Rising borrowing costs signal funding risks for global AI expansion: Jefferies report

A credit spread is the extra yield investors demand to lend to a company instead of a lower-risk borrower. A wider spread means lenders see more danger and want greater compensation. It therefore acts as a market-based warning about corporate credit risk.

A credit default swap, or CDS, is a contract linked to a company's debt. The buyer pays for protection if the company defaults. The CDS spread is the cost of that protection. If the spread rises, investors are paying more to insure against failure, suggesting weaker confidence in the borrower.

The article reports that Oracle's five-year CDS spread rose from 186 to 288 basis points. CoreWeave's jumped from 749 to 1,118 basis points. These were the highest levels since the AI capital-expenditure cycle began in early 2023. They do not prove funding has stopped, but they show that markets see greater risk in lending to these companies.

The reported increases were substantial. Oracle's five-year CDS spread rose 102 basis points, from 186 on September 11 to 288. CoreWeave's rose 369 basis points, from 749 to 1,118. Both reached their highest reported levels since the AI capital-expenditure cycle began in early 2023.

The figures matter because spreads reflect how much investors want for accepting credit risk. CoreWeave also had a 2032 bond yield of 14.8%. A bond yield is the return investors require to hold that debt. For the company, issuing comparable debt would involve a very high financing cost, assuming market conditions were similar.

That pressure can affect expansion plans. More interest expense leaves less money for data centres, computing equipment, and other projects. The report did not say financing had ended. It presented the rising spreads and high yield as early warnings that markets may become more selective about funding AI investment.

AI infrastructure requires major spending on data centres, computing capacity, and related technology. Companies often need financing before those investments produce earnings. When credit spreads rise, lenders see greater risk and demand higher returns. That makes the money needed for expansion more expensive.

The mechanism is direct. A company borrowing at a higher rate pays more interest on the same debt. More cash then goes toward servicing loans rather than building facilities or buying equipment. Projects that looked profitable at lower borrowing costs may become less attractive or require stronger future earnings.

The report highlights Oracle and CoreWeave as examples of this pressure. Their credit-risk measures reached the highest levels reported since early 2023, and CoreWeave's 2032 bond yield reached 14.8 per cent. Funding has not stopped, but rising costs could make continued AI investment harder, especially for companies reliant on external capital.

Capital expenditure, or capex, is spending on long-lasting assets and infrastructure. In this case, the AI capex boom refers to companies investing heavily to expand AI systems. The report treats this spending cycle as having begun in early 2023.

The article identifies data centres, computing capacity, and related technology as major requirements. These systems need substantial physical infrastructure before companies can deliver expanding AI services. That creates large upfront costs, often requiring companies to raise money while their expected returns remain ahead.

The scale makes financing conditions important. Jefferies urged investors to watch credit spreads for signs that markets may no longer fund the ongoing boom as readily. Oracle and CoreWeave showed sharply higher credit-risk measures. The article does not say investment has stopped. It says higher financing risks could pressure companies to justify continued spending.

Investors are watching Oracle and CoreWeave because the article reports unusually high increases in their borrowing-risk indicators. These changes offer a visible test of whether financial markets remain comfortable funding AI infrastructure. Rising indicators do not mean the companies have lost access to money, but they show that financing risk has increased.

Oracle's five-year CDS spread rose from 186 to 288 basis points. CoreWeave's rose from 749 to 1,118 basis points. CoreWeave also had a 14.8 per cent yield on a bond maturing in 2032. Higher spreads and yields mean investors demand more compensation for lending to, or insuring debt from, these companies.

The companies therefore illustrate the article's central concern. AI expansion needs continued funding for data centres and computing capacity. If markets demand increasingly high returns, companies may face costlier financing and greater pressure to prove that spending will produce sufficient earnings.

The article does not explain how Oracle, CoreWeave, or other companies would replace borrowing if debt became more expensive. It only says that higher financing risks could make AI expansion harder, especially for companies dependent on external funding. The options below are established corporate-finance alternatives, not strategies specified in the report.

Companies can use internally generated cash, raise equity by selling shares, or seek investment from strategic partners. They could also sell assets, form joint ventures, lease infrastructure, or negotiate supplier financing. These approaches can reduce immediate borrowing, but they may involve dilution, shared control, asset sales, or higher operating commitments.

Another option is to slow, stage, or cancel projects. That lowers near-term funding needs but may delay capacity and expected growth. The report's warning remains relevant under every route: markets are demanding closer scrutiny of AI spending, while earnings growth for global AI stocks is expected to moderate after 2026.

Interest rates and credit spreads determine how expensive it is to finance investment. Higher rates or wider spreads increase interest costs and make projects harder to justify. Investors then compare those costs with expected company earnings. If expected earnings weaken, lenders and shareholders may become less willing to support aggressive expansion.

Stock valuations add another layer. The article says global AI stocks trade at 23 times forward earnings, versus 13 times for non-AI stocks. That premium assumes stronger future performance. Yet consensus earnings growth for global AI stocks is expected to slow from around 100 per cent in 2026 to 30 per cent in 2027 and 22 per cent in 2028.

This combination can pressure the boom. Higher financing costs reduce returns, while slower earnings growth makes expensive shares harder to defend. The article does not predict an end to AI investment. It says rising credit risk and earnings expectations should make investors scrutinize spending more closely.