Original Source: Wall Street News
In the past two years, the market assumed that there would always be someone to pay for AI infrastructure bills. This assumption has now been rejected by the credit market.
On July 28, the KOSPI in South Korea fell by 10.84%, marking the eighth trading halt this year. NVIDIA dropped by 5%, SK Hynix fell by 7.47%, and Kioxia in Japan plummeted by 18%.
On the same day, Apple rose by 1.2%, reclaiming its title as the world's most valuable company. The Dow Jones Industrial Average increased by 0.51%, and all three major European stock indexes closed higher.
On that same day, within the same news cycle, the gains and losses were completely reversed. There was no panic in the market—it was voting. The assets being voted out were all companies deeply tied to AI infrastructure, while the assets being voted in were companies that did not depend on the AI infrastructure narrative. The signal triggering this vote was not in the stock market but in the bond market.
Who is Saying No
On Monday, NVIDIA's five-year CDS surged by 14 basis points in a single day to 82 basis points, marking the largest increase in the contract's history. Oracle's CDS rose to 215 basis points, Alphabet's CDS climbed to 67 basis points, and Broadcom, Amazon, and Meta all set new records. SpaceX's newly issued bonds from June fell below par in the secondary market.
The pricing logic of CDS is completely different from stocks. Stock pricing considers growth potential, narrative, and the future. The buyers of CDS are pension funds, insurance companies, and hedge funds; they do not read stories, only cash flows—whether the cash you generate now can cover the principal and interest due.
For the past two years, the pricing power of AI assets has been held entirely in the equity market. The assumption of equity investors is: capital expenditure equals moat; whoever has larger capex wins, every GPU is a down payment on future market value. Under this narrative, the five major tech giants are expected to spend a total of $737 billion on capital expenditures in 2026, up from $261 billion two years ago.
The buyers in the credit market have remained silent because they do not trade in "stories." They are simply waiting.
The direct event that triggered Monday's trading was NVIDIA's announcement on July 25 of a $500 billion AI infrastructure partnership with SK Group, along with a $250 billion financing guarantee from OpenAI and a $350 billion financing arrangement for chip purchases. The potential exposure on the OpenAI side alone exceeds $600 billion, surpassing Thailand's annual GDP. The market spent the weekend digesting the implications of this number—CDS jumped on Monday at the opening.
Prior to this, signals had accumulated over three weeks. On July 9, Oracle was downgraded by S&P from BBB to BBB-, barely above junk status. Alphabet reported negative free cash flow of $5.855 billion last week, the first time in its twenty years as a public company. When these two events occurred separately, the market chose to digest rather than panic—Oracle could be explained as having too concentrated a customer base, and Alphabet could be explained as still in its investment phase, each had its own issues.
NVIDIA's $600 billion revolving financing plan connected the dots. Oracle's largest customer is OpenAI, which is the core beneficiary of NVIDIA's revolving financing plan. Alphabet, Meta, and Amazon all operate under the same “borrow money for infrastructure” financing model. These are not several independent events. This is the same balance sheet, mortgaged repeatedly under different borrowers.
The credit market took three weeks to digest the signals, establish connections, and make judgments. On Monday, the judgment materialized.
How Credit Enters the Closed Loop
The traditional model of the semiconductor industry is unidirectional: NVIDIA sells chips to cloud vendors, cloud vendors lease computing power to AI companies, and funds flow from end users upstream in layers, with each layer assuming its own credit risk.
NVIDIA's current model has turned it into a cycle. NVIDIA provides credit to customers → customers use this credit to purchase chips → chip sales confirm NVIDIA's revenue growth → revenue growth supports its ratings and valuations → a stronger balance sheet supports larger guarantees → the cycle continues.
In this closed loop, part of NVIDIA's revenue growth does not come from end demand but from the credit it injects itself. Sal Naro, Chief Investment Officer of Coherence Credit Strategies, refers to such arrangements as "financial alchemy." John Flood, head of Goldman Sachs' trading department, wrote in a client memo that several traders felt "uncomfortable" and pointed out that the credit market has become a key window for assessing the current AI narrative.
This is not a problem unique to NVIDIA. The entire financing model for AI infrastructure is built on one assumption—capital costs approaching zero, and the market is always willing to pay for the "future." When CDS begins to be priced, this assumption is tested for the first time.
Chain Reaction
The credit market's doubts do not need to verify each company individually; it transmits itself down the financing chain.
Oracle is the first to touch the line. S&P downgraded its rating from BBB to BBB- on July 9, stating three reasons: AI business burn exceeds expectations (FY2026 capex increased by 162% YoY to $56 billion, with free cash flow negative at $23.7 billion, expected to further rise to $95 billion in FY2027); excessive customer concentration (OpenAI accounts for about half of its remaining performance obligations); insufficient internal capacity to absorb excess computing power. Oracle's CDS rose from 144 basis points at the beginning of the year to 215 basis points. A company with annual revenue of $67.4 billion and a whopping 77% increase in cloud infrastructure revenue is seeing its bond insurance costs double.
This risk points directly to OpenAI itself. According to Fortune and the Financial Times, OpenAI is projected to have revenue of $13.07 billion in 2025 while incurring an operating loss of $21 billion and a net loss of about $39 billion after accounting for one-time expenses. A company losing nearly $40 billion a year is Oracle's largest customer and a core beneficiary of NVIDIA's revolving financing plan.
The chain continues to transmit. Alphabet's free cash flow turned negative, and Meta's Texas data center bond at $12.55 billion bore an interest rate of 7.534%—a rate typically only seen in junk bonds, with a subscription multiple of only 1.6 times, less than half of this year's average subscription multiple for U.S. corporate bonds. The average borrowing costs for AA/A-rated companies in the entire tech sector rose from 3.9% a year ago to 5.1%, while BBB-rated companies rose from 4.7% to 7.0%, and the unsecured loan rates for lower-rated AI-related companies have reached 10.5%-12%. Meanwhile, the capital expenditure guidance for the five major tech giants continues to increase.
This is not just a problem for one company. This is the process of the entire financing model switching from "infinite cheap capital" to "market pricing." Every level of widening spread is transmitted to the subsequent levels, the next borrowers, the next markets.
The Bill of 2000
This script is not new.
In March 2000, Cisco became the world's largest company with a market value of $555 billion. Its growth engine was called "vendor financing"—lending to unprofitable internet companies to buy Cisco equipment. At its peak, the exposure was around $4 billion, covering over 700 customers. Chambers called it a "once-in-a-century revolution."
After the Nasdaq peaked, customers began to collapse in large numbers. In May 2001, Cisco wrote off $2.2 billion in inventory and set aside $800 million for bad debt. The stock price fell from $79 to $9.5. By December 2025, Cisco's stock price broke through its 2000 peak, taking 25 years. But due to continuous buybacks that compressed share capital, its current market value is still more than 40% lower than its peak in 2000—while the stock price has returned, the company's scale has not.
From 1996 to 2000, vendor financing was priced as "ecosystem empowerment" and "demand catalyst." After the Nasdaq crash in 2000, as customers began to die off in large numbers, Cisco's bad debts surfaced—those same figures were reinterpreted as "when are you going to pay us back?”
The shift in logic is abrupt: before the rupture, the equity market provided narrative legitimacy for vendor financing; after the rupture, the narrative disappeared, and the bill was reduced to numbers. Jim Bianco, president of Bianco Research, predicts that before the credit market has pricing power, equity valuations can always be debated; once the credit market starts pricing, the debate is over.
Cisco had an exposure of $4 billion. NVIDIA’s potential arrangements related to OpenAI exceed $600 billion. The CDS market and debt buyers reacted much faster this time—Cisco took 14 months from peak to write-off, while NVIDIA’s CDS began to be repriced on the next trading day after the news broke.
Rewarding the "Not Following"
Apple's market capitalization reclaiming the title of the world's largest today is not a coincidence.
Apple's latest quarterly capital expenditure was just $1.97 billion, a decrease of 36% year-over-year. During the same period, Amazon set its full-year capex guidance at $20 billion, while Alphabet raised its guidance to $205 billion. Apple's choice is to return cash to shareholders—announcing an additional $100 billion buyback in the second quarter while also raising dividends. Cook's explanation is that he has "always taken a cautious and thoughtful approach to capital expenditures."
Apple is not avoiding investment in AI. The A19 chip includes a neural accelerator, Apple Intelligence is embedded in its entire lineup, and the new Siri is restructured based on Gemini. The difference lies in the path: edge-side inference with external model access, rejecting the establishment of large-scale training clusters.
From 2024 to 2025, Wall Street constantly criticized Apple for not investing enough, with the mainstream judgment at the time being that self-built computing power equaled a competitive moat. By the summer of 2026, the credit market reversed the same judgment—large-scale capex turned from a competitive advantage into a debt burden, and a restrained balance sheet shifted from conservative to flexible. Chief strategist Jay Woods of the free capital market stated, "Apple was once criticized for insufficient AI investment, but in retrospect, it successfully avoided the capital expenditure trap."
This reversal achieved price confirmation on July 28. The market is not only punishing the over-leveraged but also rewarding those with intact balance sheets. These are two sides of the same choices.
The Eighth Trading Halt of the Year
The credit market's revaluation on Monday reached the accounts of South Korean retail investors in just one day.
KOSPI experienced its eighth trading halt of the year. Samsung and SK Hynix account for 60% of the index's weight, and these two stocks contributed approximately 70% of the KOSPI's increase over the past two years.
In May, regulators approved 16 leveraged ETFs linked to them, inflating from 4.9 trillion won to over $9 billion in two months, with individual investors holding 92%. When climbing, leverage magnifies gains, attracts incremental funds, and pushes up the underlying stocks. When falling, the same mechanism operates in reverse: stocks drop → ETFs are passively reduced → stocks drop further → margin calls → forced liquidation → continues to drop. In the first two weeks of July, 1.2 million leveraged accounts received margin call notices, with 320,000 to 460,000 fully liquidated. Retail investors aged 20 to 30 constituted 62% of the loss group.
This group could not possibly know what had happened before Monday’s opening. They did not know about NVIDIA's $600 billion revolving financing structure, did not know the reason for Oracle's downgrade, and did not know that the credit market was repricing AI assets using a framework completely different from theirs. They only knew that Samsung and SK Hynix were dropping, and leveraged accounts had received margin call notices.
This is precisely the structural feature of the entire transmission chain: information flows from professional buyers in the credit market to the equity market, from the equity market to the global semiconductor supply chain, from the supply chain to the index, from the index to the leveraged ETFs, and ultimately to retail investor accounts. Each layer is delayed, each layer adds its own leverage, and each layer bears a greater impact than the previous layer. On Monday, the CDS jumped 14 basis points, which turned into a trading halt in the KOSPI. Transmitted to leveraged accounts, it became a complete wipeout.
The credit market's repricing is not about one stock, but a whole pricing framework that made "capital always cheap" the default assumption. It begins to recalculate from the roots, with information transmitting and amplifying along the leverage chain, and those who know first take the direction, while those who know last pay the price.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
