CoreWeave's capital expenses to double, reaching $30 billion in 2026
Net losses balloon to $284 million in Q4 from $36 million last year
Revenue backlog risk and debt concerns highlighted by analyst
CoreWeave, Inc. forecast that its capital expenses for the year will double, as it spends heavily to scale up its AI cloud platform to handle the massive computing power demanded by its customers to train and deploy advanced AI models.
Its shares slumped 9% in after-hours trading.
The company expects capital expenditure of at least $30 billion in 2026, its co-founder and CEO Michael Intrator said on a call with analysts. Last year, its capital costs were $15.4 billion.
CoreWeave, a so-called neo-cloud that counts hyperscalers such as Microsoft and Alphabet's Google among its rivals, said net losses in the fourth quarter ended December 31 ballooned to $284 million from $36 million in the same quarter last year.
CoreWeave has positioned itself as a more specialized and cost-effective alternative to the big tech companies, attracting clients ranging from AI labs to large enterprises.
But it still relies heavily on large clients such as Microsoft and OpenAI, which remain critical to its growth trajectory, and faces significant revenue backlog risk.
CoreWeave faces lingering concerns regarding its backlog risk, debt and cost of capital, D.A. Davidson analyst Alexander Platt noted.
Revenue backlog was $66.8 billion as of December 31, CoreWeave said, up from $15.1 billion a year earlier.
Nvidia announced a $2 billion investment in CoreWeave in January, becoming the AI infrastructure provider's second-largest shareholder.
CoreWeave's adjusted operating income margin fell to 6% in the fourth quarter, from 16% a year earlier.
It has repurposed its high-performance GPU infrastructure, originally built for large-scale crypto mining, to establish itself as a premier cloud provider for the global AI sector.
The company reported revenue of $1.57 billion for the fourth quarter, compared with analysts' average estimate of $1.55 billion.