CoreWeave (NASDAQ:CRWV) reported a larger-than-expected fourth-quarter loss and a significant shortfall in adjusted operating margin, overshadowing modest revenue growth and sending shares down more than 10% in premarket trading on Friday.
The company posted a quarterly loss of $0.89 per share, compared with analyst expectations for a loss of $0.21. Revenue rose slightly to $1.57 billion from $1.55 billion in the same period last year.
However, adjusted operating margin came in at 5.5%, well below the 8.6% forecast by analysts. Profitability was pressured by rising costs for AI chips, increased competition for computing capacity, and expenses tied to expanding the company’s data center infrastructure.
“The narrative is dominated by a significant adj-OM miss (5.6% vs. 8.6% consensus) driven by a massive, above-guidance surge in CapEx,” Stifel analyst Ruben Roy said in a note.
Analysts said management is accelerating infrastructure spending to prepare for the upcoming Blackwell/Rubin deployment cycle, bringing forward $8.2 billion in fourth-quarter capital expenditures compared with previously guided spending of $6.0 billion.
“We are awaiting evidence of that inflection before turning more constructive,” they wrote, cutting the price target to $110 from $120.
Adjusted operating income totaled $88 million, missing analyst estimates of $139.7 million.
Chief Executive Michael Intrator said 2025 marked a milestone year for the company, as annual revenue reached $5 billion amid growing demand from customers running artificial intelligence workloads on its cloud platform.
Chief Financial Officer Nitin Agrawal added that the company’s revenue backlog expanded to $66.8 billion — more than four times the level recorded at the beginning of the year — providing stronger visibility into performance through 2026 and beyond.
CoreWeave has secured several multibillion-dollar agreements with major technology companies, including a $14 billion partnership with Meta Platforms and a $6.5 billion contract with OpenAI, reflecting accelerating demand for AI computing infrastructure.