CoreWeave
The Essential Cloud for AI
Last reviewed Mar 14, 2026
CoreWeave is a specialized cloud provider offering GPU‑accelerated infrastructure for AI/ML, VFX, and batch compute, with Kubernetes‑native orchestration and rapid scale.
Available GPUs
Hourly on-demand pricing. Click column headers to sort.
Prices last updated: September 28, 2026
CoreWeave pricing by GPU
Configurations, price rank, and alternatives for one GPU at a time.
Pros & Cons
Advantages
- Extensive selection of NVIDIA GPUs, including latest Blackwell architecture
- Kubernetes-native infrastructure for easy scaling and deployment
- Fast deployment with 10x faster inference spin-up times
- High cluster reliability with 96% goodput and 50% fewer interruptions
- Specialized support for AI, machine learning, and rendering workloads
- Storage priced with no ingress, egress, or transfer fees
- On-demand, spot, and reserved capacity plans for the same GPU fleet
- Expert support from dedicated engineering teams
Limitations
- Data center footprint concentrated in North America and Europe
- Some configurations, including GB300 NVL72 and HGX B300, require contacting sales rather than self-service on-demand
- Specialized nature may not suit all general computing needs
- Learning curve for users unfamiliar with Kubernetes
- Newer player compared to established cloud giants
Key Features
Kubernetes-Native Platform
Purpose-built AI-native platform with Kubernetes-native developer experience
Latest NVIDIA GPUs
First-to-market access to the latest NVIDIA GPUs including Hopper (H100, H200, GH200) and Blackwell/Blackwell Ultra (GB200 NVL72, GB300 NVL72, HGX B200, HGX B300)
Mission Control
Unified security, talent services, and observability platform for large-scale AI operations, including fleet and node lifecycle controllers
High Performance Networking
HPC Interconnect with GPUDirect RDMA over InfiniBand for GPU-to-GPU communication, plus VPCs and Direct Connect private links via Equinix and Megaport
SUNK (Slurm on Kubernetes)
Managed Slurm running on Kubernetes for batch and burst training workloads
Purpose-Built AI Storage
S3-compatible AI Object Storage, POSIX distributed file storage, single-tenant VAST clusters, and local ephemeral storage on GPU nodes
Compute Services
GPU Instances
On-demand and reserved GPU instances with latest NVIDIA hardware
CPU Instances
High-performance CPU instances to complement GPU workloads
Managed Software Services
Managed orchestration and platform services layered on CoreWeave infrastructure
Pricing Options
| Option | Details |
|---|---|
| On-Demand Instances | Pay-per-hour GPU and CPU instances with flexible scaling and no long-term capacity commitment |
| Spot Instances | Burst capacity at a discount to on-demand rates, published for most GPU and CPU instance types |
| Reserved Capacity | Committed usage discounts up to 60% over on-demand pricing |
| Transparent Storage | No ingress, egress, or transfer fees for data movement on CoreWeave storage |
| Inference Per-GPU Pricing | Separate per-GPU hourly rates available to CoreWeave inference platform customers |
Availability & Support
Regions
On-demand and spot GPU and CPU instances published for North America and Europe, with ongoing global AI data center expansion
Support
24/7 support from dedicated engineering teams, comprehensive documentation, and Kubernetes expertise
Getting Started
- 1
Create Account
Sign up for CoreWeave Cloud platform access
- 2
Choose GPU Instance
Select from latest NVIDIA GPUs including H100, H200, and Blackwell architecture
- 3
Deploy via Kubernetes
Use Kubernetes-native tools for workload deployment and scaling
Frequently Asked Questions
What GPU types does CoreWeave offer?
CoreWeave offers various GPU types including A100 SXM, A100 SXM, H100 SXM, H100 SXM, GH200, H200, H200, L40, L40, HGX B300, L40S, L40S, B200, B200, RTX PRO 6000, RTX PRO 6000. Check the pricing table above for current availability and pricing.
How do I get started with CoreWeave?
Create Account, Choose GPU Instance, Deploy via Kubernetes
What are CoreWeave's main advantages?
CoreWeave's main advantages include: Extensive selection of NVIDIA GPUs, including latest Blackwell architecture, Kubernetes-native infrastructure for easy scaling and deployment, Fast deployment with 10x faster inference spin-up times, High cluster reliability with 96% goodput and 50% fewer interruptions, Specialized support for AI, machine learning, and rendering workloads, Storage priced with no ingress, egress, or transfer fees, On-demand, spot, and reserved capacity plans for the same GPU fleet, Expert support from dedicated engineering teams.
What are CoreWeave's limitations?
CoreWeave's main limitations include: Data center footprint concentrated in North America and Europe, Some configurations, including GB300 NVL72 and HGX B300, require contacting sales rather than self-service on-demand, Specialized nature may not suit all general computing needs, Learning curve for users unfamiliar with Kubernetes, Newer player compared to established cloud giants.
Compare Providers
Find the best prices for the same GPUs and models from other providers
