AI infrastructure debt crisis is spreading: After Oracle's 19% plunge, who's next?

In the last week of June 2026, Oracle investors experienced a rout not seen in 25 years.

At the close on June 26, Oracle was at $148.53, down another 2.58% on the day, marking its fifth consecutive losing session. The weekly decline totaled 19.4%, the worst single-week performance since the 23.30% plunge in the week ending March 2, 2001. Compared to its all-time high near $900 billion in market cap in September 2025, the stock has retreated about 55%.

This is not an ordinary earnings miss. The market is systematically repricing the business model of AI infrastructure—"burn cash first, deliver later"—that has dominated the past two years. Oracle's collapse is just the first domino to fall in the entire AI infrastructure debt chain.

$130 Billion in Debt and $56 Billion in Capex: The Pressure on Oracle's Balance Sheet

Oracle's financial data reveals the root of the market panic.

As of the end of May 2026, Oracle's total debt stood at approximately $130 billion. Capital expenditure (Capex) for fiscal 2026 reached $55.66 billion, up 162% from the prior fiscal year. However, free cash flow was negative $23.7 billion—despite operating cash flow hitting a record high of $32 billion, nearly all of it was consumed by data center construction.

To bridge the funding gap, Oracle raised $43 billion through debt financing and $5 billion through equity financing in fiscal 2026. Management has signaled that another $40 billion will need to be raised in fiscal 2027, including a previously announced $20 billion stock offering.

Credit default swap (CDS) costs have surged to an all-time high, as bond investors demand greater risk compensation. This is a clear signal that the credit market's doubts about Oracle's debt sustainability are being priced in hard currency.

Meanwhile, Oracle laid off 13% of its workforce in fiscal 2026, reducing its global headcount to 141,000. The sales and marketing departments were the main targets of the cuts. A company still racing to ramp up capex while simultaneously conducting mass layoffs sends a jarring signal to the market: the cash burn rate of AI infrastructure has exceeded what organic growth can sustain.

The "Other Side" of Revenue Growth: The Timing Challenge of Recognizing $638 Billion in Remaining Performance Obligations

Oracle's defenders will point to a key metric: Remaining Performance Obligations (RPO) have reached $638 billion, up 363% year-over-year. Cloud infrastructure revenue (IaaS) grew 93% year-over-year, and total cloud revenue reached $9.9 billion, up 47%. Fourth-quarter total revenue was $19.2 billion, up 21%.

But these numbers are precisely the heart of the problem.

Of the $638 billion in RPO, about $75 billion comes from customer prepayments or customer-supplied AI hardware. This "revenue" reduces Oracle's short-term financing pressure, but it does not change the fact that RPO is not recognized revenue, let alone free cash flow.

The market is beginning to ask a previously overlooked question: How long is the revenue recognition cycle for these contracts? If the growth rate of AI computing demand slows, or if customers delay deployment, can these RPOs be converted into cash flow as expected? Reports that OpenAI's IPO has been delayed to 2027 further extend the payback period for Oracle's massive capex.

Evercore analysts wrote in a report: "We expect financing leverage and the pace of equity issuance to remain central to investor debate in the near term, despite strong demand signals."

CoreWeave: $15 Billion in Debt and $31–35 Billion in Capex – A Bet on Computing Power

If Oracle's debt problem raises concerns about "too big to fail," CoreWeave's situation is more like a typical high-leverage growth stock risk sample.

The AI cloud service provider has a 2026 capex guidance of $31 billion to $35 billion. As of the end of the first quarter, its contracted revenue RPO was $99.4 billion, with about 36% expected to be recognized over the next two years and 75% over the next four years.

But CoreWeave's debt burden is equally daunting. Including lease obligations, CoreWeave carries about $15 billion in debt—nearly four times its total revenue over the past twelve months. So far in 2026, the company has raised over $20 billion in debt and equity capital. In mid-June, CoreWeave announced plans to issue $3.5 billion in senior notes.

CoreWeave's credit rating is speculative grade. Its revenue is highly concentrated in OpenAI-related computing loads, with most routed through Microsoft. This is a fragile value chain: if OpenAI's computing demand falls short or Microsoft adjusts its partnership, CoreWeave's revenue base will be directly impacted.

Global banks have begun to feel "overwhelmed" by data center-related loans. Major lenders such as JPMorgan, Morgan Stanley, and Sumitomo Mitsui Banking Corporation are actively seeking to distribute these risks to a broader investor base. A $38 billion construction debt related to Oracle's data center projects in Texas and Wisconsin has taken banks more than six months to syndicate, with some institutions even seeking to sell at a discount.

The banks' caution is not unfounded. AI data centers have long construction periods, high capital intensity, and strong asset specificity—if these data centers cannot be leased by AI customers, they have few alternative uses. When the world's largest banks start saying "no" to a certain asset class, it often means systemic risk has accumulated to a critical point.

SoftBank: $40 Billion in Bridge Debt and the Dilemma of Pledging OpenAI Equity

The third link in the triangular debt chain is SoftBank.

To support its investment in OpenAI, SoftBank used $40 billion in bridge financing, which must be repaid by March 2027. Meanwhile, SoftBank has tried to secure at least $6 billion in margin loans backed by OpenAI equity—an effort that has stalled.

The root of the problem: OpenAI is still a non-public asset, making it impossible for lenders to perform real-time valuations of its equity. SoftBank initially sought a $10 billion loan but cut it by 40% to $6 billion due to waning lender enthusiasm, yet still failed to close the deal. After the news, SoftBank's stock plunged more than 8% in a single day.

S&P Global Ratings has revised its credit outlook on SoftBank to negative, citing its massive investment in OpenAI.

SoftBank faces a classic "maturity mismatch" dilemma: the realization cycle of its long-term asset (OpenAI equity) is uncertain, while the repayment pressure of its short-term debt (the $40 billion bridge loan due in March 2027) is counting down. If OpenAI's IPO cannot be completed before March 2027, SoftBank will have to sell OpenAI equity in the secondary market or seek other refinancing options—both of which are fraught with uncertainty in the current market environment.

The Systemic Spread of the "AI Debt Crisis": From Individual Stocks to the Entire Credit Market

Oracle's collapse is not an isolated event. It occurs against a broader backdrop.

In its annual report released on June 29, 2026, the Bank for International Settlements (BIS) listed the bursting of the AI bubble as one of the three "risk flashpoints" threatening global economic prosperity. BIS warned that if returns in the AI sector fall short of expectations, it could trigger a sudden tightening of financing, rapidly turning the current capex boom into a prolonged investment slump.

Signals from the credit market are equally noteworthy. In June, the risk premium on US high-grade tech bonds widened to 0.79 percentage points, up significantly from 0.74 percentage points at the end of May. Tech giants are raising capital at the most intensive pace since the dot-com bubble.

JPMorgan Chase forecasts that total spending related to AI and data centers could reach $5.5 trillion by 2030, with data centers expected to raise about $2.1 trillion in the investment-grade bond market over the next five years. The question is: who will pay for all this debt? As bond investors demand higher spreads and banks curtail their exposure to data centers due to internal risk limits, the financing cost of AI infrastructure is rising systemically.

Mapping to Crypto Markets: A Leading Indicator of Risk Appetite Contraction

The brewing risks in AI infrastructure debt are transmitting to crypto markets through multiple channels. On June 29, 2026, Bitcoin fell below the $60,000 mark to around $59,356. Ethereum weakened in tandem, trading at about $1,553.

Both are in a deep correction phase, with daily and 4-hour MACD indicators in a bearish cross below the zero line for an extended period. The green momentum bars continue to expand without showing a bottom divergence structure, suggesting that the downward momentum has not yet been exhausted.

The transmission logic is not complicated. A large portion of AI infrastructure companies' debt is provided by banks and credit markets, and the major institutional investors in crypto markets overlap heavily with those in traditional financial markets. When AI debt risks trigger expectations of systemic credit contraction, the pricing benchmark for risk assets shifts upward as a whole—crypto assets, as high-beta assets, often feel the pressure of liquidity tightening first.

Solana briefly fell below $70, while Dogecoin, BNB, and others also declined. This is not an isolated technical adjustment but a mapping of the systemic contraction of global risk appetite in crypto markets.

Who Is Next?

Oracle will not be the last company to be punished by the market for AI infrastructure debt pressure.

Following the same logic chain: those AI infrastructure companies where capex growth far exceeds free cash flow generation and which rely heavily on debt markets for financing could become targets of the next wave of selling. CoreWeave, with its speculative-grade credit rating and $15 billion in debt, is at the forefront. Nebius has raised its 2026 capex guidance to $20–25 billion. Even hyperscale cloud providers with stronger balance sheets are under increasing scrutiny—Microsoft's 2026 capex guidance is around $190 billion, and Alphabet raised its 2026 capex outlook to $180–190 billion.

The difference lies in the financing structure. Oracle and CoreWeave's AI infrastructure expansion relies more on external debt financing, while Microsoft, Amazon, and Google have stronger free cash flow and lower financing costs. But even for the latter, when capex reaches the hundreds of billions, internal cash flow is hard to cover all expenses.

JPMorgan's assessment provides a macro framework: Equity is not replacing debt but supplementing it. Behind each round of equity financing, there is often an even larger scale of debt financing. This means that the debt leverage of AI infrastructure will not decline due to tech giants' equity sales but may continue to rise.

Conclusion

Oracle's 19% single-week decline is the market's first systematic repricing of the "burn cash first, deliver later" AI infrastructure financing model. $130 billion in debt, $56 billion in annual capex, and $23.7 billion in negative free cash flow—these numbers together paint a clear picture of balance sheet pressure.

And CoreWeave's $15 billion in debt with $31–35 billion in capex, along with SoftBank's $40 billion bridge debt and the dilemma of pledging OpenAI equity, reveal the depth and complexity of this debt chain. When global banks start saying "no" to data center loans, when the BIS lists the AI bubble as a systemic risk, and when high-grade tech bond spreads continue to widen—these signals all point in the same direction: the "burn cash and stake a claim" model of AI infrastructure is entering a period of debt servicing verification.

For the crypto market, what does the spread of this AI debt crisis mean? A systemic contraction of risk appetite is underway. Bitcoin falling below $60,000 and Ethereum losing the $1,600 level are not just technical adjustments; they are the mapping of tightening global liquidity expectations in crypto asset pricing.

Oracle's 19% collapse is just the beginning. The next question is: when the credit market tightens, when financing costs continue to rise, and when the recognition cycle of remaining performance obligations is extended—who will be the next domino to fall?

FAQ

Q1: Why did Oracle's stock plummet 19% last week?

The market is worried about Oracle's staggering $130 billion debt burden and the return prospects of its AI infrastructure investments. Capital expenditure for fiscal 2026 reached $55.66 billion, free cash flow was negative $23.7 billion, and the company plans to raise another $40 billion in fiscal 2027. Investors are questioning the sustainability of its AI infrastructure expansion.

Q2: Where does Oracle's debt level stand relative to the industry?

Oracle has about $130 billion in debt, placing it at a high leverage level within the tech industry. Compared to hyperscale cloud providers like Microsoft and Amazon, Oracle lacks the same level of robust free cash flow to support capital expenditures. Its credit default swap (CDS) costs have hit an all-time high, reflecting growing credit market concerns.

Q3: How significant is CoreWeave's debt risk?

Including lease obligations, CoreWeave carries about $15 billion in debt, nearly four times its total revenue over the past twelve months. The company's 2026 capex guidance is $31–35 billion, and its credit rating is speculative grade. Its revenue is highly concentrated on OpenAI, with most routed through Microsoft, creating significant customer concentration risk.

Q4: What is the $40 billion bridge debt at SoftBank?

To support its investment in OpenAI, SoftBank used $40 billion in bridge financing, which must be repaid by March 2027. The company tried to secure a $6 billion margin loan backed by OpenAI equity, but the effort stalled because OpenAI is a non-public asset, making valuation difficult for lenders. S&P has revised SoftBank's credit outlook to negative.

Q5: How could the AI debt crisis affect the crypto market?

A large portion of AI infrastructure companies' debt is provided by banks and credit markets, overlapping with crypto institutional investors. When AI debt risks trigger expectations of credit contraction, the pricing benchmark for risk assets shifts upward. On June 29, 2026, Bitcoin fell below $60,000 and Ethereum below $1,600, reflecting the systemic contraction of global risk appetite in crypto markets.