May 7 (Reuters) – CoreWeave Inc beat analysts’ estimates for quarterly revenue on Thursday, as the specialized cloud provider tapped ‌into strong demand for its high-performance computing services used ‌to train and deploy artificial intelligence models.

The company’s shares were flat in volatile ​extended trading following a jump in operating expenses.

Demand for services from the so-called neoclouds, such as CoreWeave and peer Nebius, has skyrocketed as companies race to secure the computing capacity required to develop and ‌run AI models.

The company ⁠reported total revenue of $2.08 billion for the first quarter, compared with analysts’ average estimate of $1.97 billion, according ⁠to data compiled by LSEG.

Its operating expenses more than doubled to $2.22 billion in the quarter.

The business model for AI infrastructure is exceptionally ​capital-intensive. CoreWeave ​has been in a race ​to build out data center ‌capacity to meet demand, a strategy that requires billions in upfront investment.

CoreWeave’s core advantage lies in its specialized infrastructure and a close relationship with AI chip bellwether Nvidia, which grants it early and large-scale access to the most sought-after AI hardware.

This has made ‌it a primary destination for AI ​startups and, increasingly, enterprise clients looking to ​bypass the capacity constraints ​at larger cloud providers.

Just in the last month, ‌CoreWeave struck an expanded $21 billion deal ​for additional cloud ​computing capacity with Meta, a $6 billion deal with trading firm Jane Street and a third one with Anthropic.

The company had ​a revenue backlog ‌of $99.4 billion as of March 31, compared to $66.8 billion ​at the end of December.

(Reporting by Juby Babu in ​Mexico City; Editing by Sriraj Kalluvila)