Original Author: Xu Chao

Original Source: Wall Street Watch

CoreWeave's yield on unsecured bonds has soared to about 15%, with CDS spreads breaking 1000 basis points, raising serious doubts in the market about the systemic risks of the entire AI capital expenditure cycle.

However, top firms on Wall Street are collectively voicing: this is not a 2007 New Century Financial-style systemic collapse, but more like the transient panic triggered by Silicon Valley Bank (SVB) in 2023—a small, structurally fragile tool being used by the market as a barometer for a trillion-dollar AI capital expenditure cycle.

Ajay Rajadhyaksha, the head of global macro research at Barclays, pointed out that CoreWeave's approximately $1.5 billion unsecured bonds are structurally junior within its total debt of approximately $35 billion. The price signals from this tool contradict the entire AI cycle, similar to how a $4 billion regional bank ETF in 2023 was used to predict a recession in the U.S. Goldman Sachs believes the market has chosen the wrong benchmark— the unsecured bond market does not represent CoreWeave's true new financing costs.

Historical Precedents: SVB or the 2007 Financial Crisis?

The core of the debate is the choice of historical analogy.

In April 2007, the largest subprime mortgage lender in the U.S., New Century Financial, filed for bankruptcy, leading to a progressive collapse of the entire credit market, with the S&P 500 index dropping about 40% in the following two years. In March 2023, Silicon Valley Bank was shut down by regulators, and the market quickly formed a narrative: regional banks are the lifeline for small and medium-sized enterprises, and a credit drought would trigger a recession. The bond market quickly priced in a rate cut of 75 to 100 basis points that year, with the VIX panic index soaring even higher than in 2007.

As a result, the Federal Reserve did not cut rates at all in 2023; instead, it continued to raise rates, and the U.S. economy grew much faster than expected throughout the year, and a recession never materialized. The regional bank ETF KRE regained all its lost ground, and the S&P 500 rose about 40% in the subsequent two years.

Bloomberg columnist John Authers pointed out that both crises appeared severe in real-time, but ultimately diverged significantly. SVB was an overextended panic, not a precursor to systemic risk; the bankruptcy of New Century Financial revealed a structural flaw that ran through the entire financial system. CoreWeave's situation more closely resembles that of SVB in form.

Rajadhyaksha summarized the core error in the SVB panic: "The mistake wasn't recognizing that Silicon Valley Bank had failed. It indeed failed. The mistake was extrapolating from a small, highly volatile, and structurally weak tool to the entire U.S. economy."

Barclays: A $1.5 Billion "Tail" Wagging the Whole "Dog"

Rajadhyaksha directly compared the logical structure of current market panic to that during the SVB period.

The "tail" of 2023 was a $4 billion equal-weight regional bank ETF—KRE, with many constituent stocks trading lightly. The decline of this small ETF triggered the repricing of trillions of dollars of interest rates, investment-grade credit, and stock positions, forming a self-reinforcing panic cycle.

The "tail" of 2026 is CoreWeave's approximately $1.5 billion structural junior unsecured bond.

Rajadhyaksha outlined CoreWeave's three-tier debt structure: the main component is secured delayed draw term loans (DDTL), held in independent special purpose vehicles (SPVs), with each SPV linked to specific GPU hardware, customer contracts, and data center leases—when Meta signs a multi-year "pay whether used or not" contract and transfers it to the SPV, the actual collateral for the loan is Meta's payment commitment; next are approximately $9 billion in convertible bonds, with coupon rates between 1.75% and 2.875%; and at the bottom are unsecured notes, with hundreds of billions of dollars in senior claims ahead in the liquidation queue.

"A 15% yield tells you something about this paper; it cannot tell you that this company is about to cease operations," Rajadhyaksha wrote, using an analogy: "If a person has a $35 million mortgage along with a small credit card, and the credit card has a high interest rate, that indicates an issue with the credit card, not that all mortgage borrowers in the city are in trouble."

The numerical comparison is also clear: in March 2023, the market priced in a 100 basis point rate cut due to the decline of a $4 billion ETF; in 2026, the market began to question a trillion-dollar capital expenditure cycle due to trade in approximately $1.5 billion of structural junior bonds at 80 cents.

Goldman Sachs: The Market Chose the Wrong Benchmark for Financing Costs

Goldman Sachs analysts Gabriela Borges and Maura Hager's core argument also focuses on the misreading of debt structure.

Goldman Sachs pointed out that CoreWeave's primary financing tool is DDTL—asset-backed project financing secured by specific customer contracts, GPU infrastructure, and related cash flows. The lender guarantees the cash flow quality of the specific contracts, not CoreWeave's overall credit status; this structure transfers financing risk from the company's balance sheet to the underlying assets, thus reducing refinancing risk. The unsecured bond market for CoreWeave's new issuance has been closed for months; measuring financing costs based on the yield in that market is akin to referencing a price signal that has become obsolete.

Based on annualized interest expenses of $2.56 billion for the second quarter divided by approximately $30 billion in total debt, Goldman Sachs estimates CoreWeave's current weighted average debt cost to be around 800 to 900 basis points. For new financing, Goldman Sachs assumes that investment-grade customers (such as Meta) will be priced for DDTL at around SOFR + 300 basis points, while high-yield customers will be at about SOFR + 600 basis points, providing a blended cost range of 750 to 1050 basis points, taking a conservative benchmark of 1000 basis points, suggesting that new financing costs have only risen by 100 to 200 basis points compared to current levels.

Goldman Sachs also cited market data: DDTL 5.0 and 5.5 are trading at prices not lower than the original issue discount, indicating that the asset-backed financing market still has not lost faith in CoreWeave.

It is worth noting that Goldman Sachs disclosed in the report appendix that it is acting as the lead or co-manager for a pending underwriting business for CoreWeave and holds over a 1% stake in CoreWeave, having received investment banking fees from CoreWeave over the past 12 months, and expects to continue seeking related business in the next three months.

Goldman Sachs' Bull Market Logic: Pivot and Vulnerability - 30% Increase in Revenue per Megawatt Needed

The other half of Goldman Sachs' bull market argument is based on rising expectations for CoreWeave's computing pricing ability.

Goldman Sachs estimates that CoreWeave's current mixed monetization level is about $10 million in annual recurring revenue (ARR) per megawatt. Supporting factors include: prices for short-term contracts (3 to 6 months) have reached $40 million per megawatt; CoreWeave raised prices by 25% in July this year and again by 10% in October; contracts signed in the second quarter contribute profit margins 5 to 10 percentage points higher than recent contracts. Goldman Sachs considers $13 million per megawatt as a "reasonable starting assumption" for new long-term contracts.

Using a sample model (5-year, 200-megawatt contract, 80% debt financing, 10% financing cost, residual value of zero), the result is a 15% unleveraged internal rate of return, forming a positive spread of about 3 percentage points over the 11.8% weighted average cost of capital.

However, Goldman Sachs' sensitivity analysis also reveals the fragility of this logic: if the current actual mixed level of $10 million/megawatt cannot rise to $13 million/megawatt, the entire positive spread logic will fully reverse. Once supply catches up with demand and pricing pulls back, CoreWeave remains locked into 7 to 16-year data center leases, making exit costly.

Brian Stutland of Equity Armor Investments believes that the core of CoreWeave's problems lies in its own business model rather than the overall outlook for AI, but he also admits this is a "two sides of the same coin" situation—whether it is an isolated company-level issue or revealing a broader structural vulnerability is currently a "50-50 probability."

For investors, the key question now is not whether CoreWeave will fail, but whether computing pricing can support the financing logic of this cycle. The answer to this question will determine the ultimate outcome of this panic.

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[email protected]，本平台相关工作人员将会进行核查。