the interdependence of AI suppliers and infrastructure operators. Advanced memory is critical for training and running large AI models, while data-center operators need sustained access to chips, networking systems and electricity. Large cooperation agreements can help coordinate those pieces, but they also make it harder for markets to assess where commercial contracts end and financial support begins.
Oracle faces a more direct funding test
Oracle’s higher CDS spread tracked concerns over debt, capital spending and the period between building infrastructure and collecting enough customer cash flow to cover it. The company has become increasingly tied to cloud and AI capacity expansion, a strategy that can require significant up-front investment before the resulting data centers are fully utilized.
Nvidia’s business model has traditionally offered more insulation from that cycle because it sells high-value computing equipment rather than taking primary responsibility for constructing facilities. Any participation in customer financing would narrow that distinction. If suppliers begin using their own balance sheets or credit standing to help customers secure funding, the financial risk could move upstream from data-center developers to the companies selling the technology.
That does not mean every financing guarantee would produce a material loss. Contracts can include collateral, co-guarantors, limits on exposure and conditions that must be met before payments become due. Yet credit markets are unlikely to treat those provisions as secondary details. The enforceability of the agreements, the size of the commitments and the financial health of the underlying projects would shape how CDS spreads respond.
Capacity expected in 2027 and 2028 will test the model
Much of the planned AI data-center capacity is expected to come online in 2027 and 2028, making future cash flow a central question for lenders and credit traders. The projects will need to convert huge capital outlays into recurring revenue from cloud computing, enterprise AI services and model training workloads.
The July 27 jump in Nvidia’s CDS spread suggests that some market participants are beginning to price the possibility that the AI buildout’s financial obligations will not sit solely with the companies operating data centers. Oracle’s higher spread shows that the market has already assigned a larger credit cost to firms with more direct infrastructure responsibilities.
For cryptocurrency markets, the development is relevant less as a direct link to any token and more as a measure of risk appetite across technology finance. A disorderly repricing of highly leveraged AI infrastructure projects could pressure growth-oriented equities and other volatile assets. The more immediate signal will come from the documents behind any financing commitments: whether guarantees are formally signed, how much exposure is capped, and which companies ultimately carry the obligation if projected data-center revenues fail to arrive.
To understand CDS moves and AI risk premia more deeply, explore our explainer on what is a credit spread today.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
