The Zhitong Finance App learned that the stock price of AI cloud computing service provider Coreweave (CRWV.US) plummeted nearly 15% on Wednesday, and investors' concerns about its high-debt expansion model continued to heat up. Bernstein pointed out in a research report released on October 15 that what the market is worried about is not insufficient demand for AI computing power, but rather the financing costs of CoreWeave's continuous expansion to meet demand. As the yield on some of the company's unsecured bonds approaches 15%, its business model, which relies on external financing to drive growth, is facing an increasingly severe test.
Bernstein maintained CoreWeave's “outperforming the market” rating, with a target price of $74. The stock closed at $65.78 on Wednesday, and although the target price is about 12% higher than the current share price, Bernstein stressed that its long-term concerns mainly focus on the economic sustainability of the company's business model and future competitive position in the AI infrastructure market.
Bond yields are approaching 15%, and financing costs have become the focus of the market
Bernstein pointed out that CoreWeave is an unsecured bond that matures in 2032 and has a coupon interest rate of 9.625%, and the yield to maturity has now risen to about 14.8%. This level is usually linked to debt pricing for financially troubled firms, indicating that bond investors are demanding higher risk compensation.
Although the increase in yield in the bond market will not immediately change the interest expenses on the company's existing fixed-rate debt, it will directly affect the cost of additional future financing. This change is particularly important for a company that relies heavily on debt financing to build an AI data center.
CoreWeave's business model requires a huge investment in advance to purchase GPUs, build data centers, and related infrastructure before customers pay. Since the company lacks other businesses that can continuously generate large amounts of cash flow, rapid growth will further widen the financing gap.
According to data, in the second quarter of this year, CoreWeave's operating activities generated only US$679 million in cash flow, while adjusted capital expenditure reached US$9.4 billion. There is a gap of over US$8.7 billion between the two. Meanwhile, the company's net interest expenses for the second quarter reached US$640 million, accounting for about a quarter of the quarter's revenue.
Bernstein believes this means that even if CoreWeave can continue to receive customer orders, it must continue to rely on external funding to support business expansion, and rising financing costs may significantly reduce the profitability of future projects.
The size of debt continues to expand, and after 2030, there will be pressure to concentrate debt repayment
Another challenge facing CoreWeave is the growing debt burden. As of the end of June this year, the company's debt was about 35 billion US dollars, in addition to about 16.5 billion US dollars in leasing liabilities, which together exceeded 50 billion US dollars.
The latest debt details attached to the Bernstein Research Report further show that the total amount of outstanding debts listed by CoreWeave has reached about US$40.95 billion, the corresponding financing amount is about US$57.21 billion, the weighted effective interest rate is about 8%, and the relevant annual interest expenses are estimated at about US$3.26 billion.
Among them, some debt financing costs have reached double digits. For example, the effective interest rate for a deferred withdrawal term loan due in 2028 is about 15%, and the effective interest rate for various equipment financing and term loans is between 9% and 12%.
Judging from the debt maturity structure, CoreWeave's short-term debt repayment pressure is relatively limited, but 2030 to 2032 will usher in a clear peak of concentrated debt maturity.
According to Bernstein's data, the company's debts of about US$531 million, US$1,276 million, US$2,576 million and US$2,665 billion were due from 2026 to 2029, respectively. The size of debt maturing in 2030, 2031, and 2032 reached approximately US$11.149 billion, US$9.389 billion, and US$10.366 billion, respectively, for a total of more than US$30 billion over three years. This means that CoreWeave not only needs to continuously raise capital to support the construction of new projects, but must also consider the cost and viability of large-scale debt refinancing in the future.
Bernstein pointed out that while increasing the scale of financing will certainly drive revenue growth, it also means that more creditors will have preferential payment rights over shareholders, which may further weaken the economic returns that ordinary shareholders can eventually obtain.
Changes in customer contract structures or increased financing pressure
In addition to the cost of financing and the size of debt, Bernstein also paid special attention to the quality of CoreWeave's future customer contracts. The bank pointed out that if the company's new business gradually shifts to contracts with shorter terms, or more from non-investment-grade credit rating customers, the contract guarantees obtained by lenders will decline accordingly.
Under such circumstances, even if Coreweave can continue to sign new AI computing power service contracts, creditors may still require higher interest rates or stricter financing conditions. In other words, an increase in orders does not necessarily mean a simultaneous increase in shareholder value.
Bernstein stressed that CoreWeave needs not only customers willing to buy computing power, but also lenders willing to provide capital, and that financing conditions must be reasonable enough to leave an attractive return on investment for shareholders.
Notably, the market recently reported that CoreWeave is seeking to attract financial support from insurance companies. Bernstein believes that this further highlights the importance of financing issues in the company's investment logic, and also reflects management's urgent need to seek more and lower cost capital.
Meanwhile, research firm Gartner recently released a report once again raising concerns about the business models of emerging AI cloud service providers. Tesla CEO Musk then responded to related discussions with “yikes” on social media, which also attracted market attention.
However, Bernstein believes that the financing dependency problems faced by CoreWeave have not appeared recently. The sharp drop in stock prices on Wednesday more reflects that the market is beginning to re-evaluate existing financial risks, rather than the company disclosing major weaknesses at a new level of operation.
Strong demand for AI is also difficult to overcome long-term competitive pressure
Despite being cautious about CoreWeave, Bernstein doesn't deny the strong demand in the current AI computing power market. The bank expects that CoreWeave will still be expected to sign more large-scale customer contracts in the future, and related news may also drive up the stock price in the short term.
According to Bernstein's forecast, CoreWeave's revenue will reach $12.622 billion this year and grow further to $23.552 billion in 2027. Over the same period, the company's adjusted EBITDA is expected to increase from US$6.826 billion in 2026 to US$14.974 billion in 2027.
However, Bernstein expects CoreWeave's adjusted loss per share to reach $4.56 this year, and the loss will expand further to $4.77 in 2027. The bank believes that the core challenge CoreWeave has faced for a long time is that as global data center supply gradually increases after 2028, there is still great uncertainty about whether the company can continue to obtain large-scale hyperscale cloud computing customer contracts of the current scale.
Meanwhile, large cloud service providers such as Amazon (AMZN.US), Microsoft (MSFT.US), and Google (GOOG.US, GOOGL.US) continue to expand their AI infrastructure and may compete more actively for enterprise AI customers.
Bernstein questioned whether CoreWeave's existing software platform is sufficient to form a long-term competitive barrier, especially when large cloud service providers have stronger financial strength, a broader customer base, and stronger infrastructure integration capabilities. The bank believes that even if overall demand for AI continues to grow, these structural competition issues will not automatically disappear; they may only further delay the time when the risks actually become apparent.
Of course, if the commissioning of data centers built by large cloud service providers is delayed, or if electricity supply continues to limit the construction of new data centers, CoreWeave may still maintain a certain pricing capacity based on existing computing power resources. If its specialized software platform can form stable customer needs, it may also improve long-term competitive prospects.
But Bernstein emphasized that investors need to distinguish the difference between growth in business scale and shareholder financial returns. According to the bank, the key to whether CoreWeave can create long-term value in the future is not only how much AI customers are willing to pay for computing power, but also how expensive the company can raise the capital needed to build this computing power. As unsecured bond yields approach 15%, financing costs are becoming a key variable in determining whether CoreWeave's high growth model can continue.