stockrowWhat it does
CoreWeave provides cloud computing infrastructure and software designed for artificial intelligence. Its customers include AI research labs, technology companies and enterprises that need computing capacity to build and run AI models and applications. Michael Intrator is its chief executive officer.
The CoreWeave Cloud platform supports training, when a model learns from data, and inference, when a trained model is used to produce results. It also supports moving data, refining models and running agentic workflows. Rather than requiring customers to assemble the entire system themselves, CoreWeave combines computing hardware, networking, storage, software and managed services.
Its infrastructure uses clusters of graphics processing units, or GPUs, connected by high-speed networks. CoreWeave operates data centers designed for dense computing equipment, with access to power and, where appropriate, liquid cooling. Its network links facilities across North America and Europe.
Software helps customers put that infrastructure to work. CoreWeave Mission Control provisions resources, manages workloads and monitors performance. The company also offers managed Kubernetes services for running clusters and developer tools through Weights & Biases for training, evaluating and deploying models and applications.
Customers can use the platform for different stages of AI development and deployment. CoreWeave operates the underlying hardware and facilities while customers access the capacity and software they need.
How it makes money
CoreWeave earns revenue by providing cloud capacity and related software and services for AI workloads. It delivers its offerings through Infrastructure Services, Managed Software Services and Application Software Services. These include computing infrastructure, storage, workload management and developer tools.
Customers generally access the platform through multi-year committed contracts. Under these agreements, a customer purchases a specified amount of capacity on a take-or-pay basis over the contract term. Customers can also buy on-demand access through a pay-as-you-go model. A number of customers have expanded their initial commitments by contracting for additional capacity as their AI workloads have grown.
The infrastructure offering brings together GPU clusters, networking and storage. Managed services help customers run and control complex computing environments, while application software supports activities such as model development, evaluation and deployment. Weights & Biases provides tools spanning model training, serving, and agent evaluation and monitoring.
CoreWeave has remaining performance obligations representing contracted work still to be performed. The company expects their growth to occur in steps and says it may include periods of decline.
CoreWeave reported a net loss for the year ended December 31, 2025.
Customers and how it sells
CoreWeave serves enterprises, AI labs and technology companies deploying AI across a range of uses. Customers include organizations training models as well as those running inference and applications. CoreWeave works with customers to tailor solutions to their performance requirements, GPU selection, use case and business goals.
Revenue is concentrated among customers. Customer A accounted for 67.0% of revenue in FY2025. Customers generally purchase capacity through multi-year commitments, although on-demand access is also available. Some have increased their commitments as their workloads have grown.
CoreWeave sells and supports its services through a go-to-market organization that includes sales, marketing, field engineering, support, customer experience, partnerships and revenue operations. Its teams are aligned by industry and geography. Technical support helps customers troubleshoot problems and optimize performance.
The United States generated $4.8 billion of revenue in FY2025, representing 93.6% of total revenue. All other countries generated $330.0 million. CoreWeave operates data centers across the United States, Europe and Canada. It is expanding internationally with a focus on Europe and is evaluating launches in other regions.
| Region | Revenue, FY2025 | Share | Change on the year |
|---|---|---|---|
| United States | $4.8 billion | 93.6% | 167.2% |
| All other countries | $330.0 million | 6.43% | 179.7% |
Suppliers and costs
CoreWeave depends on semiconductor manufacturers, equipment manufacturers and software providers to source and deploy its infrastructure. Its clusters use NVIDIA GPU systems and CPUs from AMD and Intel. The company also needs networking and storage equipment, as well as components such as chillers and HVAC systems for data center cooling.
The company works with suppliers on future capacity and maintains relationships with original equipment and original design manufacturers. It also procures long-lead data center equipment needed to bring computing clusters into service. Availability, pricing and delivery of components can constrain its operations and affect its ability to meet customer commitments.
Data centers require substantial power, cooling and facilities investment. CoreWeave primarily finances infrastructure development through asset-level debt supported by take-or-pay customer contracts, supplemented by corporate equity and debt. It expects to continue incurring debt to meet its capital needs.
CoreWeave anticipates significant investments in infrastructure and go-to-market capabilities for the foreseeable future. Future investments may require significant debt or equity financing, and additional equity sales would dilute existing shareholders.
- Revenue$5.1B
- Cost of revenue−$1.5B
- Selling, general and administrative−$795.0M
- Other operating costs−$2.9B
- Operating income−$46.0M
Competitive position
CoreWeave competes primarily with hyperscalers that provide general-purpose cloud computing as part of broader product portfolios. Some of those hyperscalers are also CoreWeave customers and partners. It also competes with smaller cloud service providers. Individual competitors are not named.
CoreWeave positions its platform around AI and accelerated computing. It identifies access to new GPU generations, its proprietary orchestration software, security standards and total cost of ownership as sources of differentiation. Its infrastructure is designed to serve training and inference workloads and to let customers shift capacity as their requirements change.
Our reading: the combination of specialized infrastructure and software offers a potential advantage in operating complex AI workloads, but the evidence does not establish how durable that advantage will be against larger cloud providers.
Our reading: multi-year capacity commitments can support customer relationships, although they do not guarantee renewal or prevent customers from developing competing services. Access to hardware and power is important to CoreWeave’s ability to compete, but suppliers and facilities partners also constrain its control over deployment.
Growth opportunities
CoreWeave’s growth depends in part on continued adoption of AI and demand for the computing capacity used to train and run models. The company expects AI adoption to spread beyond early users to a broader set of enterprises, including organizations adapting models to their own data and applications. It also seeks to win more workloads from existing AI labs and enterprise customers.
To support demand, CoreWeave is expanding existing data centers and developing new sites. It expects to deploy its contracted power capacity over future periods. The company plans to continue building access to power and exploring alternative energy solutions.
Software is another area of investment. CoreWeave plans to expand its software suite to optimize AI computing. Its 2025 acquisitions included Weights & Biases, Marimo, OpenPipe and Monolith, adding tools for model and agent development and industrial applications.
CoreWeave expects remaining performance obligations to grow in steps rather than smoothly and says the measure may decline in some periods. Winning additional customers depends on factors including demand for AI computing, pricing, competition and the effectiveness of its sales strategy.
- $229.0MFY2023
- $1.9BFY2024
- $5.1BFY2025
- $12.4BFY2026 guidance
Structural risks
-
Insufficient power, higher procurement costs, outages, shortages or capacity constraints could harm the business. Data center construction can also be delayed by equipment availability, permits, site constraints and third-party partners.
-
Customer A accounted for 67.0% of FY2025 revenue. Customers may reduce their use of CoreWeave’s services, choose not to renew or expand purchases, or develop their own competing services.
-
CoreWeave has limited control over the availability, quality, performance and cost of infrastructure components. Delays or mismatches between component availability and contractual obligations could affect delivery to customers.
-
Expansion requires substantial investment. Financing may not be available in sufficient amounts or on acceptable terms, and additional equity financing would dilute shareholders. CoreWeave incurred a net loss for the year ended December 31, 2025.
-
CoreWeave faces competition from companies with greater resources. If enterprises adopt AI less broadly than the company expects, or new use cases do not arise, its opportunity may be smaller than expected.
-
Energy, environmental, export-control and other rules can increase costs or restrict services in certain geographies or for certain customers.
Bottom line
Our reading
CoreWeave combines AI cloud infrastructure with software and sells much of its capacity through long-term commitments. Its business rests on delivering and financing infrastructure quickly enough to meet those commitments, while maintaining access to power and current hardware. The large contracted backlog provides visibility into customer commitments, but concentration, losses and the funding required to build capacity remain central exposures.