CoreWeave operates a cloud platform for artificial intelligence (AI) and high-performance computing. Customers rent GPU computing power along with storage, networking, and software to train and run models. The company makes most of its revenue through committed customer contracts, while also offering on-demand access.
What CoreWeave sells
CoreWeave is a cloud service provider: it operates or secures computing infrastructure and sells customers access to it, rather than selling them GPUs to install themselves. Its platform combines GPU clusters with CPUs, high-speed connections between servers, AI-oriented storage, and software for provisioning, scheduling, orchestration, and monitoring workloads. CoreWeave describes this integrated offering in its FY2025 Form 10-K.
That matters because a large AI workload uses more than processors. Data must move between storage and GPUs, and across servers in a cluster; software must allocate resources and coordinate the work. A bottleneck in any of those layers can affect how effectively a customer uses rented compute.
Mission Control and Slurm on Kubernetes
Mission Control is CoreWeave’s proprietary orchestration and operations software. The company also offers Slurm on Kubernetes (SUNK), which supports large-scale research and training workloads. Its filing describes managed and application software services, including developer tools, as part of the broader platform.
#1 Best Overall
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
- Dual ball fan bearings last up to twice as long as sleeve bearing designs
What customers use the GPU cloud for
CoreWeave identifies model training, inference, agentic AI, agent development, and specialized workloads as use cases. Training uses compute to build or refine a model; inference runs a trained model to generate outputs. The company’s cloud targets both types of work.
Training at scale may require many GPUs working together, with fast networking and sufficient data throughput. Inference deployments can have different location and capacity needs: CoreWeave says its facilities vary in size and location, with smaller sites suited to inference near users and larger sites supporting high-density training.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
How CoreWeave makes money
Customers pay for cloud computing services, including compute enabled by CoreWeave’s software and infrastructure optimized for AI and high-performance computing. The company offers committed contracts as well as on-demand access. Its FY2025 Form 10-K describes committed contracts as take-or-pay arrangements that typically involve customer prepayment before service access.
Committed contracts represented over 98% of CoreWeave’s revenue in 2025, compared with 96% in 2024 and 88% in 2023, according to the same filing. The figures show that contracted commitments, rather than purely on-demand usage, underpin most of the reported revenue.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Revenue growth and losses
CoreWeave reported $5.1 billion in revenue and a $1.2 billion net loss for 2025. In 2024, it reported $1.9 billion in revenue and an $863 million net loss; in 2023, it reported $229 million in revenue and a $594 million net loss. These figures are from the company’s FY2025 results announcement and cover the years ended December 31 of each year.
The combination of rapid growth and continued losses is important to understanding the business model. Building or securing data-center capacity, buying servers and networking equipment, and supplying power require substantial investment before or alongside customer service. Revenue growth alone does not establish that the business has become profitable.
Rank #4
- Powered by Radeon RX 9070 XT
- WINDFORCE Cooling System
- Hawk Fan
- Server-grade Thermal Conductive Gel
- RGB Lighting
What the $66.8 billion backlog means
CoreWeave reported $66.8 billion in revenue backlog as of December 31, 2025, in its FY2025 results announcement. The company defines this measure as remaining performance obligations plus other amounts it estimates will be recognized in future periods under committed contracts. It is subject to delivery and service-availability requirements, so it is not revenue already earned or guaranteed cash.
How to assess the GPU-cloud proposition
CoreWeave’s position is that general-purpose cloud environments were not designed for the combination of high-density compute, advanced networking, optimized storage, and software required by distributed AI workloads. That is the company’s positioning, not proof that general-purpose cloud providers cannot support AI workloads.
Best Value
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- Phase-change GPU thermal pad helps ensure optimal heat transfer, lowering GPU temperatures for enhanced performance and reliability
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
- 0dB technology lets you enjoy light gaming in relative silence
For a customer comparing providers, practical fit depends on the workload and its operating requirements. Relevant questions include:
- Which GPU types and cluster sizes are available for the workload?
- Can networking and storage throughput keep the processors supplied with data?
- Do orchestration tools and software support the customer’s training or inference setup?
- Where is capacity located, and does that location suit latency and data requirements?
- What reliability commitments and contract flexibility are offered?
- What is the total cost for the workload, including the supporting infrastructure and services?
CoreWeave’s FY2025 filing does not provide a full apples-to-apples price comparison with other cloud providers. Current GPU availability, service prices, and specific contract terms therefore need to be checked with providers for the relevant workload and date.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Business risks behind the growth
Committed contracts can give CoreWeave visibility into customer demand, but they do not remove the risks of operating a capital-intensive cloud business. The company’s FY2025 Form 10-K identifies material exposures including:
- The need for substantial capital expenditure and financing to build or secure capacity.
- Access to sufficient power and the cost of that power.
- Reliance on a limited number of suppliers for important components.
- Data-center partner performance and the ability to deliver service as planned.
- Customer concentration.
- Uncertainty about whether demand for AI services will continue at its current pace.
Rapid hardware cycles add an operational challenge: capacity decisions must account for technology that can change quickly, while facilities, power, and financing commitments are large and long-lived.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
