CoreWeave is a specialized AI cloud infrastructure company. It makes money by selling GPU-accelerated cloud capacity and related software services to AI labs, hyperscalers, startups and enterprises. The company is built around large committed contracts, where revenue is recognized as CoreWeave delivers available compute, storage, networking and platform services.
The business model is asset-heavy. CoreWeave secures GPUs, data-center capacity, power, networking and financing, then turns that infrastructure into contracted cloud revenue. This creates large forward visibility, shown by $99.4 billion of revenue backlog at March 31, 2026, but it also requires heavy capital investment. At the same date, CoreWeave had $36.4 billion of net property and equipment and $24.9 billion of current plus non-current debt.
Main revenue streams and service categories include
- GPU cloud compute: CoreWeave provides accelerated infrastructure for AI training, fine-tuning, inference and other high-performance computing workloads.
- AI cloud platform services: The platform includes CoreWeave Kubernetes Service, AI-optimized object and file storage, GPU-local caching and CoreWeave Mission Control.
- Inference and capacity products: Dedicated Inference, Flexible Capacity Plans and Spot offerings help customers match AI workload needs with available GPU capacity.
- AI development tooling: Weights and Biases products, including W and B Models and W and B Weave, add model development, evaluation and observability capabilities above raw infrastructure.
CoreWeave reported Q1 2026 revenue of $2.078 billion, up from $982 million in Q1 2025. Adjusted EBITDA was $1.157 billion, equal to a 56% margin, while GAAP losses remained significant. The company posted a Q1 2026 operating loss of $144 million and a net loss of $740 million. This profile shows the core trade-off in the model: rapid revenue growth and strong infrastructure-level earnings before financing and other costs, balanced against high depreciation, interest expense and buildout requirements.
CoreWeave’s competitive advantages are concentrated in four areas
- NVIDIA alignment: NVIDIA invested $2 billion in CoreWeave Class A common stock in Q1 2026, and the two companies expanded plans to support more than 5 GW of AI factories by 2030.
- Early deployment of advanced AI systems: CoreWeave has highlighted NVIDIA Exemplar Cloud status for training and inference workloads, including inference on NVIDIA GB200 NVL72, and completed bring-up and validation of NVIDIA Vera Rubin NVL72 in June 2026.
- Large contracted backlog: The $99.4 billion backlog provides unusually high forward revenue visibility, subject to delivery, availability and customer contract performance.
- Power and infrastructure pipeline: CoreWeave surpassed 1 GW of active power in Q1 2026 and expanded contracted power by more than 400 MW to over 3.5 GW. Management targets more than 8 GW of active power by 2030.
CoreWeave competes with hyperscale cloud platforms and AI-focused infrastructure providers. Direct competitors include Microsoft Azure, Amazon Web Services, Google Cloud, Oracle Cloud Infrastructure, Nebius Group and other specialized GPU cloud providers. The hyperscalers have broader cloud ecosystems, larger balance sheets and deep enterprise relationships. CoreWeave’s differentiation is narrower and more focused: AI-native GPU infrastructure, rapid deployment of NVIDIA systems and large-scale capacity commitments for model developers and enterprises.
Nebius Group is a useful public peer because it is closer to a pure-play AI infrastructure company than diversified hyperscalers such as Microsoft, Amazon or Alphabet. Compared with Nebius, CoreWeave has a larger disclosed backlog and a highly visible NVIDIA relationship, but also carries substantial leverage and capital intensity. Compared with Oracle Cloud Infrastructure, CoreWeave is more specialized in AI GPU workloads, while Oracle operates inside a broader enterprise software and cloud strategy.
CoreWeave’s market position is that of a fast-growing specialist rather than a general-purpose cloud leader. Its customer wins and expansions in Q1 2026 included Meta, Anthropic, Cohere, Jane Street, Mistral, Perplexity and World Labs. The March 2026 Meta commitment of $21 billion and the multi-year Anthropic agreement reinforce its role as an infrastructure supplier to major AI model builders.
China is not a meaningful disclosed demand market for CoreWeave. The company emphasizes U.S. and European data-center footprints and customers, with no disclosed China revenue segment or China-specific growth strategy in the latest Q1 2026 materials. China matters more indirectly through advanced semiconductor supply, export controls, tariffs, geopolitics and supply-chain risk.