The AI, a cloud computing service provider deeply integrated with NVIDIA, has faced a double blow from both stock and bond markets. On October 14th, its stock price tumbled by 15.46% in a single day; its 2032-year unsecured bond price fell below 80 cents, pushing it into the distressed bond category.

The weakening of individual stocks further dragged down market sentiment, with the AI infrastructure sector in the US stock market experiencing a collective pullback. As of the latest report by the reporter, CoreWeave has fallen nearly 6% so far today.

The panicked selling in the market this time is due to the chain reaction caused by inflation data in the United States exceeding expectations. For companies like CoreWeave that rely on continuous borrowing and have not yet made a profit, rising interest rates will directly increase the cost of interest on new debt.

Currently, with rising expectations of interest rate hikes and increasing yields on long-term U.S. Treasury bonds, there are concerns in the market about whether this computing power company, which relies on continuous debt financing for expansion and has not yet achieved profitability, will face greater difficulties in future refinancing.

CoreWeave Both stocks and bonds suffer losses, with sectors collectively under pressure

During the US stock trading session on Wednesday, CoreWeave ( CRWV ) saw simultaneous selling of its shares and bonds. The stock tumbled significantly during the session, closing down by 15.46% at $65.78; its unsecured bonds maturing in 2032 also weakened, with prices falling below par at 80 cents, and their yield breaking through 15%, entering the distressed bond range.

Looking at the long term, CoreWeave's stock price has been fluctuating downward this year. As of the close on October 14th, the cumulative decline for the year 2026 is about 8.14%. Entering October, the volatility of the stock price has significantly increased. At the beginning of the month, it was still above $88, but after consecutive corrections on the 7th and 8th, it tumbled sharply on the 14th to close at $65.78, recording the largest single-day decline since February of this year.

The sharp decline in individual stocks also led to a collective correction in the AI infrastructure sector of the US stock market. As of the close of trading on Wednesday, Nebius, a provider of computing power services in the same sector, fell by about 10%, upstream server manufacturer Dell tumbled 11.27%, and cloud service provider Oracle dropped 8.83%, showing a clear characteristic of widespread declines in the sector.

CoreWeave is a cloud computing service provider specializing in AI liquid-cooled data center operations. They purchase high-end GPU from NVIDIA and lease the computing power to external parties, serving large-scale model training and inference. They position themselves at the supply end of computing power in the AI industry chain.

The rise of this business model is due to the real pain points of the AI industry in building its own computing power. AI companies that wish to establish large-scale computing power clusters need to invest a substantial amount of capital in purchasing GPU in the early stages. The construction period for data centers is lengthy, and they also have to face the risks of rapid hardware iteration and depreciation of idle equipment.

Industry analysis indicates that the GPU computing power leasing model is rapidly gaining popularity. Enterprises can convert one-time large capital expenditures into flexible operating costs, thereby avoiding the risks of hardware iteration and idle resources. This also opens up market opportunities for service providers such as CoreWeave.

IDC released Worldwide Quarterly AI Infrastructure Tracker in July 2026, and it is estimated that global AI infrastructure spending will reach $497 billion in 2026.

Currently, CoreWeave serves leading AI research institutions such as OpenAI, Anthropic, and Meta, as well as a large number of small and medium-sized startups. The core of its stable business development comes from strategic partnerships deeply integrated with NVIDIA.

It is reported that NVIDIA is not only its core GPU supplier but also completed a strategic capital increase of $2 billion in January 2026. This funds are specifically allocated for the expansion of computing power infrastructure, with the goal of building a total computing power of over 5 gigawatts by 2030. The two parties signed a $6.3 billion computing power guarantee agreement as early as September 2025. If the newly added computing power from CoreWeave cannot be fully leased, NVIDIA will take over the idle capacity to support their production expansion. Technically, both parties continue to collaborate on research and development, and CoreWeave can prioritize the implementation of NVIDIA's latest generation of AI infrastructure solutions.

High-leverage computing power business encounters rising interest rates

The expansion of CoreWeave is highly dependent on external financing and is naturally very sensitive to changes in interest rates. This fluctuation is not an isolated decline of a single stock, but rather a concentrated test of the market for this type of heavy-asset computing power leasing business model.

Macroeconomically, on October 14th, the U.S. Bureau of Labor Statistics released the data for September CPI. Overall, CPI increased by 3.7% year-on-year, exceeding market expectations of 3.6%. Traders increased their bets on the Federal Reserve raising interest rates within the year, which directly pushed up the yields of long-term U.S. Treasury bonds.

For companies like CoreWeave that rely on continuous borrowing and have not yet made a profit, rising interest rates will directly increase the cost of interest on new debt. The market is concerned that as interest expenses continue to accumulate due to ongoing expansion, the subsequent financing opportunities for these companies will also become tighter.

S&P Global Ratings has assigned a B+ credit rating to CoreWeave, classifying it as high-yield debt, which is commonly referred to in the market as junk debt. Gavan Nolan, Executive Director of Fixed Income Pricing at S&P Global Market Intelligence, analyzed that it is not surprising for companies like CoreWeave with high leverage to face pressure during periods of rising interest rates. In his view, although this company's capital investment scale is on par with that of large public cloud providers, its balance sheet strength is far inferior to that of industry giants. This mismatch makes it particularly sensitive to interest rate fluctuations.

Faced with a significant weakening of debt prices and widespread concerns about debt repayment capabilities in the market, CoreWeave is also taking self-help measures at multiple levels, including communication and financing.

First, management stepped in to stabilize market expectations. After the sharp drop in stock prices, CEO Michael Intrator issued an internal memo to all employees, refuting the pessimistic market pricing and emphasizing that the scale of signed orders in hand was sufficient to cover all debt repayments. The current fluctuations were attributed to irrational market reactions and the spread of negative sentiment among peers.

On the financing side, CoreWeave is the most crucial element for self-rescue. The company is actively exploring insurance funds as a long-term source of capital and promoting a 'delayed withdrawal fixed-term loan' model ( DDTL ) that uses GPU computing power assets as collateral. They hope to obtain funds with longer terms and relatively controllable costs from insurance institutions, in order to replace the current high-interest, unsecured bonds and optimize the overall debt structure. Prior to this, in September, the company already launched a combined financing plan: it intends to issue convertible preferred notes with a principal amount of $3 billion, accompanied by a $500 million over-subscription option. If fully exercised, the financing scale could reach up to $3.5 billion, and at the same time, a public offering at market price ( ATM ) will be conducted, with a maximum of 35 million ordinary shares to be sold. The aim is to supplement cash flow and replace high-cost liabilities through a combination of equity and debt. Earlier on, the company also secured several special loans based on the collateral of GPU assets, in an attempt to reduce the overall weighted financing cost.

In fact, this contradiction between capital expenditure and cash flow is not unique to CoreWeave. A cross-team analysis report released by Goldman Sachs in September showed that even the largest cloud giants with substantial funds face the problem of operating cash flows being unable to cover capital expenditures, despite their massive investments in AI computing power.

Goldman Sachs' credit strategy team estimates that by 2027, at least 35% of the capital expenditures of cloud giants will need to be raised through bond issuance, which means the market will see an oversupply of technology company bonds amounting to over $400 billion. Equity market funds are also converging on this AI sector. This year, US companies AI raised approximately $65 billion through related additional share offerings, accounting for 45% of the total additional financing in the US stock market. In an environment of rising interest rates, both leading cloud providers and third-party computing power service providers like CoreWeave will face more stringent tests with their business models that rely on continuous external financing for expansion.