Credit-default swap prices for major technology companies surged on July 28 as markets began pricing more risk into the debt of businesses financing the AI infrastructure buildout. Nvidia’s five-year CDS spread widened 14 basis points in one session to 82 basis points, while Oracle’s reached 215 basis points and Alphabet’s rose to 67 basis points, according to credit-market pricing cited in the supplied material.
The move came as equity markets split sharply between expensive AI hardware and data-center plays on one side, and companies with lighter capital-spending requirements on the other. Nvidia fell 5%, SK Hynix declined 7.47%, and Japanese memory-chip maker Kioxia dropped 18%. South Korea’s KOSPI tumbled 10.84% and triggered its eighth trading halt of the year.
Apple moved in the opposite direction, gaining 1.2% and reclaiming the position of the world’s most valuable listed company by market capitalization. The Dow Jones Industrial Average rose 0.51%, while Europe’s three largest equity indexes also closed higher, suggesting the selling was concentrated in companies most exposed to the financial demands of AI computing capacity.
Credit markets focus on AI financing commitments
CDS contracts function as insurance against a borrower defaulting on its debt. A rising CDS spread generally means traders are demanding more compensation to take on that default risk.
The sharp repricing followed disclosures around AI infrastructure funding, including financing and procurement arrangements linked to OpenAI. The package described in the supplied material included a $250 billion financing backstop and $350 billion in chip-purchase financing, creating potential commitments exceeding $600 billion.
Nvidia also announced a $500 billion AI infrastructure cooperation plan with South Korea’s SK Group on July 25. Markets had the weekend to absorb the scale of the agreement before trading resumed, and the reaction was most visible in credit markets at the next opening.
The concern is less about whether AI demand exists than about the timing of cash flows. Chip makers, cloud providers and data-center operators are committing large sums upfront for processors, networking equipment, power systems and facilities, while the revenue from AI products may arrive gradually and remain concentrated among a small number of large customers.
Oracle has become a focal point for that risk. On July 9, S&P Global Ratings downgraded Oracle to BBB- from BBB, leaving the company one step above speculative-grade status. S&P cited an expected 162% year-on-year increase in fiscal 2026 capital expenditure to $56 billion, negative free cash flow of $23.7 billion, and an estimated $95 billion in capital spending for fiscal 2027.
S&P also pointed to customer concentration. OpenAI accounted for about half of Oracle’s remaining performance obligations, according to the ratings agency, leaving Oracle more exposed if a major customer reduces commitments or delays infrastructure deployment.
Oracle’s CDS spread has risen from 144 basis points at the start of the year to 215 basis points. The company reported annual revenue of $67.4 billion, while its cloud infrastructure revenue increased 77%, showing why lenders are weighing rapid growth against the cash required to support it.
Spending plans test balance sheets
Alphabet has also faced scrutiny after reporting negative free cash flow of $5.855 billion, described in the supplied material as its first negative reading in two decades as a public company. Alphabet’s full-year capital-expenditure guidance was cited at $205 billion, while Amazon’s was put at $200 billion.
The five largest technology companies are expected to spend a combined $737 billion on capital expenditures in 2026, compared with $261 billion two years earlier, according to the figures cited in the material. That scale has changed the discussion around AI from one focused largely on product launches and chip supply to one centered on borrowing capacity, customer credit quality and returns on infrastructure.
Borrowing costs have already increased. Average funding costs for AA- and A-rated technology borrowers rose to 5.1% from 3.9% over the past year, while BBB-rated issuers saw average costs climb to 7.0% from 4.7%, according to the supplied figures. Lower-rated AI-linked companies were reported to face unsecured borrowing rates between 10.5% and 12%.
A Texas data-center financing tied to Meta illustrates the tougher terms. A $12.55 billion bond carried a 7.534% coupon, while its order book was reported at 1.6 times the deal size, below the average demand multiple for U.S. corporate bonds this year.
SpaceX bonds issued in June also traded below par in the secondary market, according to the supplied material. Broadcom, Amazon and Meta recorded their highest CDS levels in the same period.
Apple offers a different capital-spending model
Apple’s share-price rise coincided with a markedly different approach to AI spending. The company reported quarterly capital expenditures of $1.97 billion, down 36% from a year earlier, while authorizing another $100 billion in share buybacks and raising its dividend.
Apple has positioned its AI strategy around on-device processing, its A19 chip and neural accelerator, Apple Intelligence features across its operating systems, and external model access that includes a rebuilt Siri based on Google’s Gemini. That approach reduces the need to construct the large training clusters and leased data-center capacity pursued by cloud providers.
The contrast does not settle the competitive question in AI, but it gives Apple more flexibility if credit conditions remain restrictive. Companies building giant server fleets must keep spending even as debt becomes more expensive, while Apple can continue returning cash to shareholders and selectively purchase outside computing capacity.
South Korea’s leveraged market amplified hardware selling
The KOSPI decline also exposed the effect of concentration and leverage in South Korea’s market. Samsung Electronics and SK Hynix represented about 60% of the index’s weighting and accounted for roughly 70% of its gains over the previous two years, according to the figures cited in the material.
South Korean regulators approved 16 two-times leveraged exchange-traded funds tied to the two companies in May. Assets in those products reportedly expanded from 4.9 trillion won to more than $9 billion within two months, with individual traders holding 92% of positions.
During the first two weeks of July, about 1.2 million leveraged accounts reportedly received margin calls, while between 320,000 and 460,000 accounts were fully liquidated. Traders aged 20 to 30 represented 62% of the loss-making group, according to the supplied data.
The July 28 session showed how quickly AI optimism can collide with the financing requirements behind it. Equity markets may continue to reward companies that can deliver AI products with modest incremental spending, while credit markets are placing a higher premium on balance sheets carrying the cost of building the physical infrastructure.
Worried about AI-driven volatility and leverage risks? Explore funding rates in crypto to understand market stress and hidden costs.
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.
