The race to build artificial intelligence infrastructure is entering a new phase. After years of heavy spending on computing power, cloud infrastructure and data centres, investors are beginning to pay closer attention to another question: how will this expansion be financed, and at what cost?
A report by Jefferies, GREED & fear: K-shaped stresses, AI tremors and the Houthis, published on October 15, 2026, highlights rising credit market risks around the AI infrastructure buildout. Its focus is not just on the demand for computing capacity, but also on whether lenders will continue to finance the capital expenditure required to sustain the industry’s growth.
The report urges investors to monitor credit spreads for signs that the market may no longer be willing to finance the AI spending cycle. It describes the possibility of funding becoming less available as the credit card being taken away.
Oracle, CoreWeave face rising financing costs
Credit markets are already showing signs of increased concern. Jefferies points to a sharp rise in five year credit default swap (CDS) spreads for Oracle and CoreWeave, two companies closely associated with the expansion of AI infrastructure.
Oracle’s five year CDS spread rose from 186 basis points on September 11 to 288 basis points, while CoreWeave’s increased from 749 basis points to 1,118 basis points over the same period.
According to the report, these were the highest levels recorded since the AI capital expenditure cycle began in early 2023.
The yield on CoreWeave’s bond maturing in 2032 had also reached 14.8%, highlighting the elevated return investors were demanding to hold that debt.
CDS spreads reflect the market’s assessment of credit risk and the cost of protection against a potential default. Rising spreads do not necessarily mean a company is heading towards default, but they can indicate that investors are becoming more concerned about its ability to meet financial obligations.
For companies expanding rapidly to meet demand for AI computing, the cost and availability of capital are therefore becoming important variables alongside chips, power and data centre capacity.
AI infrastructure bill goes beyond chips
Building AI infrastructure requires spending across several layers of the technology stack, including computing hardware, data centres, networking, cooling and power systems. These investments must be funded before companies can generate returns from the capacity they add.
That creates a financial challenge when capital expenditure rises faster than cash flows or when borrowing becomes more expensive.
Jefferies’ analysis raises a broader question for the industry: what happens if the financing environment becomes less supportive while companies are still committing large sums to AI infrastructure?
The report does not predict that AI investment will collapse. Instead, it highlights credit spreads as an indicator investors should watch for evidence that funding conditions are tightening.
This distinction matters. Demand for AI services can continue growing even as investors become more selective about which companies can justify their spending and how that spending is financed.
Earnings growth could become harder to sustain
The financing question comes alongside changing expectations for corporate earnings.
Citing consensus estimates for AI related stocks in the MSCI AC World Index, Jefferies says earnings growth is expected to slow from approximately 100% in 2026 to 30% in 2027 and 22% in 2028.
The comparison with companies outside the AI basket is notable. Earnings growth for the global ex AI group is forecast to moderate from 16% in 2026 to around 9% in both 2027 and 2028.
These are forecasts, not confirmed outcomes. Nevertheless, they suggest investors may need to distinguish between the companies benefiting from AI adoption and those whose valuations already assume exceptionally strong growth.
The valuation gap is also significant. Jefferies puts AI stocks at 23 times forward earnings on an equal weighted basis, compared with 13 times for the ex AI basket.
If earnings growth moderates more sharply than expected, highly valued companies could face pressure even if the underlying AI market continues to expand.
What this means for data centres and infrastructure suppliers
For the data centre industry, the implications extend beyond the largest cloud platforms. Developers, equipment suppliers and infrastructure companies are all exposed to the pace at which new projects are funded and completed.
More expensive financing could affect project economics, investment timelines and the returns expected from new capacity. Companies with strong cash generation, manageable debt and clear demand visibility may be better positioned than those relying heavily on external capital.
However, the Jefferies report does not establish that data centre projects are being cancelled or that demand for AI computing is weakening. Its findings point instead to a financial risk that deserves closer monitoring.
For investors assessing the AI infrastructure supply chain, the important questions are becoming more specific: how much capital is being committed, how will it be financed, when will projects generate cash flows, and can expected returns justify the cost of that capital?
The next phase of the AI race
The AI infrastructure story has largely been defined by the scale of investment required to build computing capacity. The next phase may depend just as much on the economics behind that investment.
Rising credit spreads, high borrowing costs and moderating earnings expectations do not automatically signal the end of the AI investment cycle. They do, however, underline why infrastructure spending needs to be assessed alongside financing costs and the ability to generate sustainable returns.
For an industry racing to build the physical foundations of artificial intelligence, access to capital could become as important as access to chips and electricity.
