The companies building the infrastructure behind artificial intelligence are facing a difficult question: how long can they keep borrowing billions of dollars before the returns begin to catch up? CoreWeave, a cloud computing company that rents out computing power for AI workloads, has become a focal point of that concern as investors grow increasingly uneasy about its debt.
The cost of insuring CoreWeave’s debt against default has surged, with its five-year credit default swap spreads climbing above 1,100 basis points on Thursday, October 15. At the same time, the yield on its bonds due in 2031 briefly reached 16%, a level that signals serious concern among investors. These market movements do not mean the company is about to default, but they show how sharply perceptions of its financial risk have changed.
Why Investors Are Worried About CoreWeave
CoreWeave supplies computing capacity powered by advanced Nvidia processors. AI developers need this capacity to train models and run services, giving companies such as CoreWeave an important role in the expanding AI industry.
The challenge lies in the cost of meeting that demand. Building data centres and purchasing specialised chips requires enormous upfront spending, and CoreWeave has relied heavily on borrowing to finance its expansion. Reuters reported that the company has around $50 billion in debt and raised its capital spending forecast in August to between $35 billion and $39 billion for the year
Heavy investment can support rapid growth, but it also leaves less room for error. If financing becomes more expensive or expected revenue takes longer to arrive, a company may find it harder to balance expansion with its debt obligations. There is another complication: computing hardware evolves quickly, meaning providers must keep investing to remain competitive.
The Problem Extends Beyond One Company
CoreWeave’s difficulties have arrived at a time when investors are reassessing the cost of the wider AI build-out. Data centres, chips and computing infrastructure have attracted enormous spending, driven by expectations that AI adoption will create substantial new revenue.
Yet spending on infrastructure and earning money from AI are two different things. A company can have strong customer demand and still struggle to generate enough cash to cover its investments, interest payments and operating costs.
Research firm Gartner has warned that around 80% of specialised cloud providers, often called neoclouds, could cease operating independently by 2030. It has also raised concerns that more than $3.5 trillion in AI infrastructure spending could face impairment if investment continues at its current pace without sufficient returns.
The concerns have spread through financial markets. Reuters reported that CoreWeave’s sell-off coincided with declines in other AI-related companies and a sharp fall in semiconductor stocks, suggesting that investors are examining the financing behind the technology boom more closely.
CoreWeave Pushes Back Against the Concerns
The company’s leadership has defended its position. Chief Executive Mike Intrator told employees that CoreWeave’s deal backlog more than covered its debt obligations and argued that financial markets can sometimes misjudge a business.
That backlog is an important part of the argument in CoreWeave’s favour, but contracted business does not automatically translate into immediate cash. Investors will want to see whether the company can deliver the promised computing capacity, manage its costs and generate sufficient cash flow as its commitments grow.
The coming months will help show whether the current sell-off reflects excessive market pessimism or a deeper problem with the economics of debt-funded AI infrastructure.
For the industry, the question is becoming harder to ignore. Demand for AI computing may continue to rise, but companies will still need a sustainable way to pay for the machines, buildings and energy required to supply it.
