Why CoreWeave Is in the Crosshairs of the AI Scare

LONDON/BENGALURU, October 16 — Loss-making, heavily indebted US cloud company CoreWeave has become a focal point for investors worried about a potential artificial intelligence bubble. Its recent sell-off is raising broader questions about the sustainability of the massive investments powering the AI boom.

The Nvidia-backed company, valued around its $39 billion infrastructure business, suffered a sharp decline in its shares and bonds this week as investors reassessed the risks of financing data centers with borrowed money. CoreWeave shares fell more than 15%, while yields on its 2030 bonds climbed approximately 500 basis points to nearly 16%.

The downturn followed quarterly reports from major cloud computing companies that pointed to delays in data center projects and a slower pace of capital expenditure growth.

Heavy Debt Raises Concerns About AI Infrastructure

Investors have increasingly focused on so-called neocloud providers, including CoreWeave, Nebius, Applied Digital, Iren, Lambda Labs, Crusoe and TeraWulf. These companies build energy-intensive data centers and rent computing capacity to businesses developing and operating AI models.

The business model requires enormous upfront investment in land, electricity infrastructure and advanced processors, while substantial revenue may not arrive until facilities become operational and customer leases begin.

CoreWeave has never recorded an annual profit, according to LSEG data. In 2025, it generated $5.1 billion in revenue but reported a net loss of $1.2 billion.

The company’s capital expenditure is expected to increase fourfold this year, reaching between $35 billion and $39 billion, even as revenue is forecast to double. This expansion could intensify cash outflows and increase the company’s dependence on external financing.

James Athey, a fixed-income portfolio manager at Marlborough, warned that interconnected financing arrangements and concentrated investor exposure could amplify problems across the AI ecosystem. High leverage, he said, can turn relatively contained market concerns into more serious financial risks.

CoreWeave Defends Its Backlog Despite Mounting Pressure

CoreWeave Chief Executive Mike Intrator told employees in an internal memo that markets can sometimes “misunderstand and act irrationally.” He also argued that the company’s customer commitments provide a strong foundation for future revenue.

At the end of the second quarter, CoreWeave reported a revenue backlog of $104.2 billion, followed by more than $25 billion in additional commitments early in the third quarter.

However, investors remain concerned about whether the company can finance the infrastructure required to fulfill those contracts.

The company has approximately $50 billion in debt, and reports that Intrator and entities linked to him have sold nearly $1 billion in stock since its March 2025 initial public offering have added to scrutiny. The sales were mostly conducted under prearranged trading plans, according to regulatory filings.

Alberto Gallo, founder of hedge fund Andromeda Capital, argued that CoreWeave may need additional capital or financial assistance because its spending is outpacing the resources available to fund its expansion. His firm holds positions betting against CoreWeave bonds.

AI Financing Models Face Greater Scrutiny

The concerns extend beyond CoreWeave’s individual finances. Investors are examining the interconnected relationships between chipmakers, cloud infrastructure providers and major AI customers.

Nvidia has agreed to purchase any unsold CoreWeave computing capacity through April 2032, providing a potential source of demand for capacity the company cannot sell elsewhere.

However, critics question whether financing arrangements between technology suppliers and their customers could obscure the underlying economics of the AI industry.

The risks are compounded by uncertainty over data center construction timelines, access to electricity and the possibility that expensive graphics processing units could become outdated before facilities begin generating sufficient returns.

Tommaso Leodari, chief investment officer at hedge fund Index & Cie, said revenue growth across the sector was increasingly being supported by debt rather than equity. He warned that any slowdown in utilization or data center construction could create problems for companies trying to repay their obligations.

Industry Forecast Warns of Potential Consolidation

Research firm Gartner estimates that approximately 80% of the more than 100 neocloud providers could cease independent operations by 2030.

The forecast reflects the difficulty of balancing rapidly expanding infrastructure spending against revenue that may take time to materialize.

Concerns have also increased following warnings from major technology companies, including Microsoft and Oracle, about data center delays, power constraints and component shortages. Plans by Meta and Microsoft to develop their own chips to compete with Nvidia have added further uncertainty for infrastructure providers dependent on Nvidia processors.

Although OpenAI and Anthropic continue to generate substantial revenue growth, the pace of expansion has moderated.

Michael Frazis of Sydney-based Frazis Capital Partners argued that CoreWeave’s projected losses demonstrate that economies of scale have yet to become clearly established across the industry.

Default Protection Costs Signal Severe Distress

Despite the mounting concerns, CoreWeave retains backing from Nvidia and has several potential financing options, according to Moody’s Ratings Vice President Sean Cray. He also noted that the company does not purchase chips speculatively.

Nevertheless, credit markets have become increasingly skeptical. The cost of insuring CoreWeave’s debt against default rose to approximately 1,200 basis points this week, signaling severe financial stress.

Saurav Sen of research firm Gimme Credit warned that the company’s credit sell-off could become self-reinforcing, with concerns increasingly centered on whether CoreWeave can afford to deliver the computing capacity its customers have contracted.

What the CoreWeave Sell-Off Means for the AI Boom

CoreWeave’s difficulties highlight a central question facing the AI industry: whether strong demand for computing power will translate into sufficient cash flow to justify the enormous cost of building the infrastructure required.

The company has a substantial contract backlog and support from Nvidia, but its heavy debt burden, persistent losses and rising capital expenditure leave it vulnerable to higher financing costs and delays.

As investors become more selective, AI infrastructure companies may face greater pressure to demonstrate sustainable profitability rather than relying primarily on projections of future demand.