CoreWeave

CoreWeave is a specialized AI cloud platform for AI teams, ML engineers, data scientists, and enterprises that need high-performance GPU compute, storage, and networking for training and inference.

Key capabilities include:

• Bare-metal NVIDIA GPU instances for intensive workloads, including A100, H100, B200, and GB200 options, billed hourly (example rates include 8x H100 at $49.24/hour, 8x B200 at $68.80/hour, and A100 at ~$2.21/GPU/hour plus CPU/RAM).
• Kubernetes-native deployments with support for VPCs to orchestrate large-scale jobs and cluster-based workloads.
• AI Object Storage with automatic access-tiering based on usage patterns (Hot $0.06/GB/mo, Warm $0.03/GB/mo, Cold $0.015/GB/mo) and no fees for egress, requests, IOPS, or internal networking.
• Distributed File Storage priced at $0.070/GB/mo, also with no egress or IOPS fees; public IP is $4.00/month.

Pricing: Paid, on-demand hourly GPU/CPU compute plus per-GB/month storage; committed usage discounts are available (reported up to 60%).

Notable for its transparent, no-hidden-fee model (including zero egress) and purpose-built bare-metal infrastructure optimized for AI performance and large-scale cluster operations, positioned as an alternative to hyperscalers for GPU-heavy workloads.