CoreWeave (CRWV) reported a huge first-quarter sales jump, but the company still burned through far more money as it spent heavily on AI cloud infrastructure. Revenue hit $2.08 billion for the three months ended March 31, 2026, up from $982 million a year earlier. Analysts tracked by LSEG expected $1.97 billion, so CoreWeave came in above Wall Street’s revenue target. The company also reported a loss of $1.40 per share, though that figure may not be directly comparable with estimates. But CoreWeave posted a net loss of $740 million, compared with $315 million in the same quarter last year. Operating expenses rose to $2.22 billion from $1.01 billion, while operating loss widened to $144 million from $27 million. Its operating loss margin worsened to 7% from 3%. The net loss margin also expanded to 36%, up from 32%. Basic net loss per share stayed at $1.40, while diluted net loss per share improved from $1.49 to $1.40. CoreWeave grows revenue fast while debt costs eat deeper into earnings CoreWeave is selling into one of the loudest markets on the planet right now: AI compute. That demand showed up in the company’s backlog. Revenue backlog reached $99.4 billion by the end of March. Michael Intrator, co-founder, chairman, and chief executive officer, called it the strongest bookings quarter in the company’s history. Michael said CoreWeave had crossed 1 GW of active power and believes it can pass 8 GW by 2030. The income statement still looked painful. Interest expense, net, rose to $536 million, compared with $264 million last year. That cost is tied to the capital-heavy nature of the business. AI data centers need GPUs, power, land, networking, cooling, and long-term financing. None of that is cheap. The company’s non-GAAP numbers gave a fuller look at the same pressure. Adjusted EBITDA increased to $1.16 billion, up from $606 million. But adjusted EBITDA margin fell to 56% from 62%. Adjusted operating income dropped hard to $21 million, down from $163 million. The margin on that measure fell to 1%, compared with 17% a year ago. Adjusted net loss widened to $589 million from $150 million, while adjusted net loss margin worsened to 28% from 15%. CoreWeave signs major AI customers and lines up more power for data centers CoreWeave reported new business across AI labs, hyperscalers, and enterprise customers. The company signed several agreements with Meta Platforms (META), including a $21 billion commitment in March. It also signed a multi-year deal with Anthropic to support the development and deployment of the Claude family of AI models. The company said it expanded existing relationships with Cohere, Jane Street, and Mistral. It also listed Adaption Labs, Advaita Bio, Hudson River Trading, Perplexity, and World Labs among customers using its infrastructure. That customer base shows how wide the AI compute race has become. It is not only frontier labs. Trading firms, enterprise users, robotics teams, and app builders all need large GPU capacity. Power remained a major part of the quarter. CoreWeave passed 1 GW of active power and added more than 400 MW of contracted power. Total contracted power is now above 3.5 GW, with the company using a broader group of providers. The company also expanded its long-running relationship with NVIDIA (NVDA) to speed up the buildout of more than 5 GW of AI factories by 2030. On the product side, CoreWeave was among the first cloud providers named NVIDIA Exemplar Cloud for inference on NVIDIA GB200 NVL72 systems. It announced Flexible Capacity Plans, including Flex Reservations and Spot, so customers can better match usage with real AI workload needs. Coreweave also introduced Dedicated Inference, which lets customers choose GPU SKUs and runtimes while keeping visibility into production infrastructure. CoreWeave also launched CoreWeave Arena, a production-ready setup where customers can run and test real workloads on CoreWeave Cloud. The company expanded Weights & Biases through W&B Weave and W&B Models, with a focus on agentic AI and robotics products. To fund more growth, CoreWeave secured an $8.5 billion DDTL 4.0 facility, described as a non-recourse, investment-grade, delayed draw term loan. The floating tranche is priced at SOFR plus 2.25%, while the fixed tranche is around 5.9%. NVIDIA also closed a $2 billion Class A common stock investment in CoreWeave. Your bank is using your money. You’re getting the scraps. Watch our free video on becoming your own bank
